Citation
Bioanalytical Applications of Affinity-Based Nanotube Membranes for Sensing and Separations

Material Information

Title:
Bioanalytical Applications of Affinity-Based Nanotube Membranes for Sensing and Separations
Creator:
Caicedo, Hector
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (212 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemistry
Committee Chair:
Martin, Charles R.
Committee Members:
Lyons, Thomas J.
Powell, David H.
Omenetto, Nicolo
Dempere, Luisa A.
Graduation Date:
8/9/2008

Subjects

Subjects / Keywords:
Antibodies ( jstor )
Chemicals ( jstor )
Electronic journals ( jstor )
Fluorescence ( jstor )
Imidazoles ( jstor )
Metal ions ( jstor )
Molecules ( jstor )
Nanotubes ( jstor )
Quenching ( jstor )
Sorption ( jstor )
Chemistry -- Dissertations, Academic -- UF
affinity, alumina, antibodies, bioanalytical, dna, fluorescence, flux, membrane, nanometer, nanopore, nanotechnology, nanotube, partition, permeability, protein, sensing, separations, synthesis, template
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Chemistry thesis, Ph.D.

Notes

Abstract:
Nanotechnology has played an important role in the development of research and technology during the last two decades. The contribution of nanotechnology in different fields, along with the versatility of the constructed nanoscale materials, have made nanotechnology one of the most suitable tools to develop particular nanostructures to realize a desired function and application. A nanostructure is simply an entity at the nanometer scale with one, two or three dimensional features. Since nanotechnology covers a broad range of nanoscale materials, to simplify nanotechnology, it can be classified into two categories based on how the nanostructures are prepared: top-down and bottom-up. In the top-down methods, the nanostructures are constructed by chiseling larger bulk materials into entities of smaller size. Conversely, in the bottom-up case, small units are grown or assembled into their desired size and shape. The nanoporous materials specifically have attracted a lot of attention because they can be used for the synthesis of a variety of functional nanostructures of great usefulness in technology. These porous nanostructures usually combine many of the advantages of the top-down and bottom-up methodologies such as flexibility, size controllability, and cost. The research presented in this work utilizes nanoporous membranes to develop porous nanostructured platforms with potential applications in sensing and separations. In particular, this work is centered in fundamental studies for bioanalytical applications of affinity-based nanotube membranes for sensing and separations. A bottom-up methodology like the template synthesis was used to produce silica nanotubes inside of the pores of alumina membrane. The functionalization of the inside walls of these silica nanotube membranes allowed control of the functional behavior and properties of the nanostructured membrane during membrane-based separations and sensing. The general scheme of the work presented here, is distributed in seven chapters. The first chapter consists of a general description of the theory and fundamentals as background supporting evidence for the work presented here. The template synthesis method, the fabrication of the porous alumina membranes and the possibilities that silane chemistry offers in order to functionalize the nanostructured membranes is discussed. In addition, a brief overview of the fundamentals of biological production of proteins, antibodies, and recombinant proteins is described. After this introduction, the studies presented herein are centered in potential bioanalytical applications of the silica nanotube membranes. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2008.
Local:
Adviser: Martin, Charles R.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31
Statement of Responsibility:
by Hector Caicedo.

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Caicedo, Hector. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/31/2010
Classification:
LD1780 2008 ( lcc )

Downloads

This item has the following downloads:


Full Text





change of the membrane resistant that was detected by impedance spectroscopy which is an

effective technique to probe the interfacial properties of modified electrodes. These results

showed the feasibility of using PNA-modified nanotube membranes as a sensing-platform for

DNA hybridization methods. This assembled nanotube membrane platform present the special

characteristic that the membrane itself performed the function of the electrodes (working and

reference respectively). It was demonstrated that PNA immobilized on the nanotube membranes

function as a good probe to detect the DNA targets in condition where no electrolyte is present

during the hybridization process. It was also demonstrated the good reproducibility obtained

during the fabrication and detection with this novel sensing-platform.










transport experiments carried out at low ionic strength in antibody-modified membranes that has

been previously equilibrate with a protein that strongly binds the antibody carrier, both the

modified membrane and this permeating protein bearing similar electric double-layer that creates

double-layer repulsions when they approach each other. Besides, at low ionic strength the

overlapping between the double-layers both of the membrane pore wall, and that one of the

permeating protein affects the mobility across the membrane.193, 194 Therefore, the repulsion

effects together with the double-layer overlapping are the responsible factors of the enhanced

selectivity at low ionic strength observed in Figure 2-21B. Then, the high selectivity obtained

under these conditions might be attributed to the electrostatic exclusion of the charged anti-M-

IgG molecules from the membrane pores that were equilibrated with the same protein (anti-M-

IgG) having identical isoelectric point.162, 193


A B..
A B *



I I rI
T) ( n"
0 0
1.5. i of o *te .
2 1.0 -

0 0 0.5 -
o 0
E E
0 0.5 -

A AA.AAA"AA.A
A A 0.0 A A A A a
0.0 ..0.0
0 200 400 600 800 1,000 1,200 0 200 400 600 800 1,000 1,200
Time (min) Time (min)
Figure 2-21.Flux plots for anti-M-IgG (A) and anti-H-IgG (m) across M-IgG -modified
membranes, (A) without equilibration, and (B) after equilibration. The feed solution
concentration was 2.65 nM in both of the permeating proteins.

These data suggest that this antibody-modified nanotube-membranes at low ionic strength

can be used for situations were the transport of an unwanted protein needs to be shut down, in

order to improve selectivity. One potential application that this approach might be useful is in the

sample complexity reduction in proteomic analysis. Since the analysis and characterization of










carried out experiments to determine the effect of the feed solution concentration on the rate of

fluorescence quenching.

For these experiments, the concentration of H-IgG-FITC in the receiver solution was 3.35

nM in all cases, whereas three different concentrations for anti-H-IgG-QSY were used in the

feed solution. The three concentrations of anti-H-IgG-QSY were respectively, 16.6 nM, 33.3

nM, and 66.6 nM. These results are shown in Figure 5-7.

3.5
180- A B

S160 3.0 -
a) 0 V ~ A '7
S140 A
a vv I 2.5 7 v,
g 12 0 A v v7 A e 7 77
z I AA
1 0 0 7 7* A v 2 .0 7 A 7
L 0 .:*, 77 v d C .. "*A .. 7 ,aV V d


< 60 b 0
S1.0
40 I
0 200 400 600 800 1000 1200 1400 1600 1800 0 200 400 600 800 1000 1200 1400 1600 1800
Time (min) Time (min)
Figure 5-7.Typical plots for (A) changes of fluorescence signal and (B) concentration of
unquenched protein in function of time for three different feed solution
concentrations: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control).

Figure 5-7 shows the influence of feed solution concentration on the rate of fluorescence

quenching in the receiver solution. As can be observed from Figure 6-7, transport experiments

with higher concentration of anti-H-IgG-QSY yields to a faster rate of the fluorescence

quenching. These results confirmed that the molecular recognition reaction between the labeled

antibodies generated the appropriate distance between the donor (FITC) and acceptor (QSY-7)

molecules necessary for this fluorescence quenching-mediated detection. This was also verified

by comparing this effect with the control experiment (plot d in Figure 5-7). The significant

differences in fluorescence intensity for the experiments shown in Figure 5-7were observed only

during the initial four hours of the transport experiments. After this period of time, the changes in









It is known that in the position of the critical angle, the angle of refraction, 02, is 90.

Consequently, the angle of incidence is the critical angle and the Snell's equation is reduced to

Equation 1-9:


sin 0c = n- (1-9)
ni
The critical angle, 0c, can therefore be found simply by knowing the refractive indexes of

the two mediums.

It is worth mentioning that the evanescent wave is nontransverse, which allows all its

vector components to interact with the dipoles in all directions, thus this wave is a very

informative probe of the sample material.134

During FTIR-ATR analysis, the infrared radiation is passed through an infrared

transmitting crystal characterized by having a high refractive index, allowing the radiation to

reflect within the crystal several times. Along with each reflection in the transmitting crystal, at

the top surface, the evanescent wave penetrates the sample in order to explore the surface

material. It should be pointed out that the depth of penetration and the total number of

reflections along the crystal can be controlled either by varying the angle of incidence or by

selection of crystals. According to Equation 1-10, the penetration depth of the evanescent wave

in the sample material can be controlled by the selection of transmitting crystal, since each

crystal posses its own refractive index.


d = A (1-10)



Where 2 is the wavelength of the IR radiation, n2 and nl are the refractive index of the

sample and the refractive index of the crystal respectively, and 0 is the angle of incidence.

Therefore, crystals with higher refractive index allow lower penetration depths. The depth of









[73] Masuda, H.; Yamada, H.; Satoh, M.; Asoh, H.; Nakao, M.; Tamamura, T. Applied
Physics Letters 1997, 71, 2770-2772.

[74] Asoh, H.; Nishio, K.; Nakao, M.; Tamamura, T.; Masuda, H. Journal of the
Electrochemical Society 2001, 148, B152-B156.

[75] Diggle, J. W.; Downie, T. C.; Goulding, C. W. ChemicalReviews 1969, 69, 365-405.

[76] Jessensky, O.; Muller, F.; Gosele, U. Applied Physics Letters 1998, 72, 1173-1175.

[77] O'Sullivan, J. P.; Wood, G. C. Proceedings of the Royal Society ofLondon, Series A:
Mathematical, Physical and Engineering Sciences 1970, 317, 511-543.

[78] Parkhutik, V. P.; Shershul'skii, V. I. Journal of Physics D: Applied Physics 1992, 25,
1258-1263.

[79] Siejka, J.; Ortega, C. Journal of the Electrochemical Society 1977, 124, 883-891.

[80] Hench, L. L.; West, J. K. ChemicalReviews 1990, 90, 33-72.

[81] Livage, J.; Henry, M.; Sanchez, C. Progress in Solid State Chemistry 1988, 18, 259-341.

[82] Patrissi, C. J.; Martin, C. R. Journal of the Electrochemical Society 1999, 146, 3176-
3180.

[83] Sides, C. R.; Martin, C. R. Modern Aspects ofElectrochemistry 2007, 40, 75-126.

[84] Sides, C. R.; Li, N.; Patrissi, C. J.; Scrosati, B.; Martin, C. R. MRS Bulletin 2002, 27,
604-607.

[85] Lakshmi, B. B.; Patrissi, C. J.; Martin, C. R. Chemistry ofMaterials 1997, 9, 2544-2550.

[86] Lin, J.; Brown, C. W. TrAC, Trends in Analytical Chemistry 1997, 16, 200-211.

[87] Collinson, M. M. Mikrochimica Acta 1998, 129, 149-165.

[88] Avnir, D. Accounts of Chemical Research 1995, 28, 328-334.

[89] Walcarius, A. Electroanalysis 1998, 10, 1217-1235.

[90] Brinker, C. J. Current Opinion in Colloid & Interface Science 1998, 3, 166-173.

[91] Walcarius, A. Electroanalysis 2001, 13, 701-718.

[92] Stoeber, W.; Fink, A.; Bohn, E. Journal of Colloid and interface Science 1968, 26, 62-
69.










silica has been ranged from 1.9 to 4.5.327 Because of the different electrical characteristics of

these materials, a change in the surface characteristics of the membranes should be expected

when silica nanotubes are synthesized within the alumina membrane, as was observed in the

Nyquist plots in Figure 6-5.

In order to assess the stability of the impedance measurements a study of the wettability of

the bare alumina membrane was carried out by monitoring the changes in resistance with the

time. These results are summarized in Figure 6-7.



8000

1 6000 ,

r 4000 -

2000


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Wetting Time (hours)
Figure 6-7.Effect of the wetting time on the resistance impedance measurement for an assembled
unmodified alumina membrane 200 nm pore diameter.

As can be seen from Figure 6-7, the changes in resistance from impedance measurements

of alumina membrane were pretty stable with the time. These results confirmed the stability of

the sputter-coated membranes and the stability itself of the impedance measurements.

Functionalization of the Silica nanotube Membranes

The differences in membrane resistance before and after each modification step during the

immobilization process were analyzed in order to characterize the accomplishment of the various

modifications (Figure 6-8). The obtained values of Rm from the Nyquist plots in Figure 6-8 are

compiled in Table 6-1.









Functional Behavior of the M-IgG- and H-IgG-modified Membranes

In these studies, we have been exploring the transport properties, and separation of

different anti-IgGs through antibody-modified nanotube membranes, and how the type of

immobilized antibody alters the functional behavior of these functionalized membranes.

Although, porous membranes have been used for size-based protein separation with great

success, separation of proteins with similar size is more difficult.7 This process is even more

challenging, when the separation is between proteins that have practically the same molecular

structure and size as in the case of different anti-IgGs.

Flux plots for the transport of anti-M-IgG and anti-H-IgG across the M-IgG-modified

nanotube membrane (Figure 2-7A) are linear, as are the flux plots for these permeating proteins

for a H-IgG-modified membrane (Figure 2-7B). The fluxes obtained from the slopes of these

plots are shown in Table 2-1. As mentioned above, the selectivity coefficient (a) across of the

membranes was evaluated by comparing the flux of a permeating species that binds weakly to

the carrier to the flux of a permeating species that binds more strongly.

Table 2-1.Fluxes and selectivity coefficients a for anti-M-IgG and anti-H-IgG across the M-IgG
and H-IgG-modified membranes
IgG-attached to the Flux of anti-M-IgG Flux of anti-H-IgG
membrane (pmoles h-1 cm-2) (pmoles h-1 cm-2)

M-IgG 0.60 1.31 2.2

H-IgG 1.33 0.48 2.8



It was observed that for both membranes the flux of the species that binds strongly to the

carrier (e.g., anti M-IgG for the M-IgG-modified membrane) is lower than the flux of species

that binds weakly to the carrier. This is a clear indication that the transport properties for these









Thus, at the high concentration range, the Langmuir model approaches to a constant value

that is equal to Cst,,e, so no more sorption will take place when all the sites are filled. On the

other hand, at very low concentrations, the Langmuir model obeys the correct thermodynamic

boundary condition of Henry's law over infinitely dilute concentration range.215 216

In the limit of zero concentration the Langmuir model leads to the following equation

which satisfies the condition of a linear isotherm (Equation 3-4):

lim- C = CsutebC,ee kDC Fed (3-4)
C-O
Hence, a good estimation of Henry's law coefficient, kD, may be written as the product of

CsItes and b. Consequently, this kD parameter can be used for the calculation of the partition

coefficient (kp), which is defined as the product of kD times the ratio of volumes between volume

inside the membrane (VMembrane) and volume in the free equilibration solution (VFree Solution)

(Equation 3-5).


k = kD VMembrane (3-5)
SFree-Soluhon )
It should be pointed out that for linear isotherms, the Henry's law coefficient, kD, can be

determined from the slope, while for dual-mode adsorption isotherms, it can be determined from

the initial slope at infinite dilute concentration.

The partition coefficients can also be calculated using Equation 3-6, which expresses the

ratio of the intramembrane concentration to external concentration at equilibrium.


k V MembraneCMembrane (3-6)
VFree-Soluhon Free-Soluhon )
All this dual-mode sorption model theory explained until here can be validated using

linearization methods for the Equation 3-2. Several linear plots of the dual-mode sorption

model have been studied,217 and for instance, the equation can be rearranged and transformed

into the Equation 3-7.










Figure 4-6 compares the fluorescence spectra for imidazole functionalized membranes

before and after coordination with CuC12. Simultaneously with the fluorescence enhancement, a

small bathochromic shift of the maximum emission wavelength was observed in the fluorescence

spectra of the copper immobilized membrane from 443 nm to 550 nm.

45000
S40000
35000
1 30000 -
25000 a
0
0 20000
S15000 -
1 0000 -
5000 -


400 450 500 550 600 650 700
Wavelength (nm)
Figure 4-6. Fluorescence spectra of imidazole functionalized membranes (a) before and (b) after
metal ion immobilization using an ethanol solution that was 1 mM in CuC12.

Besides, a dependence on the copper concentration for the fluorescence enhancement was

found (Figure 4-7). As can be seen from Figure 4-7 the addition of the chloride salt of copper

leads to an increase in fluorescence intensity of the functional membrane and at a certain

concentration the fluorescence intensity remains constant. Such an increase in fluorescence

enhancement can result from conformational restriction induced upon binding.274

It is beyond the scope of the present work to discuss in detail the mechanism of interaction

of the functional imidazole membrane with these metal ions. Nevertheless, the observed

fluorescence intensity changes can be reasonably explained on the basis of the complex structure

formed after coordination takes place. It was ascribed that for the situation of the imidazole

membranes coordinated with the nitrate salts, the nitrate anions in [M(Im)n](N03)2 complexes

are strictly counterions, hence they are no part of the sphere of coordination. Therefore, the










Transport across Antibody-Modified Silica Nanotube Membranes

Transport experiments were also carried out across antibody-functionalized nanotube

membranes. For these experiments silica nanotube membranes were modified both with H-IgG

or M-IgG. The results for these experiments are compared in Figure 5-12.

160 3.5
o A o B
0 0 0 A



P 0 2.5
140 A 0
S120 0 a

o (M A
S120 A 0 A
A 0 AAA



00ncentration ws3.nf t 2
< 80 O00ooooooooob ooooob
I I 1 1.5 '
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Time (min) Time (min)
Figure 5-12.Typical plots for (A) changes of fluorescence signal and (B) concentration of
unquenched protein in function of time illustrating the transport of anti-H-IgG-QSY
across (a) H-IgG-, and (b) M-IgG-modified membranes. The feed solution
concentration was 33.3 nM for both experiments.

The data analysis for the transport across antibody-modified membranes based on Stern-

Volmer plots and log-log plots are summarized in Table 5-3.

Table 5-3.Data for the Stem-Volmer quenching constant (Ksv) and slope of the log-log plots for
antibody-modified membranes
Antibody-Modified Membrane* Ksv (nM-') Slope of log-log plots
H-IgG 0.382 -0.0995
M-IgG 0.415 -0.1218
*The feed solution concentration of anti-H-IgG-QSY was 33.3 for both experiments.

The results shown that the fluorescence quenching in the receiver solution was higher for

the transport experiments carried out across M-IgG modified than H-IgG modified membranes.

This is due to the faster rate of transport of anti-H-IgG-QSY across M-IgG as a result of the

lower binding constant between these two proteins. Therefore, it demonstrated that antibody-

modified membranes can be use to modulated the rate of fluorescence quenching for any specific









information about the extent of changes of the surfaces properties resulting from coupling of

biomaterial to the studied interface.142

A typical impedance data is represented in the form of Nyquist plot. The shape of this plot

includes a semicircle region lying on the Zre-axis followed by an incline line. The semicircle

region observed at higher frequencies corresponds to the kinetic limited process. Whereas the

linear part characteristic of the lower frequency range represents diffusional limited process. In

the Nyquist plot Rs can be found by reading the real axis value at the high frequency intercept.

While the sum of Rm and Rs can be found by fitting the semicircle until the real axis value at the

lower frequency intercept.

Nyquist plot present one shortcoming, when looking at any point on the plot, it cannot tell

what frequency was used to record that point. There is another popular representation of the

impedance data called the Bode plot. For this, the impedance is plotted with log frequency on the

x-axis, and both the absolute value of impedance and the phase-shift on the y-axis.141









At first glance observing the plots in Figure 3-14, one can conclude that according to the

dual-mode model there is not total immobilization on the Langmuir-mode sites for these two

permeating proteins across M-IgG modified membranes. If total immobilization occurs in the

Langmuir-mode sites (F=0) as the dual-mode model explain, the permeability become

independent of the concentration. These studies demonstrated a high concentration dependence

on permeability both for anti-H-IgG and anti-M-IgG. These also demonstrated that the M-IgG

carriers in the membrane indeed enhance the permeability at low concentration for both of the

permeating proteins as shown in Figure 3-14.

It was found that the permeability enhancement for these two proteins was only up to a

certain threshold of the binding affinity and appeared to depend strongly on the concentration.

This enhancement in permeability at low concentration is caused by a contribution of the

diffusion due to the Langmuir mode species.204' 211

As a consequence of its lower binding affinity, the permeabilities for anti-H-IgG were

higher than those for anti-M-IgG. At the beginning we set a hypothesis that the differences in

permeation for anti-H-IgG and anti-M-IgG observed could be due to a sort of antifacilitated

transport. This was initially considered based on our previous works,9' 10 which showed that

higher fluxes were always observed for the molecule that binds stronger to the carrier in the

membrane. The mechanism for the facile kinetic conquered, in our previous works, was ascribed

by the sequential binding and unbinding events of the transported permeating species with the

carrier in the membrane. This behavior might be mathematically rationalized in a more detailed

manner by the model described here.

As indicated in Figure 3-14, the experimental and calculated data were in good agreement

at concentrations higher than 2.0 x 10-1 moles per cm3. Nevertheless, at very low concentration











membranes. It has been demonstrated that mercapto compounds are appropriated modification

reagents for this strategy.106 260 263 Furthermore, it has been confirmed that the use of aromatic

thiols leads to degrees of modification which are much higher than those achieved with aliphatic

ones. In addition, when aromatic thiols are used the possibilities of elimination as a side reaction

are eradicated.261

In order to promote the substitution nucleophilic bimolecular mechanism (SN2) the

imidazole modification was carried out in an aprotic solvent (THF). Since aprotic solvents do not

solvate the nucleophiles, the reactivity of the 2-mercaptoimidazole can be enhanced making it

more accessible to approach to the primary halide carbon as the conditions in 3-chloropropyl

silica nanotube modified membranes. A high concentration of 2-mercaptoimidazole was also

used to promote the SN2. It is worth mentioning that some indications that the reaction is taking

place were the sulfur smell during the reaction and the lower pH of the solution at the end of it.

Fluorescence microscopy characterization

We found that after imidazole immobilization the nanotube membranes were fluorescent,

and this observation become in our first control to monitor the success of the modification

chemistry.


25000
o) :
i 20000 -

S15000
09 20000
15000 -
0 I
10000 -,

5000 '



400 450 500 550 600 650 700
Wavelength (nm)
Figure 4-3.Fluorescence spectra of (a) silica nanotube and (b) imidazole functionalized
membranes.










dual-mode model,204 206, 211, 220 since the permeability in terms of the dual-mode model decreases with

increasing feed concentration.221


1.2x10' -

1.0x10'

7 8.0x108

6.0x10-



2.0x108

0.0 ,i- i I t -
0.0 1.0x10" 2.0x10" 3.0x1O0 4.0x10" 5.0x10"
Feed Concentration (moles / cm')
Figure 3-6. Experimental obtained permeability versus feed concentration for anti-M-IgG (A)
and anti-H-IgG (m) across M-IgG modified membrane.

At first sight from the data of concentration dependence of the experimental permeability

in Figure 3-6, we might anticipate that the transport mechanism for these two proteins across M-

IgG modified membranes approaches to the dual mode model mechanism. In order to prove this

transport mechanism for these two proteins across the M-IgG modified membranes, an

evaluation of the sorption and permeation parameters were carried out, and they are discussed

next.

Experimental evaluation of the dual-mode model parameters

The mathematical solution that Koros and Paul found for the permeability coefficients

based on the dual mode model is known as the partial immobilization model (Equation 3-17).204

The use of the model requires the determination of both DD and F, which were evaluated

experimentally by using the permeability coefficients in Figure 3-6. Additionally, to use the

model is a prerequisite to have the information of the three dual-mode sorption parameters, kp,

Csites, and b, which were also determined experimentally via protein-adsorption isotherms.










anti-M-IgG binds weakly to this carrier. We proved these assumptions to be true by conducting

competitive binding assays.

Figure 2-6A shows the fluorescence spectra of the permeating proteins bound to a M-IgG-

modified membrane after equilibration with a solution that was 21.2 nM in both anti-M-IgG and

anti-H-IgG. The membrane shows strong fluorescence from anti-M-IgG and very weak

fluorescence from anti-H-IgG. Calibration data allowed us to convert these fluorescence

intensities to moles of each protein bound to the surface. In agreement with our design

paradigm, 10-times more anti-M-IgG was bound to the M-IgG-modified membrane than anti-H-

IgG. The small amount of anti-H-IgG bound is due to antigen cross reactivity, as a result of the

polyclonal characteristic of the antibodies.169- 172 Similarly, 14-times more anti-H-IgG was

bound to the H-IgG-modified membrane than anti-M-IgG (Figure 2-6B).


35000 45000
A B
S3000040000
.9 30000 b P a
35000
U 25000 -
C 30000
( 20000 25000
15000 20000
1L 15000
10000 -
10000 -
5000 5000
0 I 0
400 500 600 700 800 400 500 600 700 800
Wavelength (nm) Wavelength (nm)
Figure 2-6.Fluorescence spectra of (A) M-IgG- and (B) H-IgG-modified membranes after
equilibration with a solution that was 21.2 nM in both (a) anti-H-IgG, and (b) anti-M-
IgG.

These experiments also confirmed the success of the chemistry immobilization and the

capability of the antibodies immobilized in the nanotube membranes to preserve their bioactive

properties on the porous surface.









Today, recombinant protein purification represents one of the principal protein purification

markets, and many systems have been developed in recent years for the purification of these

proteins.241 250,257 The system proposed here, exploits the advantages of IMAC separations, but

in a nanopore-membrane- platform. The use of imidazole-modified silica nanotube membranes

for the controlled transport of histidine-tagged recombinant proteins is described.

Imidazole is a functional aromatic nitrogen heterocycle important in ligand coordination

chemistry. The structure of imidazole is a five-membered ring with two different nitrogen atoms.

One of the nitrogen is bound to hydrogen, having two lone pair of electrons in an unhybridized

p-orbital donated to the aromatic ring, and it is structurally related with the pyrrole nitrogen. The

second nitrogen structurally related with the pyridine nitrogen, donates its single electron in an

unhybridized p-orbital to the ring, containing also a free lone pair of electrons in a hybrid

orbital.258 In aqueous solution imidazole behaves like a weak organic base with pKa = 6.99.259

Therefore, the lone pair of electrons in the pyridine-like nitrogen atom, are only involved in

coordination when the pH of aqueous solutions is close or higher than 6.99. It is worth

mentioning, that the mode of coordination of imidazole can be also influenced by the nature of

the metal.

In this investigation, the silica nanotube membrane was prepared by sol-gel template

synthesis of silica nanotubes within the pores of a nanopore alumina template. Imidazole was

attached to the inside walls of chloropropyl-modified silica nanotubes by nucleophilic

substitution reaction with 2-mercaptoimidazole in anhydrous tetrahydrofuran. The modified

membrane was then exposed to solutions of the desired divalent metal ion (copper, nickel, cobalt

or zinc) to attach the metal via coordination with the imidazole. The histidine-tag on the
































To my parents Herlinda and Luis, my wife Jessica, and my daughter Isabela









LIST OF FIGURES


Figure pe

1-1. Scanning electron micrograph pointing up the honeycomb structure of nanoporous
alu m in a ................... ...................2...................5..........

1-2. Structure of nanoporous alumina. .................................................................................. 26

1-3. Electrochemical cell setup for nanoporous alumina growth............................................28

1-4. Scanning electron micrographs of a nanoporous alumina membrane anodized at 50 V
in 5% oxalic acid, (A) before and (B) after elimination of the barrier layer. ..................29

1-5. Scanning electron micrograph of a highly-ordered nanoporous alumina membrane
obtained at 50 V in 5% oxalic acid. ...... ......................................................................30

1-6. Outline of the steps involved in the sol-gel synthesis............. ........... ............... 32

1-7. Fluorescence spectra obtained from the biofunctionalized (a) alumina template and
(b) templated silica nanotube membranes with anti-Human IgG labeled with Alexa
4 8 8 .......................................................... .................................. . 3 4

1.8. Reactions involved in the alkoxysilane functionalization. Scheme obtained at
http://www.dowcoring.com/content/publishedlit/SILANE-GUIDE.pdf ......................35

1-9. Types of specialized transport in living membranes (A) carriers, and (B) channels.........37

1-10. Flux concentration dependence for passive and facilitated transport.............................38

1-11. Concentration gradients formed across the membrane during passive and facilitated
tran sport ........................................................ ..................................39

1-12. Basic structure of the immunoglobulin molecule. .......... ..................... ........... .... 42

1-13. Intermolecular attractive forces involves in the antibody-antigen binding. Scheme
obtained from www.fleshandbones.com/readingroom/pdf/291.pdf. ............................44

1-14. Jablonski diagram of the electronic states during the fluorescence process....................47

2-1. Immobilization of antibodies on the template synthesized silica nanotube
m em branes. ................................................................................56

2-2. Flow-through fluorescence system used to measure the flux of the permeating
proteins. ................................................................. ........... ............ 59

2-3. Electron micrographs of (A) branch side, (B) solution anodization side, and (C)
cross-section of the alumina template membranes, and (D) silica nanotubes after
rem oval from the m em brane ...................................................................... ..................63









The ribosome produces an unfolded linear chain of denatured peptides. Then, the chain of

aminoacids folds to form structures such as alpha-helices and beta-sheets, which are the elements

of the protein secondary structure. The various secondary structures then fold into a compact

tridimensional structure, which correspond to the tertiary structure of a protein. This tertiary

structure, in many cases contains the functional form of the protein. In some cases, the

combination of two or more subunits of tertiary structure peptides may be required for the

functional form of the protein; this assembly is called the quaternary structure.

Antibodies

Antibodies, also called immunoglobulins, are proteins produced by the immune system for

the defense of the organism against foreign species.120 They exhibit very specific binding

capacity for specific molecules. Immunoglobulins are divided in five classes: IgG IgM, IgA,

IgD, and IgE.120 They vary in size, charge, aminoacid composition and carbohydrate content.

Immunoglobulins are made of two identical heavy and two identical light chains of aminoacids,

forming a characteristic Y-shape (Figure 1-12).


antigen-binding site VH domain


Domain Fab fragments
-- VL domain


hinge region
light chain Fc fragment




heavy chain


Figure 1-12.Basic structure of the immunoglobulin molecule.









as paradigm to show the power of the template method for preparing nanotubes for

biotechnological applications.o5 Silica nanotubes are easy to make, and they can be derivatized

with an enormous variety of different chemical functional groups using simple silane chemistry

with commercially available reagents.

Mitchell et al. developed the concept of smart nanotubes that can be used for biological

extraction and biocatalysis.38 Nanotubes removed from the template and differentially

functionalized both in the inner and outer nanotube surfaces were used for this application. Such

differentially functionalized nanotube was used as smart nanophase extractor to remove specific

molecules from solution. Thus, hydrophobic functionalized nanotube on the inside, with a

hydrophilic nanotube exterior was used to extract lipophilic compounds from aqueous solutions.

In the same work, Mitchell et al. showed that nanotube functionalized with an enantioselective

antibody selectively extracts the enantiomer of a drug molecule that binds to the antibody from

the racemic mixture.38

Lee et al. demonstrated the use of silica nanotube membranes for enantioseparations.9 They

found that nanotubes embedded within the pores of the alumina template and functionalized with

a Fab fragment antibody that binds the RS isomer of a drug molecule over the SR isomer,

facilitated the transport of the RS enantiomer. They also found that the transport selectivity and

binding affinity of RS over SR could be tuned by controlling the inside diameter of the

membrane containing silica nanotube, and by addition of DMSO to the protein buffer solution,

respectively.

Gasparac et al. explored the template synthesized nano test tubes as an alternative payload-

release strategy by using the template synthesis within the pores of nanopore alumina films.51

These hollow nanotubes were closed on one end and open on the other. Therefore, if these nano











binds more sites but those are displaced when the anti-M-IgG molecules catch up to them.

Additional confirmation for this conclusion is obtained by comparing the binding capacities for

these two proteins (Table 3-2).

1.6x109
? 8.0x108 A
0 1.4x10 -
E E
I 1.2x10-9 A
7.0x10
MA
.- 1.0x109
o .o
o 0
S6.0x10 8.0x10" A
o

0 5.Ox1 O A
4.0x10" A
0.0 1.0x101 2.0x10" 3.0x10 4.0x10" 5.0x10' 0.0 1.0x10.' 2.0x10.' 3.0x10.' 4.0x10(O 5.0x10~
Feed Concentration (moles/cm3) Feed Concentration (moles/cm3)
Figure 3-15.Experimental obtained diffusion coefficients versus feed concentration for (A) anti-
H-IgG and (B) anti-M-IgG across M-IgG modified membrane.

At the region of low limit concentration it was also observed that as the concentration

increases the permeability increases as well (Figure 3-14). It was deduced that between the two

set of binding sites (well-oriented and random) present in M-IgG membranes, there is a transition

stage in concentration before the random binding sites with a higher binding capacity become in

a positive cooperative binding sites. This transition involves a mutual cooperation of these two

binding sites in the membrane. This effect was observed at concentrations lower than 2.0 x 10-11

moles per cm3 (Figure 3-14). These results are in favorable agreement with the positive

cooperative contribution observed in the adsorption isotherms in Figure 3-8. Besides as

mentioned previously, the Langmuir binding sites were the dominant factor in determining the

total solubility of anti-H-IgG and ant-M-IgG in the M-IgG modified membranes. And regardless

of the protein these Langmuir sorption sites contributed the most to the solubility at low

concentration, as was demonstrated in Figure 3-12.









Since the values Ka, T and R the gas constant (8.314 Joules moles- K-1) are known, by

using Equation 2-2 was possible to calculate AGo. The results of these energetic calculations are

summarized in Table 2-2.

Table 2-2.Thermodynamic parameters of AGo associated with the affinity interaction of anti-M-
IgG and M-IgG-modified membranes as a function of the temperature
Temperature Temperature Ka AGO
(C) ( K) (M-1 ) (KJ / moles)
23 296 3.76 x 108 -48.59

32 305 7.81 x 107 -46.08

42 315 3.19 x107 -45.24



As can be seen from Table 2-2, the resulting AGo values were negative, which indicate that

the process is exothermic and proceed spontaneously toward the formation of the complex

protein-antibody. In addition, we can see how small changes in free energy generate significant

changes in the binding affinity constant. Therefore, these changes in energy can be associated

with the variations in the fluxes observed in Figure 2-17 when the temperature was changed.

Now that that the changes of free energy as a function of the temperature were established,

from a plot of AGo vs. Tthe values of AHo and ASo can be extracted from Equation 2-1, by

assuming that AHo is independent of temperature. From the slope and intercept of the obtained

plot with a linear correlation of 0.918 the estimated values for ASo and AHo were 0.17 0.06

KJ/mol and -99.95 17.12 KJ/mol, respectively. Those estimated values confirm the spontaneity

of this biomolecular reaction toward the formation of the protein-antibody complex. Interesting

is that ASo is still positive even though the loss of rotational-translational degrees of freedom

after complex formation. This effect has been extensively studied and had been associated with









aprotic medium. It was demonstrated that transport of histidine-tagged recombinant proteins

across these imidazole-modified silica nanotube membranes is controlled by the nature of

immobilized metal ion on the imidazole surface. Besides, the extensive characterization of the

imidazole functionalized membrane allowed the understanding of how coordination chemistry

with different divalent metal ions affects the functional behavior of the membrane and transport

properties during the permeation of six-histidine recombinant proteins. Furthermore, additional

results shown that by the appropriate choice of the transition metal ion for coordination with the

imidazole ligand, the binding strength and optical properties within the functional membrane are

also modified. Hence, these studies revealed how the presence of the immobilized metal ion

affects the functionality of the imidazole membrane and its transport properties. The metal ion

affinity nanotube membrane described here has the potential to be a versatile analytical platform,

where different metal ions can be immobilized on the same prevailing conditions, additionally

offering the possibilities of regeneration without apparent loss of the metal coordination

properties of the functional membrane. These studies demonstrate an alternative way of

immobilized metal ion affinity by using nanotube membranes.









In the Martin group there is a commonplace methodology for the electroless deposition of

gold within the pores of polycarbonate (PC) 10, 37 and polyethylene terephthalate (PET) 16, 46

template membranes. Depending on the plating time, it is possible to generate gold nanowires or

nanotubes. Thus, in the nanotube situation, by controlling the plating time, the inside diameter

can be varied. The electroless deposition of gold on the pore wall of these polymeric membranes

is currently the best known method to precisely control the inside diameter of the template-

synthesized nanotubes.47

Martin et al. have shown that because the inside diameter of these gold nanotube

membranes can be small relative to molecular dimensions, the nanotube membrane can be used

to cleanly separate small molecules on the basis of molecular size.47 By varying the rate of the

plating reaction the diameter of the nanotubes was smaller in the faces of the template

membrane. Thus, the flux and selectivity were determined for the bottleneck of the nanotubes.

Together with size-based selectivity, they also demonstrated that generic chemical transport

selectivity can be introduced into this membranes.48'49

More recently, Martin et al. have investigated gold nanotube membranes where the inner

pore surfaces were systematically modified for biochemical functionalization to accomplish

selective permeation of DNA.10 They demonstrated a DNA-functionalized nanotube membrane

for the separation of DNA by single-base mismatches selectivity via facilitated transport.10 These

DNA-functionalized nanotube membranes selectively recognize and transport the DNA strand

that was complementary to the immobilized hairpin-DNA (carrier) within the pores, relative to

DNA strands that are not complementary to the immobilized DNA.

Another synthetic method used for the manufacture of nanotubes is the sol-gel chemistry to

produce nanotubes of silica and other inorganic materials. The silica nanotubes have been used









[180]Dancette, O. P.; Taboureau, J.-L.; Toumier, E.; Charcosset, C.; Blond, P. Journal of
Chromatography, B: Biomedical Sciences and Applications 1999, 723, 61-68.

[181]Rutherford, S. W.; Do, D. D. Adsorption 1997, 3, 283-312.

[182]Paul, D. R. Journal of Polymer Science, Polymer Physics Edition 1969, 7, 1811-1818.

[183]Tsuru, T.; Izumi, S.; Yoshioka, T.; Asaeda, M. AIChE Journal 2000, 46, 565-574.

[184]Nomura, T.; Nakao, S.; Kimura, S. Kagaku Kogaku Ronbunshu 1987, 13, 811-817.

[185] Chandler, D. Nature 2005, 437, 640-647.

[186]Lin, F.-Y.; Chen, W.-Y.; Hearn, M. T. W. Journal of Molecular Recognition 2002, 15,
55-93.

[187]Palma, E.; Klapper, D. G.; Cho, M. J. Pharmaceutical Research 2005, 22, 122-127.

[188]Norde, W.; Lyklema, J. Journal of Colloid and Interface Science 1979, 71, 350-366.

[189]Norde, W. Advances in colloid and interface science 1986, 25, 267-340.

[190]Norde, W.; MacRitchie, F.; Nowicka, G.; Lyklema, J. Journal of Colloid and Interface
Science 1986, 112, 447-456.

[191]Hamilton, R. G. Production and epitope location of monoclonal antibodies to the human
IgG subclasses. In Molecular Aspects of Immunoglobulin Subclasses; Pergamon Press:
Oxford, 1990, pp 79-91.

[192] Schmuhl, R.; Keizer, K.; van den Berg, A.; ten Elshof, J. E.; Blank, D. H. A. Journal of
Colloid and Interface Science 2004, 273, 331-338.

[193] Saksena, S.; Zydney, A. L. Biotechnology and Bioengineering 1994, 43, 960-968.

[194]Zydney, A. L.; Pujar, N. S. Colloids and Surfaces, A: Physicochemical and Engineering
Aspects 1998, 138, 133-143.

[195]Lopez, M. F.; Melov, S. Circulation Research 2002, 90, 380-389.

[196] Sun, L.; Crooks, R. M. Langmuir 1999, 15, 738-741.

[197]Bath, B. D.; Lee, R. D.; White, H. S.; Scott, E. R. Analytical Chemistry 1998, 70, 1047-
1058.

[198]Baltus, R. E. Journal of Membrane Science 1997, 130, 157-172.









(Ro), and this radius corresponds to the distance at which fluorescence resonance energy transfer

from the donor dye to the acceptor dye is 50% efficient (Equation 5-4).131,298

E = + 6 -> r = distance between fluorophore and quencher (5-4)

Ro
The magnitude of Ro is dependent on the spectral properties of the donor and acceptor dyes

and the typical value reported for the FITC and QSY-7 pair is 61 A.303

The suitable proximity for the fluorescence quenching process between these fluorescence

donor and acceptor molecules was accomplished through a molecular recognition reaction using

two antibodies with specific affinity for each other. This antigen-antibody immunoassay is

illustrated in Figure 5-4.






10'(D
0 P + i4


tWH40g G4Y H4gGTC Compl.x An n.nAnttbody
Figure 5-4.Molecular recognition reaction used to promote the fluorescence quenching.

Thus, the flux of anti-H-IgG-QSY was monitored by using a known concentration of H-

IgG-FITC as fluorescence emitter in the permeation compartment.

Verification of the Fluorescence Quenching System via H-IgG-FITC and anti-H-IgG-QSY
Reaction in Solid State

In order to verify that the molecular recognition reaction between H-IgG-FITC and anti-H-

IgG-QSY leads to an inhibition of the fluorescence process, a silica nanotube membrane was

modified with H-IgG-FITC. The fluorescence signal for this modified membrane was observed

under microscope before and after six hours of exposure to a solution of anti-H-IgG-QSY that









[93] Hench, L. L. Sol-Gel Silica-Properties, Processing and Technology Transfer; Noyes
Publications: New Jersey, USA, 1998.

[94] Iler, R. K. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface
Properties and Biochemistry, John Wiley & Sons, 1979.

[95] Pyell, U.; Stork, G. Fresenius'Journal ofAnalytical Chemistry 1992, 343, 576-581.

[96] Jal, P. K.; Patel, S.; Mishra, B. K. Talanta 2004, 62, 1005-1028.

[97] Kiselev, A. V.; Lygin, V. I. Infrared Spectra of Surface Compounds; Wiley: New York,
1975.

[98] Kiselev, A. V. Kolloidnyi Zhurnal 1936, 2, 17-26.

[99] Chuang, I. S.; Maciel, G. E. Journal of Physical Chemistry B 1997, 101, 3052-3064.

[100]Grushka, E.; Kikta, E. J., Jr. Analytical Chemistry 1977, 49, 1004A-1008A, 1010A,
1012A, 1014A.

[101] Van der Voort, P.; Gillis-D'Hamers, I.; Vansant, E. F. Journal of the Chemical Society,
Faraday Transactions 1990, 86, 3751-3755.

[102] Tundo, P.; Venturello, P.; Angeletti, E. Journal of the American Chemical Society 1982,
104, 6551-6555.

[103]Plueddemann, E. P. Silane CouplingAgents, 2nd ed.; Plenum Publishing Corporation:
New York, 1991.

[104]Loy, D. A.; Baugher, B. M.; Baugher, C. R.; Schneider, D. A.; Rahimian, K. Chemistry
ofMaterials 2000, 12, 3624-3632.

[105]Arkles, B. Chemtech 1977, 7, 766-778.

[106]Dias Filho, N. L.; Gushikem, Y. Separation Science and Technology 1997, 32, 2535-
2545.

[107]Lehninger, A. L. Principles ofBiochemistry, 2nd ed.; Worth Inc.: New York, 1993.

[108]Kemp, N. J.; Noble, R. D. Separation Science and Technology 1983, 18, 1147-1165.

[109]Lakshmi, B. B.; Martin, C. R. Nature 1997, 388, 758-760.

[ 110]Noble, R. D.; Koval, C. A. Materials Science of Membranes for Gas and Vapor
Separation 2006, 411-435.









The decision of using organosilanes with one or more than one hydrolyzable groups for

silane functionalization depends on the particular set of properties desired in the new

functionalized surface. Thus, when a hydrophobic surface is preferred, organosilanes with one

hydrolyzable group are used.103 If surfaces with higher degree of surface modification are

desired, organosilanes with two or three hydrolyzable groups are used.103

An alternative way of generating one monolayer of the desired silane monomer on silica

substrates is by conducting silane chemistry of chlorosilanes in aprotic solvents.105, 106 In

conditions where the aprotic solvent and surface are water free, no alkoxysilane is produced, and

no polymerization of the chlorosilane can take place. Therefore, the reaction must be promoted

by nucleophilic substitution; where the nucleophile hydroxyl groups in the surface attack the

primary halide carbon in the chlorosilane moiety. Surface modification in those cases takes

longer time, usually 12-24 hours.

Thereby, surface modification via silane functionalization provides a unique opportunity to

regulate the interfacial properties of silica substrates while their basic structure and mechanical

strength are preserved.

Nature's Transport across Membranes: An Inspiration for Membrane-Based Separations

Biological membranes are vital components of all living systems, forming the external

boundary of the living cells and the internal organelles into the cytoplasm.107 They are

constituted mainly of various lipids arranged in a bilayer that imparts a fluid character. Thus,

both lipids and proteins are free to move within a bilayer. The proteins embedded in the bilayer

and the carbohydrates attached to its surface, facilitate communication and transport across the

membrane. In the living membranes, there are two types of membrane proteins which are

involved in transport and signaling activities. Those are the peripheral and integral (or intrinsic)










estimated number of OH groups accessible on porous silica surface is between 4 and 5 OH

groups per nm2 with an average surface area of 20.4-21.7 A2 per silanol group.94 99-101 In the

research reported here, we took advantage of this special attribute of the higher silanol groups on

silica surface for further silane functionalization, because when compared with alumina, the

number of estimated hydroxyl groups accessible on alumina surface is only 0.1-2.5 OH groups

per nm square.30'102

In order to prove the benefit of the silica nanotubes in the alumina template, some

fluorescence experiments were performed. For these studies, both a templated silica nanotube

membrane and alumina membrane were biofunctionalized with anti-Human IgG fluorescent

labeled with Alexa 488. This anti-Human IgG was covalently attached via the Schiff s base

reaction to the walls of the aldehyde functionalized nanotube membranes.9 These results are

shown in Figure 1-7, where the Alexa 488 fluorescence spectra from alumina membrane and

from silica nanotube membrane are compared. The Alexa 488 fluorescence signal from silica

nanotube templated membrane was approximately 6 times higher than the analogous alumina

membrane. Hence, these results suggest that the deposition of silica nanotubes is essential to

provide higher degree of functionalization on the membranes.

60000 b
Ssoooo -

4 30000



0 I -- -



Wavelength (nm)

Figure 1-7.Fluorescence spectra obtained from the biofunctionalized (a) alumina template and
(b) templated silica nanotube membranes with anti-Human IgG labeled with Alexa
488.
g 20000
a


400 500 500 700 Soo
Wavelength (nm)

Figure 1-7.Fluorescence spectra obtained from the biofunctionalized (a) alumina template and
(b) templated silica nanotube membranes with anti-Human IgG labeled with Alexa
488.














", ------------ piB




Figure 5-1.Flow-through fluorescence system used to measure the changes of fluorescence
intensity in the receiver solution.

This system consisted of an L-7485 fluorescence detector (Hitachi). The flow stream was

delivered using an L-7100 HPLC pump (Hitachi). We discovered that if the receiver solution

was sent continuously through the fluorescence detectors during the entire course of a transport

experiment, significant photobleaching of the FITC dye occurred. To prevent this, we turned the

pump on to deliver receiver solution to the detectors for only 3 minutes per hour of transport

experiment. That is, we used a duty cycle of 57 minutes with pump off followed by 3 minutes

with pump on (flow rate = 3 mL per min) to collect fluorescence data. The whole system was

controlled using Chromeleon 6.50 chromatography management software (Dionex) through a

UCI-100 universal chromatography interface (Dionex).

FITC was excited at 495 nm, and the emission was detected at 515 nm. Calibration curve

was obtained with the same flow through system on solutions containing known concentrations

of the flourophore-labeled antibody. The calibration curve was used to obtain the correct

concentration of the unquenched fluorescent labeled protein in solution.

Results and Discussion

Experimental Principle Used for the Fluorescence Quenching Detection

For these studies, the transport analysis design of the anti-H-IgG-QSY across silica

nanotubes was mainly based on the FRET process. Therefore, the analysis required the use of a










transport experiments, the feed solution concentration was in all cases 2.65 nM in both of the

permeating proteins anti-H-IgG and anti-M-IgG (Figure 2-11A).

0.16
A 4.5 -
0.14 A 4
o.r / 34.0 -
E 0.12 3.5 -

0.10 3.0
E 0
0.08 -2.5
X '^ 2.0 --
S0.06 .
S1.5- *
0.0 1.0
201200 200120 100 200 20/200 200/20 100 200
Pore Size (nm) Pore Size (nm)
Figure 2-11.Effect of the pore size on (A) the flux, and (B) selectivity coefficients for M-IgG
modified membranes. The feed solution concentrations was 2.65 nM both in anti-H-
IgG and anti-M-IgG.

At this specific feed concentration (2.65 nM) not appreciable differences for the flux of

anti-M-IgG across these M-IgG modified membranes were observed. However, lower fluxes

were obtained for anti-H-IgG as the pore size decreases (Figure 2-11A). This reduction on flux is

due mainly to the resistance that the transported antibodies experience in the pores that arises

from the hydrodynamic interaction of the protein molecules with the wall of the pore, which

include far field and near field interactions.177 178 Interestingly, the flux for the 200 nm

membrane decreases slightly compared with the 100 nm membrane. This reduction in flux was

ascribed to the fact that 200 nm pore size membranes are much easier of functionalize than 100

nm one. Therefore, the data suggests that the number of binding carrier sites immobilized in this

membrane is possibly higher than the 100 nm, which in turn decreases the effective pore size of

the membrane. Additional evidence for this conclusion was obtained from the comparison of

fluorescence spectra of the membranes after transport experiments, which showed higher

fluorescence intensities for the 200 nm M-IgG modified membrane.









also observed. In the latter region, the differences between the bending vibrations of the O-H for

the alumina membrane and the silica nanotube membrane were not significant (spectra a, and b

in Figure 4-11). This was reasonable because part of the OH groups for the alumina are replaced

for the silanols OH groups in silica; besides silica surface introduce higher silanol groups on

surface for further silane functionalization.30 94, 99-102

In the case of chlorosilane modification (spectrum c in figure 4-11), lower absorption for

the silane-modified membrane was observed in the bending O-H region compared with alumina

and silica. This difference in transmittance clearly confirmed the success of the silanization

reaction where the O-H free groups in the alumina matrix and the silanol groups in the silica

nanotubes were replaced by the condensation reaction with 3-chloropropyltrimethoxysilane.

The differences in transmittance were even more evident for the imidazole functionalized

membrane (spectrum d in Figure 4-11). Additionally this exhibited two observable absorption

bands at 1118 cm-1 and 474 cm-1 assigned respectively to the anti-symmetric C-S-C, and to the

C-S stretching vibrations, which confirmed the imidazole functionalization.

These differences in the intensity of the bands might be attributed once more to changes of

the electric dipole moment after each chemistry modification.287 Besides, each chemistry

modification introduces a variety of vibrational motions, and degrees of freedom characteristic of

its component atoms that should produce a net change in the dipole moment. It should be also

considered the fact that may be after the modifications the membranes become less hydrophilic

than the bare alumina template, thereby the contribution of water on surface must be less

significant, even though all the membranes were dried at the same conditions.

These transmission FTIR data in Figure 4-11 may perhaps be interpreted on the basis of

the infrared absorption spectral region. Thus, the shorter wavelengths can penetrate less the









BIOANALYTICAL APPLICATIONS OF AFFINITY-BASED NANOTUBE MEMBRANES
FOR SENSING AND SEPARATIONS




















By

HECTOR MARIO CAICEDO


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2008









CHAPTER 3
MODELING AND CHARACTERIZATION OF THE TRANSPORT PROPERTIES OF
PROTEINS ACROSS ANTIBODY-FUNCTIONALIZED NANOTUBE MEMBRANES BY
USING THE DUAL-MODE MODEL

Introduction

In the design and development of more effective membranes for membrane-based

separations, the value of the transport mechanism must be considered in order to understand the

different interactions that govern transport through the pores of the membranes. In addition, the

transport mechanism, supported by mathematical models and experimental evidences provides a

grater understanding of the underlying transport phenomena.

The separation properties of molecules through membranes are based on the difference of

permeation. Therefore, it is important to understand permeation behavior, not only to understand

the permeation mechanism but also to elucidate the separation properties.

In order to elucidate the transport properties across selective membranes a wide variety of

models have been developed.196- 202 However, a model that explains the sorption and permeation

mechanism is of fundamental importance to fully understand the transport properties in

membrane separation.

The well-known dual-mode model allows evaluation of both sorption and permeation

mechanism, for those cases where the model applies. The dual-mode model was initially

developed by Vieth and Sladek203 and later modified by Paul and Koros204 to describe the

sorption and permeation of gas molecules in glassy polymeric films. In literature there are

various modified dual-mode models that have also been explored.182, 205 208

The model assumes that the membrane contains two distinct regions in which sorption can

occur.204 One region exhibits solubility of the molecule based on Henry's law and the second









6-8. Nyquist plots (Zim vs. Zre) of impedance measurement in highly purified water: (a)
alumina, (b) after synthesis of silica-nanotube, (c) after silanization, (d) after
biotinylation, (e) after streptavidin immobilization, (f) after PNA immobilization,
and (g) after DNA hybridization. The plots were recorded applying 5 mV alternative
voltage in the frequency range from 1 MHz to 0.05 Hz. ............................. ............... 181

6-9. Nyquist plots (Zim vs. Zre) of impedance measurement in highly purified water for a
PNA-modified nanotube membrane: (a) before, and (b) after 10 minutes of DNA
hybridization. The plots were recorded applying 5 mV alternative voltage in the
frequency range from 1 MHz to 0.05 Hz............................................... 183

6-10. Effect of the membrane pore size on the membrane resistance: (A) 200 nm, and (B)
100 nm. The resistance values were obtained from the impedance data recorded by
applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz.......184

6-11. Study of the hybridization process stability with time: (A) unmodified control
alumina membrane, and (B) PNA-modified membrane. The resistance values were
obtained from the impedance data recorded by applying 5 mV alternative voltage in
the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of
com plem entry D N A ...................... ........................ .. .... .................. 185

6-12. Nyquist plots (Zim vs. Zre) of impedance measurement in highly purified water for a
PNA-modified nanotube membrane: (a) before, (b) after adding 3 nmoles of DNA,
and (c) after 6 nmoles of DNA. The plots were recorded applying 5 mV alternative
voltage in the frequency range from 1 MHz to 0.05 Hz. ............................. ............... 186

6-13. Reproducibility study of the PNA-modified nanotube membranes, before and after
the hybridization process. The resistance values were obtained from the impedance
data recorded by applying 5 mV alternative voltage in the frequency range from 1
MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA. .................... 187










development of functionalized-membranes with a selective transport function executed through a

specific molecule "carrier" to modulate the transport during membrane-based separations.9'10

The "carrier" is a mediator that selectively binds to the specific chemical solute, the target

compound to be separated from a sample solution. Depending on its kinetic interactions with the

target compound, the carrier allows enhanced or retarded transport across the membrane.

Here we describe the use of antibody-functionalized nanotube membranes for the

separation of proteins. Since antibodies that selectively bind proteins are well known,165, 166 it

occurred to us that this antibody-based approach might provide a general route for preparing

highly selective membranes for protein separations. To test this hypothesis, membranes

containing silica nanotubes were prepared using the template synthesis method.9' 25, 85 In the

inside walls of these nanotubes, purified mouse immunoglobulin IgG (M-IgG) or human

immunoglobulin IgG (H-IgG) were covalently attached, using an aldehyde-containing silane

reagent (Figure 2-1).9






NH,
0 N-H N
II Ij -H20 II
H-C -- H-C-OH H-C
Covalent Schlff
coupling base by
Aldehyde-Silane dhydraton
Nanotube Membranes
Figure 2-1.Immobilization of antibodies on the template synthesized silica nanotube membranes.

To study the functional behavior of the IgG-functionalized membranes, competitive

transport experiments were done, using antibodies of anti-mouse IgG (anti-M-IgG) and anti-

human IgG (anti-H-IgG) fluorescent labeled with Alexa 488 or Alexa 594. The studies of

transport properties for these fluorescent labeled antibodies demonstrated that the nature of

immobilized antibody changes the functional behavior of the functionalized membranes and the









(Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes.

The purified mouse immunoglobulin (M-IgG) immobilized within the silica nanotube

membranes was obtained from Sigma Aldrich. Membranes modified with purified human

immunoglobulin (H-IgG) obtained from Sigma Aldrich were also prepared. A trimethoxysilyl

propyl aldehyde (United Chemical Technologies) was used to attach the M-IgG or H-IgG to the

nanotubes. A purified human immunoglobulin conjugated with the FITC dye (H-IgG-FITC)

obtained from Sigma Aldrich was used as emitter during the fluorescence quenching detection. A

purified rabbit immunoglobulin conjugated with the FITC dye (R-IgG-FITC) obtained from

Sigma Aldrich was used as control emitter during the fluorescence quenching detection. To

investigate the transport properties across silica nanotube membranes the whole antibody raised

in goat against H-IgG was used was used as permeating protein. This anti-human IgG (anti-H-

IgG) with specific affinity for the Fab fragments of the H-IgG was obtained from Sigma Aldrich.

In order to observe the fluorescence quenching of the donor-fluorescent labeled protein through a

molecular recognition reaction, the anti-H-IgG permeating protein was conjugated with the

QSY-7 quencher. This protein labeling was accomplished via acylation reaction of an N-

Hydroxysuccinimidyl (NHS) ester of QSY-7 to the primary amine residues of the protein to form

a stable amide bond. This NHS ester of QSY-7 was purchased from Molecular Probes (Eugene,

OR). The BioGel P-30 resin provided in the amine-reactive protein labeling kit obtained from

Molecular Probes was used for purification of the anti-H-IgG conjugated with QSY-7.

SuperBlock blocking buffer solution in phosphate-buffered saline was obtained from Pierce.

SuperBlock solution contains a proprietary protein for blocking binding sites for non-specific

protein adsorption. Tween 20 was obtained from Sigma Aldrich. All other chemicals were of









Frequently, the introduction of a metal binding site in the expressed recombinant peptide is

placed for further purification. It has been demonstrated that an aminoacid sequence consisting

of 6 or more histidine residues in a row operate as a metal binding site for recombinant

proteins.124, 125 This hexa-His sequence is called a His-Tag. By using commercially available

vectors, the His-Tag sequence can be placed on the N-terminal of a target protein. This expressed

His-Tag recombinant protein can be simply purified by using immobilized-metal affinity

chromatography (IMAC).126, 127

Synopsis and Analytical Visualization of the Instrumental Techniques

Fluorescence spectroscopy

The transport performance in membrane-based separation is normally investigated by

permeation experiments of the systems in buffered-aqueous solutions of the permeant solutes to

separate. One of the analytical methods used in this investigation to characterize the fluxes of the

transported proteins across the membranes is fluorescence spectroscopy. This was achieved

through the use of simple photophysical probes such as fluorescent-labeled antibodies. The use

of fluorescence techniques for the study of proteins in solution is widespread. Fluorescence

spectroscopy is a very sensitive technique, making it possible to perform experiments on protein

solutions at very low concentration, minimizing the amount of molecules required for the

analysis and eliminating problems associated with aggregation and low solubility.128- 130

Fluorescence is a characteristic process that occurs in certain type of molecules called

fluorophores that have the property of absorbing energy and then immediately releasing this

energy in the form of light.131 The fluorescence process basically consists of three stages:

excitation, lifetime of the excited state, and fluorescence emission.









BIOGRAPHICAL SKETCH

Hector Mario Caicedo was born in Cali, Colombia. He graduated with a B.S. in chemistry

from the Universidad del Valle, in Cali. For his bachelor's thesis he conducted research in

analytical chemistry focusing on environmental pollutants under the direction of Dr. Nelly de

Palacios (his mentor in chemistry), developing an analytical methodology for the evaluation of

chromium in urine and blood by using electrothermal atomic absorption spectroscopy (ETAAS).

After that, he worked as a chemist in the industrial field, for a combined time of almost four

years, for both a biotechnology and pharmaceutical company. Then, he decided to go back to

academia to conduct his Master of Science in analytical chemistry at the Universidad del Valle.

In the summer of 2000, he visited the University of Florida as an international exchange student,

conducting research in electrochemistry, as a part of his master's research work, using carbon

fiber ultramicroelectrodes under the direction of Dr. Anna Brajter-Toth. After returning to the

Universidad del Valle, he conducted research in electroanalysis of xanthine by using carbon fiber

ultramicroelectrodes with surfaces electrochemically activated and polymer-modified, in the

group of Dr. Alonso Jaramillo. Once he graduated with a M.S. in chemistry from the University

del Valle, he continued his graduate studies in the Department of Chemistry at the University of

Florida to pursue his Ph.D. He worked under the supervision of Dr. Charles R. Martin, where he

conducted research in bioanalytical applications of affinity-based nanotube membranes for

sensing and separations. He completed his final work in April 2008, when he received a Doctor

of Philosophy degree.


212









54%, 42%, and 53%, respectively. Amazingly was to observe the stability of these membranes

after three days of the duplex formation during the hybridization process.


THREE PNA-MODIFIED NANOTUBE MEMBRANES

60000 54618
50000 41831
40000 35097 25459
E 30000 19440 20435 22126
1 20000 11913 15892
10000


.6 Z6 i- ,




Figure 6-13.Reproducibility study of the PNA-modified nanotube membranes, before and after
the hybridization process. The resistance values were obtained from the impedance
data recorded by applying 5 mV alternative voltage in the frequency range from 1
MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA.

The results presented in these studies demonstrated the good stability and reproducibility

of the PNA-modified nanotube membrane describe here.

Conclusions

It was described the use of PNA-modified nanotube membranes to study a label-free

detection for DNA hybridization. This was achieved by exploiting the unique structural and

hybridization features of PNA. This approach was characterized by observing the changes in the

membrane resistance using impedance spectroscopy. This method allowed carrying out an in situ

characterization of the interfacial properties of the modified nanotube membranes during the

hybridization process in the duplex formation PNA-DNA. The novel idea was based on the fact

that PNA is a neutral molecule; thereby during hybridization events achieved in just highly

purified water without any electrolyte in the solution, the only source of new charges in the

system during the hybridization process emerge from the DNA molecules. This event exhibited a









coefficient kp. Since, the sorption coefficient given in the Equation 3-1 is a gas-phase coefficient;

in solution therefore the sorption coefficient should be considered as a liquid-phase coefficient.

The adapted dual-mode model is shown in Equation 3-2.

C bC
C = CD C = kpCfeed + tes feed (3-2)
l+bC feed
Where C is the total protein concentration in the membrane, CD is the protein concentration

based on Henry's law sorption, CH is the protein concentration based on Langmuir sorption.

Whereas, kp is defined as the partition coefficient, b and Cstes are the Langmuir affinity constant

and maximum capacity parameters, respectively.

The Langmuir affinity constant, b, characterizes the sorption affinity for a particular

protein-antibody membrane system, and Cstes is used to measure the amount of antibody

immobilized in the membrane.

The three dual-mode sorption parameters, kp, Cstes, and b, can be determined through the

experimental evaluation of protein-adsorption isotherms. The estimation for Cstes, and b,

sorption parameters can be accomplished by analyzing the data according to the monolayer

Langmuir model (Equation 3-3):167

CBd -C Stes Free (3-3)
1 +bCFree
Where CBound is the specific concentration of protein bound to the carriers in the

membrane, which is related to each specific concentration of free protein (Cfree) in the

equilibration solution.

The other of the three sorption parameters contained in the dual-mode model is the

partition coefficient. Through the analysis of the boundary conditions of the Langmuir-mode

model, an optimum parameter value for the partition coefficient could be estimated.









DNA was obtained from Alpha-DNA. All other chemicals were of reagent grade and used as

received. Purified water, obtained by passing house-distilled water through a Barnstead, E-pure

water-purification system, was used to prepare all solutions.

Preparation of the Silica Nanotube Membranes

Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-

based template-synthesis method. Briefly, a sol-gel precursor solution was prepared by mixing

absolute ethanol, TEOS and water (40:5:10 by volume), and this solution was allowed to

hydrolyze for 1 hour. The alumina membranes were then immersed into this solution for two

minutes under sonication, and they were kept in this solution for ten additional minutes. The

membranes were then removed and the surfaces swabbed with ethanol to remove precursor from

the alumina surface. The membranes were then air dry for ten minutes at room temperature and

cured in an oven for 12 hours at 150 oC.

Preparation of PNA-Modified Nanotube Membranes

The biotinylated-PNA probe was immobilized on nanotube membrane surfaces using the

streptavidin-biotin affinity system.297 324 Streptavidin is a 60-kDa protein purified from the

bacterium Streptomyces avidini. The interest in streptavidin is the remarkable and unique

interaction with its low-molecular weight ligand biotin (Kd_10-15 M). Because of its extremely

strong affinity, the streptavidin-biotin affinity system has been developed as a general tool to use

in various experimental systems.325

The immobilization process used in this approach consisted of three basic steps. The first

and second steps were respectively, a silanization which introduced amino groups to the surface,

and conjugation between the amino groups and the NHS-LC-Biotin. The latter acylation reaction

of an N-Hydroxysuccinimidyl (NHS) ester of biotin to the primary amino groups of the silane-

modified surface yields to a stable amide bond. The next step was the coupling of streptavidin to










proteins are the same, which in turn means that the proteins are identical in size, as would be

expected for two IgG molecules. Therefore, the transport properties across these control

membranes are the same for these two antibodies. These points are further reinforced by the

concentration dependence of the flux. The concentration used for the lower two curves was

eight-times lower than the concentration used for the upper two curves. Correspondingly, the

flux at the high concentration (1.45 pmoles h-1 cm-2) is eight times higher than the flux at the low

concentration (0.18 pmoles h-1 cm-2). After the permeation experiments, the membranes were

observed under the fluorescence microscope, but no fluorescence was detected. This indicates

PEG-Si-modification suppresses non-specific protein adsorption to the silica nanotubes.

10
9 -
8d


7
0 6 }
5 -
0
i8
4 4
0 4
o 3


A0 A

l;lelrlsin--is
0 200 400 600 800 1000 1200 1400
Time (min)

Figure 2-5. Flux plots for anti-M-IgG (A) and anti-H-IgG (m) through PEG-Si-modified control
membranes. The feed concentration was 2.65 nM (lower curves) and 21.2 nM (upper
curves).

Competitive Binding Studies

As noted above, our experimental design assumes that the permeating protein anti-M-IgG

binds strongly to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to

this carrier. Likewise we assume that anti-H-IgG binds strongly to the carrier H-IgG and that










obtained by the proteins that strongly bind to the carrier. The binding constant errors found in

Figure 2-16 for the proteins that weakly bind the carriers may result not only from the possible

lack of equilibrium due to the weak affinity with the carrier, but also from the random orientation

effect of the immobilized antibody. It was also demonstrated that by controlling the orientation

of the immobilized antibody using protein G the potency affinity of the antibodies is effectively

preserved (bars in the right side of Figure 2-16).

90 1600
M K = Dissociation Cta.
80 C = Binding Capacity 1400

70 anti-M-IgGi antiIgG 1200

60 T I-ntiH-lgG N
1000
50 -
anti-H-gG 800
40 -antiM
30 600 ,
30 U
400
20 -

10 200


H-IgG H-IgG M-IgG M-IgG Protein G
Membrane Modification M-IgG
Figure 2-16.Dissociation constants and binding capacities from adsorption isotherms of anti-M-
IgG and anti-H-IgG on H-IgG- and M-IgG-modified membranes, according to the
Langmuir monolayer model.

Effect of the Temperature on the Flux and Binding Constant

For these studies, nanotube membranes templated on the 100 nm pore size commercially

alumina were modified with M-IgG well-oriented immobilized via the protein G immobilization

technique. Transport experiments (Figure 2-17), and adsorption isotherms (Figure 2-19) were

carried out at three different temperatures: 23, 32, and 42 C.

Figure 2-17 shows the flux plots for the transport experiments carried out at the three

studied temperatures. It was observed that as the temperature increases the flux for anti-M-IgG









The results revealed that the different immobilization steps introduced significant changes

in the membrane resistance. The most significant change was found in the membrane resistance

associated with the hybridization process of DNA with the PNA-modified nanotube membrane.

This decrease in resistance was associated with the increase of negative surface charge density

due to the duplex formation PNA-DNA on the surface.

-125000






-75000



-- (a)
---- (b)
/ -- (C)
-25000 / (d)
--o- (e)

-- (g)

0 25000 50000 75000
Z'
Figure 6-8. Nyquist plots (Zim vs. Zre) of impedance measurement in highly purified water: (a)
alumina, (b) after synthesis of silica-nanotube, (c) after silanization, (d) after
biotinylation, (e) after streptavidin immobilization, (f) after PNA immobilization, and
(g) after DNA hybridization. The plots were recorded applying 5 mV alternative
voltage in the frequency range from 1 MHz to 0.05 Hz.

The increase of Rm after the deposition of silica nanotubes was consistent with the fact that

after deposition smaller effective pore size in the membrane might be reached. Conversely, the

decrease in membrane resistance after silanization was explained through the formation of

protonated amino groups on the surface at the pH of the experiment (highly purified water).

However, part of this amino protonated groups were neutralized by the conjugation reaction with

the biotin linker, which was confirmed for the increase in resistance after the biotinylation. The









that agreement was missing. The explanation for this behavior was that the dual-mode model

assumes a constant diffusion coefficient, which in practice does not happen. Therefore, the

model is not able to discriminate a different regimen for the concentration dependence on the

diffusion coefficient as the case at low concentration observed here. As was said above, at low

concentration the adsorption of these two proteins in M-IgG modified membranes is dominated

basically for the binding sites with stronger binding affinity (Figure 3-12). Then, these

Langmuir-mode sites impart higher sorption rates for the permeating proteins, slowing down the

diffusion coefficients. Consequently this is also reflected in the permeability. Thus, at low

concentration the experimental permeabilities were lower than the calculated permeabilities. This

reduction in permeability is mostly due to a combined effect of both the lower diffusion

coefficients, and concentration-dependence of diffusion coefficients at this lower region in

concentration (Figure 3-14).

Additional evidence for this conclusion was also demonstrated by observing the

concentration dependence on the experimentally obtained diffusion coefficients in Figure 3-15.

As can be seen from Figure 3-15, lower diffusion coefficients were obtained at very low

concentration. Thus, these results demonstrate highly concentration-dependent diffusion

coefficients at low concentration as a result of the two types of binding sites present in the M-

IgG modified membranes (well-oriented and random).

Lower permeabilities were also observed for anti-H-IgG at the low concentration region

(Figure 3-14). This effect was explained taking into consideration the differences in diffusion

coefficients between anti-H-IgG and anti-M-IgG. Due to anti-H-IgG diffuses faster and thereby

reaches the binding sites inside the pores faster; the concentration of anti-H-IgG is initially

greater in the membrane than that of anti-M-IgG. Thus, anti-H-IgG initially at low concentration









test tubes could be filled with a payload and then the open end corked with a chemically labile

cap, they might function as a universal delivery vehicle. More recently, Hillebrenner et al.

demonstrated nano test tubes corked by chemical self-assembly.52' 53

Buyukserin et al. have also prepared silica nano test tubes by using nanopore photoresist

films as templates.54 They used a plasma etch method, by using a nanopore alumina film as a

mask, to etch a replica of the alumina pore structure into the surface of a photoresist polymer

film. They found that the length of the test tubes is determined by the thickness of the porous

part of the photoresist film.

It is very important to control the inside diameter of the fashioned nanotubes, when the

alumina membranes are used as template. In order to control the nanotube inside diameter a

substitute template synthetic methodology has been considered, where the deposition of the film-

forming material is based on layer-by-layer deposition. In this case, the wall thickness, and

equally the inside diameter of the nanotubes would be determined by the number of layers of the

material deposited along the pore wall. Using this approach, Ai et al. attempted to deposited

layers of polyelectrolyte in the pores of a nanopore alumina membrane.5 Kovtyukhova et al.

have developed a method based on alternate SiC14/H20 deposition/reaction cycles to make and

control the thickness of silica nanotubes.56 Hou et al. used Mallouk's alternating a,co-

diorganophosphonate Zr(IV) chemistry to deposit layered nanotubes along the pore walls of an

alumina template membrane.40-41 For these nanotubes, the layers were held together by bonds

between the phosphonate groups and the Zr(IV) ions, and the polyvalence of this interaction

provided cross-linking that imparts the structural integrity of the layered nanotubes.40 Hou et al.

also used this synthetic strategy for the synthesis of DNA nanotubes where the bulk of the tube










driving force is the electrochemical potential. In addition, the flux of uncharged molecules across

the membrane is linearly dependent on concentration.

In facilitated transport, the transported species moves also according to a

thermodynamically favored direction, but with the collaboration of a specific transport system

"the carrier". Facilitated transport systems display saturation behavior, where the flux is still

dependent on concentration, but at high concentration the dependence is no longer linear. This

behavior is due to the saturation of the binding carrier sites as the concentration of the

transported species increases. Figure 1-10 compares the dependence on concentration of the flux

for the passive and facilitated transport systems.



Facilitated
Transport


jj Passive
u- / Transport






Concentration

Figure 1-10.Flux concentration dependence for passive and facilitated transport.

The third class of transport in living membranes corresponds to the active transport

systems. These use energy in order to transport the species against their thermodynamic

potential. The most usual energy supplier is the ATP hydrolysis, but light energy and the energy

stored in ion gradients are also used.107

Thus, inspired in the strategies that biological membranes use to accomplish transport,

scientists have explored the use of analog synthetic membranes with specific functionalization to

selectively conduce transport. Likewise, these systems function on the base of facilitated










The permeation process through membranes can be divided into its transient and steady

state components.181,182 The transient component can be characterized by the time lag. This time

lag observed phenomenon correspond to the time at which the permeating specie enters the

membrane to the time that the flow rate of diffusing species into the close volume reaches a

steady state of permeation.181, 182

Figure 2-14A shows the transport of anti-M-IgG and anti-H-IgG across a PEG-Si modified

control membrane. It is clearly observed a time lag event for both of the permeating proteins.

This effect was even more noticeable for the antibody-modified membrane as can be seen in

Figure 2-14B. By just comparing the time in which anti-M-IgG reaches a steady state of

permeation across PEG-Si control and M-IgG-modified membranes, ten and twelve hour

respectively, we can see that antibody immobilization has a significant effect in the functional

behavior of the membrane.

Therefore, by taking advantage of this time lag effect, transport of antibodies with different

sizes were carried out in order to prove the concept of a simultaneous combination of size- and

affinity-based separation of proteins (Figure 2-15).

0.6

0.5 -

o 0.4- -

0.3

0 0.2 -
0.
o A A

"" 0.1 A

0.0 fLIAAAAAAAA
0 200 400 600 800 1000 1200 1400
Time (min)
Figure 2-15. Flux plots for anti-M-IgG (A) and F(ab)2-anti-H-IgG (m) in well-oriented M-IgG -
modified membrane. The feed solution concentration was 21.2 nM in both of the
permeating proteins.










arrangement of neighboring pores in hexagonal arrays by repulsive interactions between the

pores during the steady-state film formation after the pore initiation process.

Sol-Gel Chemistry at a Glance

The sol-gel processing of inorganic ceramic and glass materials has been known for a long

time but it has been commonly used since the mid-1900s, and comprehensively studied during

the past 30 years.80 81 In recent years, the sol-gel chemistry has evolved as a powerful approach

for preparing micro- and nano-structured materials.8285 Sol-gel materials display attractive

properties such as ease of preparation, low temperature synthesis, multiple structures with large

open spaces of tunable porosity, controllable hydrophobicity, capacity for chemical modification

of entrapment of various components, mechanical stability, and optical transparency.86-89

The molecular precursors for preparation of sol-gel materials are essentially alkoxysilanes.

The basic steps involved in the sol-process using tetraethoxysilane (TEOS) as a precursor

molecule are shown in the Figure 1-6. The initiation step consists in the hydrolysis of the

alkoxide precursor which leads to the formation of silanol groups (Si-OH) and ethanol. The

silanols can then undergo polymerization reactions with other silanols or with other

alkoxysilanes and this condensation reaction leads to the formation of siloxane groups (Si-O-Si)

along with water and ethanol as additional reaction products.

1 Hv.h-rolysis
S,(OEt)4+ H20+H* -~ ... r, OHI), + nEtO

2)('oindensation
~Si(()Ol:f}OH)+'.-, ,. ), -- (OH).,(1,,t ... 4./ ,OH)_ + H,
01
",,' !'- ,_ ,(H),I +'."' I"; -- '{(H)_,(O.Er),_,,Si~-SSI(OEtn,_ (G( \_, +tlWH


l--Si-O--Si- --- i--()--SiO-Si-()-Si --S--()-, + 1.1p


rrI[-Si-O-Si--St-O-Si-O-L -*Gdl

Figure 1-6.Outline of the steps involved in the sol-gel synthesis.









This chapter describes the use of PNA-modified nanotube membranes to study a label-free

detection for DNA hybridization. This was achieved by exploiting the unique structural and

hybridization features of PNA. This approach was characterized by observing the changes in the

membrane resistance using impedance spectroscopy. This method allowed carrying out an in situ

characterization of the interfacial properties of the modified nanotube membranes during the

hybridization process in the duplex formation PNA-DNA. The novel idea was based on the fact

that PNA is a neutral molecule; thereby during hybridization events achieved in just highly

purified water without any electrolyte in the solution, the only source of new charges in the

system during the hybridization process emerge from the DNA molecules. This event exhibited a

change of the membrane resistant that was detected by impedance spectroscopy which is an

effective technique to probe the interfacial properties of modified electrodes.

Experimental

Materials

Commercially available nanopore alumina membranes (60 rm thick, nominal pore

diameter 100 nm and 200 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999%

(TEOS) (Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these

membranes. A trimethoxysilyl propyl diethylenetriamine (DETA) was used to introduce amino

groups to the nanotubes was obtained from United Chemical Technologies. Sulfosuccinimidyl-

6-(biotinamido) hexanoate (NHS-LC-Biotin) obtained from Pierce was used as biotin-

streptavidin linker. A purified streptavidin was obtained from Sigma Aldrich. A biotinylated-

PNA (PNA) with twelve nucleotides bases (Biotin- GCG CTA GCC TAC) obtained from

Applied Biosystems was used as the probe to detect DNA. To investigate the hybridization

properties of the PNA-modified nanotube membranes a complementary twelve bases DNA (3'-

CGC GAT CGG ATG -5') with a MW=3607 g/moles was used. This complementary target









modified membranes. In this case the two control IgGs were the anti-M-IgG. A F(ab)2 fragment

antibody raised in goat against human IgG (F(ab)2-anti-H-IgG) and labeled with FITC

fluorophore was also used for transport experiments. This F(ab)2-anti-H-IgG antibody was

obtained from Molecular Probes. SuperBlock blocking buffer solution in phosphate-buffered

saline was obtained from Pierce. SuperBlock solution contains a proprietary protein for

blocking binding sites for non-specific protein adsorption. Tween 20 was obtained from Sigma

Aldrich. All other chemicals were of reagent grade and used as received. Purified water,

obtained by passing house-distilled water through a Barnstead, E-pure water-purification system,

was used to prepare all solutions.

Preparation of the Silica Nanotube Membranes

Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-

based template-synthesis method.9'25, 85 Briefly, a sol-gel precursor solution was prepared by

mixing absolute ethanol, TEOS and 1 M HC1 (50:5:1 by volume), and this solution was allowed

to hydrolyze for 30 minutes. The alumina membrane was then immersed into this solution for

two minutes under sonication. The membrane was then removed and the surfaces swabbed with

ethanol to remove precursor from the alumina surface. The membrane was then air dry for ten

minutes at room temperature and cured in an oven for 12 hours at 150 oC. This yields silica

nanotubes, with wall thickness of 3 nm, embedded within the pores of the alumina membrane.9

Antibody Immobilization

The trimethoxy aldehyde silane was first attached to the inside walls of the silica

nanotubes to introduce aldehyde terminal groups onto the pores of the membrane.9 The

membrane was then immersed into a solution that was 1.0 mg per mL in either M-IgG or H-IgG

dissolved in 10 mM phosphate-buffered saline with pH = 7.4 that also contained 150 mM

sodium chloride and 5 mM sodium azide. The membrane was incubated overnight at 4 oC,









Determination of Static Protein Binding Capacity and Binding Constants through
A dsorption Isotherm s ................... ... .......... .... ... .. .... ............ ................... 6 1
Effect of Equilibration Time on Flux and Selectivity via Dynamic Binding
C a p a c ity ..............................................................................6 2
R results and D discussion ......................... ..................................... ............. ..............62
Electron M icroscopy and Calibration Data................................................................ 63
Transport Properties of the anti-IgGs through PEG-Si-modified Control
M e m b ra n e s ............................................................................................................. 6 4
C om petitive B finding Studies........................... .................................................... 65
Functional Behavior of the M-IgG- and H-IgG-modified Membranes...........................67
Observed Functional Behavior is not a Transient Phenomenon ...................................68
Flux vs. Feed C concentration ........................ ........... ......................................... 70
Effect of the Pore Size on the Flux and Selectivity Coefficient.............................. 71
Effect of the Immobilized-Antibody Orientation on the Flux .......................................73
Separation Based on a Simultaneous Combination of Size and Affinity Selectivity......74
Binding Capacities and Binding Constants.................................................................. 77
Effect of the Temperature on the Flux and Binding Constant............... ... ...............78
Effect of the Low Ionic Strength on the Selectivity ..................................................83
C onclu sions.......... ............................... ................................................85

3 MODELING AND CHARACTERIZATION OF THE TRANSPORT PROPERTIES
OF PROTEINS ACROSS ANTIBODY-FUNCTIONALIZED NANOTUBE
MEMBRANES BY USING THE DUAL-MODE MODEL ...............................................87

Introduction ............................... ...... ... ......................................87
D definition of D ual-M ode M odel...................... ...... ...................................... ............... 89
R results and D iscu ssion ................ .. .. .. .... .... .... ... ......... ................................95
Simulations of Permeability vs. Feed Concentration Based on the Dual-Mode
M odel .............................. ............. .................... ............ 95
Experimental Description of the Dual-M ode M odel.................................................... 100
Effect of feed concentration on the flux........................................... ...........100
Experimental permeabilities as function of the feed concentration .......................101
Experimental evaluation of the dual-mode model parameters............................. 102
Experimental evaluation of Csites and b ...................................... ...............103
Experimental evaluation of the partition coefficients k, ..................................107
Experimental evaluation of the diffusion coefficients and solubility ....................110
Fitting Experimental Data with the Theoretical Dual-Mode Model ...........................114
Validation of the Dual-M ode M odel ....................... ........................... .................... 119
C onclu sions.......... .............................. ...............................................12 1

4 METAL ION AFFINITY NANOTUBE MEMBRANES: PREPARATION,
CHARACTERIZATION, AND CONTROLLED TRANSPORT OF SIX-HISTIDINE-
TAGGED RECOM BINAN T PROTEIN ................................................... ..................... 122

Introdu action ................... ......................................................................... 122
E x p erim mental ............................................................................... 124
M a te ria ls .........................................................................................................1 2 4









[256]Hochuli, E.; Doebeli, H.; Schacher, A. Journal of Chromatography 1987, 411, 177-184.

[257]Gaberc-Porekar, V.; Menart, V. ChemicalEngineering & Technology 2005, 28, 1306-
1314.

[258] Sundberg, R. J.; Martin, R. B. Chemical Reviews 1974, 74, 471-517.

[259] Gasparac, R.; Martin, C. R.; Stupnisek-Lisac, E.; Mandic, Z. Journal of the
Electrochemical Society 2000, 147, 991-998.

[260]Meier, M. A. R.; Schubert, U. S. Journal of Polymer Science, Part A: Polymer Chemistry
2003, 41, 2964-2973.

[261]Herrero, M.; Tiemblo, P.; Reyes-Labarta, J.; Mijangos, C.; Reinecke, H. Polymer 2002,
43, 2631-2636.

[262] Tiemblo, P.; Guzman, J.; Riande, E.; Mijangos, C.; Reinecke, H. Macromolecules 2002,
35, 420-424.

[263]Dias Filho, N. L. Mikrochimica Acta 1999, 130, 233-240.

[264]Kemlo, J. A.; Shepherd, T. M. ChemicalPhysics Letters 1977, 47, 158-162.

[265] Varnes, A. W.; Dodson, R. B.; Wehry, E. L. Journal of the American Chemical Society
1972, 94, 946-950.

[266]de Silva, A. P.; Gunaratne, H. Q. N.; Gunnlaugsson, T.; Huxley, A. J. M.; McCoy, C. P.;
Rademacher, J. T.; Rice, T. E. ChemicalReviews 1997, 97, 1515-1566.

[267]Fabbrizzi, L.; Licchelli, M.; Pallavicini, P.; Sacchi, D.; Taglietti, A. Analyst 1996, 121,
1763-1768.

[268]Huang, F.; Tian, Y.; Chen, C.-Y.; Cheng, Y.-J.; Young, A. C.; Jen, A. K. Y. Journal of
Physical Chemistry C 2007, 111, 10673-10681.

[269]McFarland, S. A.; Finney, N. S. Journal of the American Chemical Society 2001, 123,
1260-1261.

[270] You, C.-C.; Hippius, C.; Gruene, M.; Wuerthner, F. Chemistry--A European Journal
2006, 12, 7510-7519.

[271] Shen, Y.; Mackey, G.; Rupcich, N.; Gloster, D.; Chiuman, W.; Li, Y.; Brennan, J. D.
Analytical Chemistry 2007, 79, 3494-3503.

[272]Hartwig, J. F.; Pil, P. M.; Lippard, S. J. Journal of the American Chemical Society 1992,
114, 8292-8293.










of hydrogen bonds results in hydrogen bridges between appropriated atoms. In the same way, the

attractions between the opposite charged groups in both sides of the proteins contribute with the

electrostatic forces, while the Van der Waals forces are created by the interaction between

electrons clouds. The hydrophobic forces on the other hand, which may contribute up to half of

the total strength of the antigen-antibody bond, rely in the association of non-polar hydrophobic

groups, minimizing the contact with water molecules. All these attractive forces are weak if they

are considered individually and compared with covalent bonds. However, if they are put together

they become a large component of the total binding energy.

antibody antigen
hydrogen H-
oondnbg from www.fles.
N -- H 1 JT_
electrostaticcmbint
i -^fl/ -+ .
Van der Waals
I ____'_ + "- l-. \ |I +."t- *
hydrophobic
watere r"
excluded


Figure t-13.Intermolecular attractive forces involves in the antibody-antigen binding. Scheme
obtained from www.fleshandbones.com/readingroom/pdf/291.pdf

Recombinant proteins

Proteins produced by in vitro transcription and translation can be expressed in organisms

such as Escherichia coli (E. coli). Since, the genome and structure of the E. coli have been long

studied and well understood by biologists and geneticists. One of the main characteristics that

make the E. coli as a very powerful and versatile system for genetic engineering and protein

expression is its ability to rapidly proliferate; E. coli cells can be doubled in number under

optimal conditions every five minutes.122,123









a fluorophore suppressing some of the fluorescence deactivation pathways, producing

fluorescence enhancement.

Thus, those changes in fluorescence intensity showed how the optical properties of the

functional imidazole membranes changes depending on the transition metal ion immobilized.

Besides, these results confirm the accomplishment in the preparation of imidazole functionalized

membranes.

Characterization by FTIR-ATR spectroscopy

The successful modification of the silica nanotube membrane with the imidazole moiety

was also demonstrated by FTIR-ATR spectroscopy. Figure 4-8 compares the FTIR-ATR spectra

for bare alumina and imidazole functionalized membranes. It should be pointed out that all

spectra were corrected for spectral distortion using the OMNIC ATR software. However, by

using this ATR correction it is possible to lose spectral information of the fundamental

vibrational modes of a molecule, overall when the entire spectral region is examined at once. To

avoid this, it is a common method to analyze the data in limited spectral regions in order to give

a more detailed assignment of the characteristic bands. The Figure 4-8B illustrates this issue, and

a narrow spectral region, from 2000 to 600 cm-1, for the imidazole functionalized membrane is

used.

The differences observed in the spectra of bare alumina and imidazole-modified

membranes clearly demonstrated the modification of the nanotube membranes by using 2-

mercaptoimidazole. For the FTIR-ATR spectrum of bare alumina two broad bands appear in the

3900-3200 cm-1, and 3000-1600 cm-1 regions, which are attributed mainly to the O-H stretching

modes and O-H deformation modes, respectively.278'279 Although, for these two broad bands the

contribution of water (at about 3600 cm-1 and about 1600 cm-1) also has to be considered, since it

is known that hydrogen bonds with water molecules contribute to broaden peak areas.280'281









By changing the anodization conditions such as electrolyte used, its concentration, the

applied voltage and the period of anodization, the pores length, radius and the cell population

density can be easily varied. The pore diameter depends on the applied voltage used for the

anodization, and the thickness is a function of the anodization time. Pore diameters can be varied

between 20 and 200 nm, with lengths between few tens of nanometers up to hundreds of

microns. The pore size can be also controlled by the pore-widening treatment of dipping the

anodized porous alumina in an appropriate acid solution after the anodization. This acidic

solution is generally used to remove the barrier layer, which is a dense insulated oxide layer that

separates the underlying aluminum surface and the porous oxide film (Figure 1-2). 58 71 However

before the barrier layer is eliminated, the porous alumina film has to be removed from the

underlying aluminum metal in order to collect a free-standing membrane. There are two methods

of removing the remaining aluminum substrate in order to obtain the free-standing membrane.

The first one is by dissolution with a saturate HgC12 solution, and the second is a voltage-

reduction method. The latter entails stepwise reduction of the potential to obtain an interfacial

oxide layer that has a network of much smaller pores.44 After formation of this highly porous

layer, the alumina film is immersed into an acidic detachment solution which results in rapid

dissolution of the interfacial oxide.44











Figure 1-2.Structure of nanoporous alumina.


Figure 1-2.Structure of nanoporous alumina.










between carrier and permeating antibodies at the very low feed concentration used during these

studies.

A key way to prove that transport across a membrane is facilitated is to plot the flux vs. the

feed concentration.9' 10' 109 For facilitated-transport membranes, such plots are linear at low feed

concentration but flatten (Langmuirian shape) 9, 10, 109 at high concentrations. Data for both of the

permeating proteins show the characteristic Langmuirian shape of facilitated transport.

Membranes that contained attached PEG-Si, instead of attached M-IgG showed linear plots as is

shown in Figure 2-10. Therefore, the antibody-functionalized membranes showed here exhibit a

type of behavior for facilitated-transport membranes during protein separations.


8 -

E
Q 6-

4-
04
E
0

0
L2



0 20 40 60 80 100
Feed Concentration ( nM )
Figure 2-10.Typical plot of flux versus feed concentration for anti-M-IgG or anti-H-IgG across
PEG-Si modified membrane. The data plotted is an average of the fluxes for anti-M-
IgG and anti-H-IgG for each specific feed concentration.

Effect of the Pore Size on the Flux and Selectivity Coefficient

The influence of the pore size on the flux and selectivity coefficient was explored by using

three different sizes of commercially available (Whatman) nanopore alumina membranes having

60 rm thickness, and nominal pore diameters respectively of 200, 100, and 20 nm. To study this

effect, nanotube membranes of the specified pore size were modified with M-IgG. For the










concentration-dependence can be used to elucidate the separation properties for the solute

molecules that permeate a carrier-modified membrane. For the reason that differences in

permeability are a consequence of the sorption mechanism, the separation properties of one

solute might differ from that of another, depending on the affinity solute-carrier interaction.

a




6.0xtO -
1 4.0x10' b
2.OxlOX


0.0
0.0 1.0X10'" 2.0X10 10X11" 4.0X0lIO' 5.OX0
Feed Concentration (moles/cm')
Figure 3-4.Permeability plot simulations for the transport of BM across carrier-modified
membranes showing the effect of the binding affinity on the permeability. (a) b = 1 x
1011 cm3/moles, and (b) b = 3 x 1010 cm3/moles. The other parameters are summarized
in Table 3-1.

Experimental Description of the Dual-Mode Model

Effect of feed concentration on the flux

A number of competitive permeation experiments were carried out at different feed

concentration, ranging from 2.65 x 10-12 to 5.0 x 10-10 moles cm-3 using M-IgG modified

membranes (Figure 3-5). Throughout every experimental permeation measurements, the flux per

-22
unit area (moles cm-2 sec-1) was obtained, using 0.31 cm2 as the area of membrane exposed to the

contacting solution phases.

Figure 3-5 shows the studies of the effect of feed concentration on the flux during

competitive permeation experiments of anti-H-IgG and anti-M-IgG across M-IgG-modified

membranes. The flux for anti-H-IgG was at all times higher than the flux for anti-M-IgG. The

linear dynamic range for anti-M-IgG was 3 times lower than for anti-H-IgG, as a consequence of









modified with protein G that was previously immobilized via aldehyde silane chemistry on the

nanotube walls. The results for the transport experiments of anti-M-IgG and anti-H-IgG across

this well-oriented M-IgG modified-membrane are shown in Figure 2-12. The feed solution

concentration was 21.2 nM in both of the permeating proteins.

The fluxes of anti-H-IgG and anti-M-IgG across this well-oriented M-IgG modified

membrane were 0.64 and 0.20 pmoles h-1 cm2, respectively. As the M-IgG-modified membranes

immobilized by direct immobilization on the nanotube surface, this membranes also

preferentially transport the protein that binds weakly to the immobilized carrier. The fluxes for

anti-H-IgG and anti-M-IgG were respectively two times and almost three times lower compared

with those fluxes for membranes modified with M-IgG but where the immobilization was

directly attached to the silane aldehyde surfaces in the membrane. The reason for these lower

fluxes was attributed to the decrease in the effective pore size in the membrane with the addition

of protein-G and the possibility of having all the M-IgG molecules in a well-oriented

immobilization which also contributes to the decrease in the effective pore size. Besides, the

well-oriented immobilization contributes with a stronger binding affinity, because this helps to

preserve the binding activity of the antibody, which at the same time lowers the diffusion

coefficients of the permeating proteins.

Separation Based on a Simultaneous Combination of Size and Affinity Selectivity

In order to prove this concept of a simultaneous combination of size- and affinity-based

separation of proteins, a nanotube membranes templated in a monodisperse pore size alumina of

-40 nm crosswise the whole 60 |tm thickness was used. Scanning electron micrographs of the

surface and cross section of this membrane are shown in Figure 2-13.









Impedance Characterization of the Unmodified Membranes ............. .... ...............179
Functionalization of the Silica nanotube Membranes ........... ......... ............... 180
Impedance Sensing of DNA-PNA Hybridization .............. ...................182
Effect of the Pore Size on R ................................................................................... 183
Stability and Reproducibility of the PNA-modified Membranes .............................. 185
C onclu sions.......... .............................. ...............................................187

7 C O N C L U SIO N S ................. ......................................... .......... ........ .. ............... .. 189

L IST O F R E F E R E N C E S ..................................................................................... ..................194

B IO G R A PH IC A L SK E T C H ............................................................................. ....................212









[309]Maruyama, K.; Mishima, Y.; Minagawa, K.; Motonaka, J. Analytical Chemistry 2002,
74, 3698-3703.

[310]Zhao, X.; Tapec-Dytioco, R.; Tan, W. Journal of the American Chemical Society 2003,
125, 11474-11475.

[311]Landers, J. P. Analytical Chemistry 2003, 75, 2919-2927.

[312]Miao, W.; Bard, A. J. Analytical Chemistry 2004, 76, 5379-5386.

[313]Castro, A.; Dalvit, D. A. R.; Paz-Matos, L. Analytical Chemistry 2004, 76, 4169-4174.

[314]Wang, R.; Minunni, M.; Tombelli, S.; Mascini, M. Biosensors & Bioelectronics 2004, 20,
598-605.

[315]Bui, M.-P. N.; Baek, T. J.; Seong, G. H. Analytical and Bioanalytical Chemistry 2007,
388, 1185-1190.

[316]Watson, J. D.; Crick, F. H. C. Nature 1953, 171, 737-738.

[317]Marmur, J.; Doty, P. Journal of molecular biology 1961, 3, 585-594.

[318]Nielsen, P. E.; Egholm, M.; Berg, R. H.; Buchardt, O. Science 1991, 254, 1497-1500.

[319]Nielsen, P. E. Current opinion in biotechnology 1999, 10, 71-75.

[320]Wang, J. Biosensors & Bioelectronics 1998, 13, 757-762.

[321]Arlinghaus, H. F.; Kwoka, M. N.; Jacobson, K. B. Analytical Chemistry 1997, 69, 3747-
3753.

[322]Wang, J.; Palecek, E.; Nielsen, P. E.; Rivas, G.; Cai, X.; Shiraishi, H.; Dontha, N.; Luo,
D.; Farias, P. A. M. Journal of the American Chemical Society 1996, 118, 7667-7670.

[323]Aoki, H.; Umezawa, Y. Analyst 2003, 128, 681-685.

[324] Abel, A. P.; Weller, M. G.; Duveneck, G. L.; Ehrat, M.; Widmer, H. M. Analytical
Chemistry 1996, 68, 2905-2912.

[325]Methods in Enzymology: Aviding-Biotin Technology, 1990, 184, 1-617.

[326] Macdonald, J. R. Impedance Spectroscopy; Emphasizing Solid Materials and Systems;
Wiley Interscience publications: NY, 1987.

[327]Winkler, B. H.; Baltus, R. E. Journal of Membrane Science 2003, 226, 75-84.









fluorescence signal were less evident. This behavior can be rationalized in that only a small

fraction of the labeled proteins fulfilled an appropriate orientation of the dipole-dipole interaction

between the donor-acceptor molecules for FRET. However, the possibility of contact quenching

should be considered as well.

Data Analysis Based on the Stern-Volmer Plots

The role of fluorescence quenching can be experimentally studied by determining the

quenching rate parameters using the Stem-Volmer plots that are drawn in accordance with the

Stern-Volmer equation (Equation 5-5)131


S=I+Ks[Q] (5-5)
I
From the experimental data of this investigation, the terms Io and I in Equation 5-5 are

known, and they are respectively the fluorescence intensity in the absence and presence of the

quencher-conjugated protein. However, the concentration of the quencher-conjugated protein

(anti-H-IgG-QSY) is not absolutely defined. Therefore, some assumptions were made in order to

quantitatively describe the fluorescence quenching process in terms of the Stern-Volmer

equation. By assuming that during the time of the transport experiment the degree of labeling

(DL) remains constant for the fluorescently protein, the fluorescence intensities were converted

to molar concentration of unquenched protein via calibration curve. It was also assumed that the

reaction between anti-H-IgG-QSY and H-IgG-FITC during the quenching phenomenon was 1:1,

since the molar concentration of unquenched protein was already established. Besides, it was

assumed that the differences in molar concentration of unquenched protein with time represent

the concentration of the quencher-conjugated protein (anti-H-IgG-QSY) that was able to have the

right conditions to interact with the donor-labeled protein (H-IgG-FITC) to yield the quenching










in the membrane. Therefore, at low concentration the solubility of the proteins was highly

increased by the presence of the M-IgG carriers, so those can be used as facilitated partition

domains of the transport. In addition, the data revealed that as the concentration increases, the

solubility decreases and reaches a plateau where the solubility in the membrane is only

controlled by the solubility in the Henry's mode.

100

80 -

60 -

Z 40
.2
20


0II
,. *.----. ------- -

0.0 1.0x101~ 2.0x10" 3.0x10 4.0x10 5.0x1010
Feed Concentration (moles/cm3)
Figure 3-12. Concentration-dependence of the solubility in term of the dual-mode model for anti-
M-IgG (A), and anti-H-IgG (m) across M-IgG modified membranes.

As was described before in Equation 3-19, the permeability is the product between

diffusion and solubility. Since both the concentration dependence of the permeability and

solubility were experimentally determined, they were used to estimate the effective values for the

diffusion coefficients of anti-M-IgG and anti-H-IgG across M-IgG modified membranes. By

using this correlation method a plot of permeability against solubility is expected to be linear,

with a slope equals to the diffusion coefficient (Equation 3-19).

Figure 3-13 shows the correlation plots of permeability and solubility based on the

concentration dependence for the anti-H-IgG (Figure 3-13A) and the anti-M-IgG (Figure 3-13B)

across M-IgG modified membranes. These plots in Figure 3-13 were carried out by using the

experimental concentration-dependence data from Figure 3-6 and Figure 3-12 both for the









transport properties of the permeant proteins across them. Additional characterization of these

antibody-based membranes showed a Langmuirian shape plot characteristic of a typical

facilitated transport membrane when the effect of feed solution concentration on the fluxes of

these fluorescent-labeled antibodies was investigated.

Experimental

Materials

Commercially available nanopore alumina membranes (60 [m thick, nominal pore

diameter 100 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999% (TEOS)

(Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes.

The purified mouse immunoglobulin (M-IgG) immobilized within the silica nanotube

membranes was obtained from Sigma Aldrich. Membranes modified with purified human

immunoglobulin (H-IgG) obtained from Sigma Aldrich were also prepared. A recombinant

protein G obtained from Sigma Aldrich was used for well oriented antibody-immobilization

membranes. A trimethoxysilyl poly(ethylene glycol) (PEG-Si, MW 460-590 Da) obtained from

Gelest (Morrisville, PA) was used to modify the silica nanotubes control membranes. A

trimethoxysilyl propyl aldehyde (United Chemical Technologies) was used to attach the M-IgG

or H-IgG to the nanotubes. Two whole antibodies raised in goat against M-IgG were used to

investigate the transport properties of the M-IgG-modified membranes. These anti-M-IgGs were

obtained from Molecular Probes (Eugene, OR) conjugated with either the dye Alexa 488 or the

dye Alexa 594.

The transport properties of two control IgGs, raised in goat against human IgG (H-IgG),

were also investigated. These anti-H-IgGs were also obtained from Molecular Probes

conjugated with either Alexa 488 or Alexa 594. It is worthy to mention that these four

conjugated antibodies were also used to investigate the transport properties of the H-IgG










membranes will be discussed, which consists of three stages: polishing of the aluminum

substrate, the two anodization steps, and the detaching step in order to have a free-standing

membrane.

Polishing of aluminum foils: In order to obtain a high quality alumina film, the starting

aluminum material, must be very smooth. Initially, the high purity aluminum foils (99.999%) are

mechanically polished with slurry of aluminum particles. If, to the naked eye, the aluminum foil

shows severe scratches a mechanical polishing with fine sand paper is applied until the scratches

disappear. The mechanical polishing is then followed by an electrochemical polishing. In this

process, a potential difference of 15 V is applied between the anode aluminum foil and a lead

plate which serve as cathode, by using a polishing electrolyte solution (95% concentrated

phosphoric acid, 5% concentrated sulfuric acid and 20 g/L chromic oxide) heated to 70 OC.

Electropolishing steps of 5 minutes are successively repeated as many times as it is necessary,

until the aluminum foil has a mirror-like surface.

Two anodization steps: Electropolished aluminum is oxidized by using an electrochemical

cell setup like the one presented in Figure 1-3.


Anode Cathode
2AI(s)+ 3H20() A203()+ 6Hq)+ 6e 6H' iq) + 6e" 3H2)

Supply
Aluminum

si S t a in le s s
Porous Steel
Alumina

A Electrolyte




Stirring
Bar _


Figure 1-3.Electrochemical cell setup for nanoporous alumina growth.











situation the reduction was of 53% (Figure 6-10B). These differences in the resistance during the

hybridization process might be attributed to the fact that membranes with higher resistance

present longer dynamic range. However, the possibility of a higher immobilization of PNA

might be also considered.

Stability and Reproducibility of the PNA-modified Membranes

The stability of the PNA-modified membrane was evaluated by comparison the effect of

DNA hybridization in time with an unmodified control membrane. These results are compared in

Figure 6-11.

16000 45000
A o B
A 40000 B
14000
35000
12000_ _
S0 30000
100000
3 25000
8000
E 20000
6000
6000 15000
4001500 --- -- -
4000 10000

2000 5000
0 L I F I I I I I IF 1 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Time after adding DNA (hours) Time after adding DNA (hours)
Figure 6-11. Study of the hybridization process stability with time: (A) unmodified control
alumina membrane, and (B) PNA-modified membrane. The resistance values were
obtained from the impedance data recorded by applying 5 mV alternative voltage in
the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of
complementary DNA.

Figure 6-11 shows the stability of the impedance measurement for a control (Figure 6-

11A), and PNA-modified membrane (Figure 6-11B) after adding 3 nmoles of complementary

DNA to the experimental cell. It is clearly seen that the changes in membrane resistance are

mainly due to the hybridization process (Figure 6-11B). However, some reduction on the

membrane resistance was observed for the control membrane after the addition of 3 nmoles of

DNA. This effect might be attributed to nonspecific interactions on the membrane surface. The










complementary to the DNA target with a sequence 3'- CGC GAT CGG ATG -5', was used

as receptor in the nanotube membrane sensing-platform. These studies begin, however, with the

impedance measurements for the alumina template, and silica nanotube membranes and

investigations of the equilibration times required for the impedance measurements.

Impedance Characterization of the Unmodified Membranes

The Nyquist plots for bare alumina template (200 nm nominal pore size) and silica

nanotube membranes are shown in Figure 6-5. The x-axis and y-axis in this plot corresponds to

real impedance, Zre, and imaginary impedance, Zim, respectively. Zre and Zim are mainly

originated from the resistance and capacitance of the cell system, respectively.

-40000


-30000



N -20000 b -
Sa

-10000


0
0 10000 20000 30000 40000
Z'
Figure 6-5.Nyquist plots (Zim vs. Zre) for impedance measurement of (a) alumina and (b) silica-
nanotube assembled membranes in highly purified water.

As shown in Figure 6-5, changes after the deposition of the silica nanotubes in the alumina

template were observed in semicircle part at high frequency region (usually associated with

kinetic controlled process), whereas the linear tail at lower frequencies (usually associated with

mass transfer process) maintained almost the same slope. The variations in the high frequency

region were probably due to a change in the isoelectric point (pI) of the nanotube membrane.

Since it is known that the pi of the alumina membranes range from 6.9 to 10.5, while the pi for










log plots) confirmed also that transport experiments with higher concentration of anti-H-IgG-

QSY yields to a faster rate of the fluorescence quenching. These results are summarized in Table

5-2.

Table 5-2.Relationship between slope of the log-log plots for the fluorescence signal vs. time
with the feed solution concentration of anti-H-IgG-QSY.
Feed Concentration of
danti-H-IgGC SY (nM) Slope of log-log plots Intercept of log-log plots
anti-H-IgG-QSY (nM)
0 (control) -0.1198 2.4360
16.6 -0.1374 2.3406
33.3 -0.1693 2.3488
66.6 -0.2095 2.3344


The relation between the slopes obtained from the log-log plots with the feed concentration

is shown in Figure 5-11.


-0.12


"- -0.14 -


-0.16 -

-0.18 -


m -0.20

0 10 20 30 40 50 60 70
Feed Cocentration of anti-H-IgG-QSY (nM)
Figure 5-11.Linear relationship between slopes obtained from the log-log plots with the feed
concentration of anti-H-IgG-QSY.

A linear relationship was also obtained between the slopes of the log-log plots with the

feed concentration of anti-H-IgG-QSY. It showed excellent linearity (correlation coefficient =

0.996) over the range from 0 nM to 66.6 nM. Therefore, by this simple data analysis using log-

log plots, it is also possible to determine the concentration of the analyzed protein on the feed

solution by simply knowing the slope of the log-log plot.









2-4. Typical plots of (A) calibration data, and (B) calibration curve for Alexa-488 labeled
anti-H -IgG ...................... ..................................... .. .................. ........ 64

2-5. Flux plots for anti-M-IgG (A) and anti-H-IgG (m) through PEG-Si-modified control
membranes. The feed concentration was 2.65 nM (lower curves) and 21.2 nM
(upper curv es). .......................................................... ................. 65

2-6. Fluorescence spectra of (A) M-IgG- and (B) H-IgG-modified membranes after
equilibration with a solution that was 21.2 nM in both (a) anti-H-IgG, and (b) anti-
M -IgG .......................................................... .................................. 66

2-7. Flux plots for anti-M-IgG (A) and anti-H-IgG (m) in (A) M-IgG-modified
membrane, and (B) H-IgG-modified membrane. The feed solution concentration was
21.2 nM in both of the permeating proteins.............. ............................................. 68

2-8. Dynamic binding capacity study using M-IgG-modified membranes. The flux and
selectivity coefficient vs. days of equilibration are shown for the transport of anti-M-
IgG and anti-H-IgG. The feed solution concentration was 21.2 nM in both of the
perm eating proteins .................. ................ ........................... .. ....... ..... 69

2-9. Flux versus feed concentration for anti-M-IgG (A) and anti-H-IgG (m) across M-IgG
modified membrane. The feed solution concentrations were equimolar in both of the
permeating proteins............... .......... .. ....... ............. 70

2-10. Typical plot of flux versus feed concentration for anti-M-IgG or anti-H-IgG across
PEG-Si modified membrane. The data plotted is an average of the fluxes for anti-M-
IgG and anti-H-IgG for each specific feed concentration............................................... 71

2-11. Effect of the pore size on (A) the flux, and (B) selectivity coefficients for M-IgG
modified membranes. The feed solution concentrations was 2.65 nM both in anti-H-
IgG and anti-M -IgG ................... .. .................... ... .............72

2-12. Flux plots for anti-M-IgG (A) and anti-H-IgG (m) in well-oriented M-IgG-modified
membrane. The feed solution concentration was 21.2 nM in both of the permeating
p ro te in s ................... ................................ ......................... ................ 7 3

2-13. Scanning electron micrograph of (A) the surface, and (B) cross section of a highly-
ordered nanoporous alumina membrane obtained at 30 V in 5% oxalic acid. .................75

2-14. Flux plots for anti-M-IgG (A) and anti-H-IgG (m) in (A) PEG-Si-modified
membrane, and (B) well-oriented M-IgG -modified membrane. The feed solution
concentration was 21.2 nM in both of the permeating proteins.................. ...............75

2-15. Flux plots for anti-M-IgG (A) and F(ab)2-anti-H-IgG (m) in well-oriented M-IgG -
modified membrane. The feed solution concentration was 21.2 nM in both of the
perm eating proteins ................................................................. .. ....... ..... 76









which allowed free amino sites on the antibody to attach via the Schiff's base reaction to the

nanotube-immobilized aldehyde groups9. After immobilization, the membranes were rinsed with

buffer and then incubated for 5 hours in the SuperBlock solution that also contained 0.05%

Tween 20.

Transport Experiments

The M-IgG- or H-IgG-modified membranes were sandwiched between two pieces of

Scotch tape that had 0.31 cm2 area holes punched through them. These holes defined the area of

membrane exposed to the contacting solution phases. This membrane assembly was mounted

between the two halves of a U-tube permeation cell, with half-cell volumes of -10 mL.4 6, 10,22,23






i'e
I -






Figure 2-2.Flow-through fluorescence system used to measure the flux of the permeating
proteins.

The transport properties of three types of membranes were investigated membranes

containing immobilized M-IgG, membranes containing immobilized H-IgG, and control

membranes with attached PEG-Si. The permeating proteins, whose transport properties were

investigated, were the fluorescently-labeled anti-M-IgG and anti-H-IgG. The feed solution for

all experiments was equi-molar in these two proteins, dissolved in 10 mM phosphate-buffered

saline pH 7.4 that also contained 150 mM sodium chloride and 5 mM sodium azide (buffer used

for all studies). The receiver solution on the other side of the membrane was identical except it

initially contained no protein.









3-9. Adsorption isotherm for anti-M-IgG (A) binding on M-IgG-PG modified
m em branes. ...............................................................................106

3-10. Extrapolations of the adsorption isotherms at infinite dilute concentration for anti-M-
IgG (A), and anti-H-IgG (m) in M-IgG modified membranes............................ 108

3-11. Permeability plot simulations illustrating the effect of the partition coefficient on the
dual-mode model. Simulation parameters: D = 5 x 10-8 cm2/sec, b = 9 x 1010
cm3/moles, Cstes = 3.5 x 10-8 moles/cm3, and kp were respectively, 1 (V), 0.1 (A),
0 .0 1 (e ), an d 0 .0 0 1(* ).................................................................... ..... ..........10 9

3-12. Concentration-dependence of the solubility in term of the dual-mode model for anti-
M-IgG (A), and anti-H-IgG (m) across M-IgG modified membranes...........................111

3-13. Correlation plots between permeability and solubility based on the concentration
dependence for (A) anti-H-IgG and (B) anti-M-IgG across M-IgG modified
m em b ran e s ............. .................................................................... 1 12

3-14. Experimental and calculated obtained permeability versus feed concentration for
anti-M-IgG (A) and anti-H-IgG (m) across M-IgG modified membrane. ....... .......... 114

3-15. Experimental obtained diffusion coefficients versus feed concentration for (A) anti-
H-IgG and (B) anti-M-IgG across M-IgG modified membrane................................ 117

3-16. Plots for experimental and calculated obtained permeability against (1 + bCfeed)-1 for
(A) anti-H-IgG, and (B) anti-M-IgG across M-IgG modified membrane .....................120

4-1. Reaction mechanism of imidazole functionalized nanotube membranes......................126

4-2. Flow-through UV-Vis detection system used to measure the flux of six-histidine-
tagged ubiquitin human recombinant protein. ...................................... ............... 129

4-3. Fluorescence spectra of (a) silica nanotube and (b) imidazole functionalized
m em branes. ...............................................................................130

4-4. Fluorescence spectra of imidazole functionalized membranes before (upper) and
after (lower) metal ion immobilization: (A) Co+2, (B) Cu+2, (C) Ni+2, and (D) Zn+2. ....131

4-5. Fluorescence quenching (F/Fo) for imidazole functionalized membranes in the
presence of nitrate solutions prepared in ethanol that were 1 mM of the respective
divalent m etal ion.............................................................................................. 132

4-6. Fluorescence spectra of imidazole functionalized membranes (a) before and (b) after
metal ion immobilization using an ethanol solution that was 1 mM in CuC12 ...............133

4-7. Copper concentration dependence on fluorescence enhancement for imidazole
functionalized m em branes. ..................................................................... ...................134









way to tune the affinity of a binding protein is to perform rational mutations at the binding site238

240. Thus, depending on the mutations performed in the binding pocket of binding proteins,

mutants with different binding affinities can be obtained;240 this approach would allow the

development of membrane-based separation systems with diverse dynamic ranges.

Moreover, if this tuning affinity can be integrated with a system of membranes where the

number of carriers can be augmented in a way that the permeability of the desired permeating

protein can be selectively facilitated with respect to other present proteins in the separation

sample, a selective and amplified separation system of membranes can be developed. This can be

accomplished by taking advantage of the enhancement in permeability at low concentration

caused by a contribution of the diffusion due to the Langmuir mode species as was explained

above.204,211 Thus, this system would permit to the large number of accessible binding sites to

promote higher concentration gradients inside the membrane that would enhance the flux of the

selected target proteins, which imply a notable key aspect in designing more effective

membranes for membrane-based separation.

Validation of the Dual-Mode Model

The validity of the above results can be assessed by the plot of experimental obtained

permeabilities against (1 + bCfeed)-1. From the slope of the straight line in this plot, the Langmuir-

mode diffusion coefficient (DH) can be evaluated and consequently the mobile fraction of the

Langmuir species F.211 The results for these data are shown and summarized in Figure 3-16 and

Table 3-4, respectively.

Figure 3-16 shows and compares the experimentally and calculated obtained plots of anti-

H-IgG and anti-M-IgG permeabilities vs. (1 + bCfeed)-1. Taking into consideration the cooperation

of the two binding sites at low concentration, the examination of the linear relationship of these

plots demonstrates that the concentration dependence of the permeability for these two proteins









capacities of the antibody-modified membranes and the binding constants for the permeating

proteins to the antibody carriers confirmed that the permeating protein anti-M-IgG binds strongly

to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to this carrier,

likewise, the permeating protein anti-H-IgG binds strongly to the carrier H-IgG and that anti-M-

IgG binds weakly to this carrier. It was also demonstrated that by controlling the orientation of

the immobilized antibody using protein G the potency affinity of the antibodies is effectively

preserved. Through these studies was also demonstrated that by selecting a template membrane

with an appropriated pore size diameters these antibody-modified nanotube membranes have the

ability to be used with a combination of both size and affinity selectivity for protein separation.

Additional studies on the effect of temperature on the fluxes and the binding constant

demonstrated that the functional behavior of these antibody-modified membranes is governed for

the changes in free energy before and after interaction of the permeating proteins with the carrier

antibody, which was attributed to the changes in enthalpy and entropy of the system. Other

important finding during this investigation was that by controlling the ionic strength of the

transport solution, these antibody-modified nanotube membranes can be used for situations were

the transport of an unwanted protein needs to be shut down, in order to improve selectivity. This

latter approach might be useful in the sample complexity reduction in proteomic analysis. Since

the analysis and characterization of complex mixtures of proteins are the central aims of

proteomics. It is also noteworthy to mention the low pM detection limits achieved during these

studies. In general, the ability to alter the surface biochemistries and thus tailor the membrane

for specific analytes, promises the use of these membranes for the analysis of a wide range of

proteins.



























Figure 4-14.XPS multiplexes analysis at high resolution for the imidazole functionalized
membrane.


Figure 4-15 compares the high-resolution C s peaks for the bare alumina membrane, the


chlorosilane modified membrane and the imidazole functionalized membrane, respectively.


4000



3000



2000



1000


294 292 290 288 286 284 282
Binding Energy (eV)

Figure 4-15.High-resolution XPS data of the C s peak for (a) bare alumina template, (b)
chlorosilane modified, and (c) imidazole functionalized membranes.


It was clearly seen that the C s peak for Imidazole modified membranes consisted both of


a well defined peak at 287 eV which was ascribed to the C-C peak and a shoulder at higher


energy binding where the C-N, C-S peaks are overlapped. On the other hand, the C s peak for


chlorosilane modified membranes revealed a better indication of a second peak at 291 eV which


may be assigned to the C-C1 environment. The high-resolution scan of C s peak for the bare


-a i
-


-' ~`""
rar rrnw
~II
NI)
\II)










the biotin-modified surface. It should be pointed out that the NHS-LC-Biotin had a spacer arm of

22.4 A that can reduce steric hindrance associated with the binding many biotinylated molecules

on one streptavidin, which in turns may enhance the sensitivity of the system.


SIu C LC
S NHS LC-Biotin Streptavidin




o o/ DN
ALUSAN MEMBRANE ALM 9'"






P -


DETA: a T
Sulfosuccinimidyl-6-(biotinamido) hexanoate BRAN








membranes.
1 (v/v) solution of DETA in 1 mM acetic acid for 30 minutes at room temperature. The DETA-
OH OH H
0 LiC I




DETA: -oThe a- mio--si
Trimethoxysilylpropydiethylenetriamin e d u r non ad fd at 10 -ov o
NHS-LC-Biotin: .0 a_
Sulfosuccinimidyl-6-(biotinamido) hexanoate L. LLM iioUIE,
Figure 6-2.Immobilization of biotinylated-PNA on templated synthesized silica nanotube
membranes.

During silanization the silica nanotube membranes were immersed into freshly prepared

1% (v/v) solution of DETA in 1 mM acetic acid for 30 minutes at room temperature. The DETA-

modified membranes were rinsed with water to remove the unreacted DETA. The amino-silane

modified nanotube membranes were dried under nitrogen and fixed at 120 C in the oven for 15

minutes. For the biotinylation process the amino-silane modified membranes were exposed for 6

hours to a 0.5 mg/mL solution of NHS-LC-Biotin prepared in 0.1 M bicarbonate buffer pH 8.5.

The biotinylated nanotube membranes were then incubated overnight at 4 OC in a 1 mg/mL










Experimental evaluation of Csites and b

These Langmuir sorption parameters were determined experimentally using adsorption

isotherms. The relationship that exists between concentrations of protein in the porous membrane

isothermally in equilibrium with those in the feed solution is called adsorption isotherm. In order

to assess the equilibration time for the sorption experiments, a sorption dynamic study was

initially carried out (Figure 3-7). For this, a solution that was 21.2 nM of anti-M-IgG labeled

with Alexa 488 was presented to M-IgG modified membranes. The amount of anti-M-IgG

adsorbed was investigated observing the changes in fluorescence intensity as a function of time.


600

500

C 400

C 300
0-
200-
0
S100



0 2 4 6 8
Time (Hours)
Figure 3-7.Curve of adsorption dynamics for anti-M-IgG (A) in a M-IgG modified membrane.
Inset shows the changes of fluorescence signal with time.

As can be seen from Figure 3-7, the amount of anti-M-IgG adsorbed in the M-IgG

modified membrane increases rapidly during the first 3 hours. After that, the rate of adsorption

decreases and the system reaches a plateau of apparently equilibrium.

All the adsorption isotherms were carried out with M-IgG modified membranes at 23 C in

stirred U-tube permeation cells using the same buffer conditions used for the permeation studies,

via solutions containing single concentrations of anti-M-IgG or anti-H-IgG fluorescent labeled









[162]Nakatsuka, S.; Michaels, A. S. Journal of Membrane Science 1992, 69, 189-211.

[163] Saxena, A.; Shahi, V. K. Journal of Membrane Science 2007, 299, 211-221.

[164]Aravind, U. K.; Mathew, J.; Aravindakumar, C. T. Journal of Membrane Science 2007,
299, 146-155.

[165]Poetz, O.; Schwenk, J. M.; Kramer, S.; Stoll, D.; Templin, M. F.; Joos, T. O. Mechanisms
ofAgeing and Development 2005, 126, 161-170.

[166]Poetz, O.; Ostendorp, R.; Brocks, B.; Schwenk, J. M.; Stoll, D.; Joos, T. O.; Templin, M.
F. Proteomics 2005, 5, 2402-2411.

[167]Langmuir, I. Journal of the American Chemical Society 1918, 40, 1361-1402.

[168]Yang, L.; Chen, P. Journal of Membrane Science 2002, 205, 141-153.

[169]Weiss, M.; Welsch, W.; Schickfus, M. v.; Hunklinger, S. Analytical Chemistry 1998, 70,
2881-2887.

[170] Saha, K.; Bender, F.; Gizeli, E. Analytical Chemistry 2003, 75, 835-842.

[171]Igarashi, T.; Ishigatsubo, Y.; Ohno, S.; Shirai, A.; Aoki, I.; Okubo, T.; Klinman, D. M.
Clinical Immunology and Immunopathology 1994, 72, 307-311.

[172]Klinman, D. M.; Banks, S.; Hartman, A.; Steinberg, A. D. Journal of Clinical
Investigations 1988, 82, 652-657.

[173]Gueell, C.; Davis, R. H. Journal of Membrane Science 1996, 119, 269-284.

[174]Mueller, J.; Davis, R. H. Journal of Membrane Science 1996, 116, 47-60.

[175]Ho, C.-C.; Zydney, A. L. Journal of Membrane Science 1999, 155, 261-275.

[176]Jones, K. L.; O'Melia, C. R. Journal of Membrane Science 2000, 165, 31-46.

[177]Dechadilok, P.; Deen, W. M. Industrial & Engineering Chemistry Research 2006, 45,
6953-6959.

[178]Bhattacharjee, S.; Sharma, A.; Bhattacharya, P. K. Industrial & Engineering Chemistry
Research 1996, 35, 3108-3121.

[179]Johnson, C. P.; Jensen, I. E.; Prakasam, A.; Vijayendran, R.; Leckband, D. Bioconjugate
Chemistry 2003, 14, 974-978.









organization of the pores which is considered to be triggered by mechanical stress at the

alumina/aluminum interface.76

In a first step the barrier layer is formed jointly with numerous defects in the aluminum

surface for the pore initiation. Due to the decreasing current with the increase of the thickness of

the barrier layer the formation of pores is initiated at defect positions. Then, assisted by the

electric field, the alumina film developed at the metal-oxide interface at the pore walls are

preferentially dissolved by the acidic electrolyte. For this reason, when the electric field

increases by increasing the voltage, the dissolution process is accelerated and larger pores are

formed. Thus, pores grow perpendicular to the surface via field-assisted oxide dissolution at the

oxide/electrolyte interface in dynamic equilibrium with the oxide growth at the metal/oxide

interface.77'78 It is believed that the anodization current is mainly related to the movement of

ionic species (02-, OH-, Al3+) through the oxide layer at the bottom of the pore.79 Therefore, as a

consequence of the pore development, the electric field and the ionic current become

concentrated in the barrier layer underneath the major pores. The oxide growth at the metal/oxide

interface is due to the migration of oxygen containing ions (O-2/OH-) from the electrolyte

through the oxide layer at the pore bottom, with simultaneous ejection of Al3+ ions drifted

through the oxide layer into the solution at the oxide/electrolyte interface.79 The fact that Al3+

ions are lost into the electrolyte and that the atomic density of aluminum in alumina is by a factor

of two lower than in metallic aluminum, have suggested a possible origin of forces between

neighboring pores that contributes to the mechanical stress which is associated with the

expansion during oxide formation at the metal/oxide interface.76 Since the oxidation takes place

at the entire pore bottom simultaneously, the material can only expand in the vertical direction,

so that the existing pore walls are pushed upward. This somewhat explain the self-organized









becomes less dependent on the concentration. This region resembles to the time lag region for

gases in glassy polymers.181' 182

In general, as stated by Koros and Paul the limits of the dual-mode model are confined to

two situations.204 One is the limit where total immobilization occurs, and only the Henry's mode

species contributes to the permeability. The other limit is a situation with a nonlinear isotherm

with no distinctions for sorption mechanism or transport behavior between any of the adsorbed

solutes. However, the concentration-dependence of the permeability is a consequence of the

shape of the adsorption isotherm, and therefore a result of the binding affinity strength between

Langmuir permeating species and Langmuir binding sites.

In order to demonstrate the concentration dependence on the permeability as consequence

of the shape of the adsorption isotherm, permeability plot simulations were carried out

comparing two different sorption mechanisms via two different binding affinities (Figure 3-4).

The plot (a) in Figure 3-4 clearly indicates that a carrier-modified membrane with stronger

affinity for the solute molecules displays higher concentration-dependence as a result of the

sorption isotherm. So, the concentration dependence of permeability is determined by the

strength of the binding affinity for the solute-carrier complex in the membrane. However, the

total permeability is highly dependent on the diffusion coefficient.

It is worth mentioning that by using simulations was verified that the order of influence of

the dual-mode model parameters in the permeability was D > b > C,t,, >>> kp, respectively.

According to these results, the dual-mode model permeation analysis is a useful model to

explain transport across carrier-modified membranes. The analysis showed until here

demonstrated how the fluctuations in permeability are a result of the strength of the sorption

mechanism. Therefore, the differences in permeability and the differences in permeability










with Alexa 488. Data were analyzed according to the monolayer Langmuir model (Equation 3-

3).167

By using the experimental values of Cfee and Cbound in moles per cm3 in Equation 3-3 the

parameter Csites was calculated in moles per cm3 while b in cm3 per moles. Figure 3-8 shows and

compares adsorption isotherms for anti-M-IgG and anti-H-IgG carried out with M-IgG modified

membranes; also the adsorption isotherm for anti-M-IgG using unmodified nanotube membranes

is showed. Table 3-2 summarizes the corresponding binding constant and protein binding

capacities, as calculated by fitting the data using nonlinear regression methods.


1.2x109 Control Membrane
12x -- M-IgG Modified Membrane
o -A- M-IgG Modified Membrane
1.0x109 [--

E 8.0x10 -

Santi-M-IgG
6.0x10 10
0
S4.0x10 1 anti-H-IgG
A
2 2.0x10" anti-M-IgG
S A / /AAA
0.0 -*-*.-0 ,---- =

0.0 5.0x1012 1.0x11 1.5x101 2.0x1011 2.5x101

Free Concentration (moles I cm3)
Figure 3-8. Adsorption isotherms for anti-M-IgG (A), and anti-H-IgG (m) binding on M-IgG
modified membranes. Also the adsorption isotherm for anti-M-IgG (*) using a
control membrane is shown.

Figure 3-8 shows the adsorption isotherms obtained from the fluorescence intensity of the

wet membranes exposed to different protein concentrations. For the unmodified silica nanotube

membranes the adsorption of anti-M-IgG labeled with Alexa 488 was negligible in comparison

with the antibody-modified membranes. It was also observed, that the adsorption for the

unmodified membranes increases proportionally with the concentration of fluorescent-labeled









carrying out fluorescence measurements at diverse temperatures. Thus, the quenching

mechanism is mainly ascribed to a collisional process when at higher temperatures the system

present larger values ofKsv. Conversely, for the cases of quenching via complex formation,

higher temperatures show smaller Ksv; in this case Ksv is related to the association constant of

the complex (Ka). In situations, where the fluorescence ratio as a function of the quencher

concentration reveals a positive deviation of the linear Stern-Volmer plot, it corresponds to a

combined process of both collisional and static quenching. For such a case a modified Stern-

Volmer equation is given by Equation 1-7.

2 =(1+ Ksv[Q])(1+ Ka[Q]) (1-7)
I
Therefore, the upward curvature in the Stern-Volmer plot is due to the contribution of the

square term (second power) of [Q] in the Equation 1-7.

FTIR-ATR spectroscopy

Fourier Transform Infrared Attenuated Total Reflection spectroscopy is an appropriate

technique for the characterization of the surface of materials. This technique combines the

capabilities of chemical structure identification offered by the infrared spectroscopy mutually

with the fundamental properties of the total internal reflection techniques. The origin of the

internal reflection techniques is involved with the fact that radiation propagating in an optical

medium having higher refractive index (nj) experiences total internal reflection at the interface

of a contiguous sample medium with lower refractive index (n2). This formed propagated wave,

which penetrates a small portion of the sample medium, is called evanescent.133 This

phenomenon takes place only when the angle of incidence of the radiant wave exceeds a critical

angle, 0c, which can be determined by using Snell's law (Equation 1-8).

sin 01 n
sin 0(1-8)
sin 82 n1









CHAPTER 5
TRANSPORT ANALYSIS OF PROTEINS ACROSS SILICA NANOTUBE MEMBRANES
THROUGH A FLUORESCENCE QUENCHING-BASED SENSING APPROACH

Introduction

Molecular recognition is a very important and commonly used tool in both clinical

diagnostics and pharmacology. The practical methods based on recognition depend

fundamentally on specific biomolecular recognition reactions. In some cases, these biomolecular

recognition reactions confer such extremely high specificity that it has allowed the development

of analytical assays for the determination of many chemicals and biochemicals at very low

concentrations.290 293 This is important because in this era of proteomics in particular, the

quantitative determination and identification of proteins at low concentrations are essential

requirements. Many very sensitive methods have been developed for the detection of proteins

and other biological molecules on the basis and capacity of fluorescent compounds to transfer

energy absorbed from photon radiation to close molecules by taking advantage of the molecular
294-297
recognition.294 297

This called fluorescence resonant energy transfer (FRET) phenomenon takes place when

the energy in the excited state of a donor fluorophore is transferred to a close acceptor molecule

via an appropriate orientation of the dipole-dipole interaction between these donor-acceptor

molecules.298, 299 One of the basic conditions for the FRET process to occur is that the emission

spectrum of the donor must overlap the absorption spectrum of the acceptor. Therefore, FRET

efficiency is a distance-dependent interaction process.131' 298 This distance dependence of FRET

has been extensively used to study both structure and dynamics of proteins and nucleic acids, as

well for the detection and visualization of intermolecular associations, and for the development

of biomolecular devices based on intermolecular binding interaction.298 300'301









Experimental evaluation of the diffusion coefficients and solubility

The most direct method of experimental determination of S and D requires steady-state

permeation data, complemented with equilibrium sorption data.199 Therefore, effective values of

S, D, and hence P characterized for a membrane-penetrant system that has been restricted to

equilibrium sorption and steady-state permeation conditions represent appropriate averages of

the transport parameter of the system.224 To use the dual-mode model requires knowing D and S.

This was accomplished experimentally through an evaluation of the permeability and solubility

in terms of the dual-mode model as will be shown next.

As was state before, the solubility for a penetrant protein in the membrane can be

expressed in term of the dual-mode sorption parameters by the Equation 3-20.210 For the reason

that the three sorption parameters required for this calculation were already established, an

experimental study of the concentration dependence of the solubility was conducted (Figure 3-

12).

Figure 3-12 shows the concentration-dependence of the solubility in term of the dual-mode

model for anti-M-IgG and anti-H-IgG in M-IgG modified membranes. The concentration-

dependence of the solubility in term of the dual model is a consequence of the shape of the

adsorption isotherm. Hence, from Figure 3-12 M-IgG modified membranes possess higher

affinity for anti-M-IgG than for anti-H-IgG, as was demonstrated before (Table 3-2). These data

also indicate where a protein molecule prefers to dissolve, whether in the Langmuir binding sites

or in the membrane matrix based on Henry's law. According to Figure 3-12, anti-M-IgG

molecules would rather dissolve in the Langmuir sorption mode (upper plot), while anti-H-IgG

molecules in the Henry's mode (lower plot). It was also observed that regardless of the protein at

low concentrations the Langmuir mode contributes the most to the total solubility, which is

probably due to the strong affinity exerted for that fraction of well-oriented immobilized carriers









[199]Galiatsatou, P.; Kanellopoulos, N. K.; Petropoulos, J. H. Journal of Membrane Science
2006, 280, 634-642.

[200]Funke, H. H.; Noble, R. D.; Koval, C. A. Journal of Membrane Science 1993, 82, 229-
236.

[201] Smith, K. A.; Meldon, J. H.; Colton, C. K. AIChE Journal 1973, 19, 102-109.

[202] Teramoto, M. Industrial & Engineering Chemistry Research 1994, 33, 2161-2171.

[203]Vieth, W. R.; Sladek, K. J. Journal of Colloid Science 1965, 20, 1014-1033.

[204]Paul, D. R.; Koros, W. J. Journal of Polymer Science, Polymer Physics Edition 1976, 14,
675-685.

[205]Vieth, W. R.; Howell, J. M.; Hsieh, J. H. Journal of Membrane Science 1976, 1, 177-220.

[206] Petropoulos, J. H. Journal of Polymer Science, Polymer Physics Edition 1970, 8, 1797-
1801.

[207]Mauze, G. R.; Stern, S. A. Journal of Membrane Science 1984, 18, 99-109.

[208]Riedl, C. M.; Spencer, H. G. Journal ofAppliedPolymer Science 1990, 41, 1685-1693.

[209]Noble, R. D. Journal of Membrane Science 1990, 50, 207-214.

[210]Kanehashi, S.; Nagai, K. Journal of Membrane Science 2005, 253, 117-138.

[211] Tsujita, Y. Progress in Polymer Science 2003, 28, 1377-1401.

[212]Barrer, R. M.; Barrie, J. A.; Slater, J. Journal of Polymer Science 1958, 27, 177-197.

[213] Stern, S. A.; Zhou, S.; Araux-Lara, J. L.; Ware, B. R. Journal of Polymer Science, Part
C: Polymer Letters 1989, 27, 427-431.

[214]Koros, W. J.; Paul, D. R. Journal of Polymer Science, Polymer Physics Edition 1978, 16,
1947-1963.

[215]Huh, J.-K.; Song, D.-I.; Jeon, Y.-W. Separation Science and Technology 1999, 34, 571-
586.

[216]Huh, J.-K.; Song, D.-I.; Jeon, Y.-W. Separation Science and Technology 2000, 35, 243-
259.

[217]Wang, J.-S.; Kamiya, Y. Journal of Membrane Science 1999, 154, 25-32.









Transport Experiments

The bare silica nanotube membranes, M-IgG- or H-IgG-modified membranes were

sandwiched between two pieces of Scotch tape that had 0.31 cm2 area holes punched through

them. These holes defined the area of membrane exposed to the contacting solution phases.

This membrane assembly was mounted between the two halves of a U-tube permeation cell, with

half-cell volumes of 10 mL.

The feed solution contained the quencher conjugated antibody (anti-H-IgG-QSY)

dissolved in 10 mM phosphate-buffered saline pH 7.4 that also contained 150 mM sodium

chloride and 5 mM sodium azide (buffer used for all studies). The receiver solution on the other

side of the membrane contained the buffered solution of the fluorescence emitter labeled protein

the H-IgG-FITC. The concentration of the transporting protein in the feed solution was in all

cases higher than the emitter-donor-protein in the receiver side. Thus, the chemical potential was

preferentially preserved from the feed to receiver solutions as the actual driving force for the

flux. A protocol was established to carry out all the transport experiments.

First, the membrane mounted in the U-tube permeation cell was equilibrate for one hour in

the buffer solution, and the receiver solution were frequently measured using a flow-through

fluorescence detector until a stable base line for the fluorescence signal was reached. After that, a

known concentration of anti-H-IgG-QSY was placed in the feed solution, and it was permitted to

equilibrate for another fifteen minutes. Then, a known concentration of H-IgG-FITC was put in

the receiver solution, after five minutes when the fluorescence signal was stable the experimental

data for the transport experiment started being recorded.

The changes of fluorescence intensity in the receiver solution were periodically measured

using a flow-through fluorescence detector (Figure 5-1).









some favored orientations of the imidazole moiety in the functional membrane after

coordination.

The observed band at 1110 cm 1 was also assigned, and it is clearly illustrated in Figure 4-

8B. This band was labeled to the asymmetric C-S-C stretching vibrations.284, 288 For the correct

assignment of this band, the contribution of Si-O stretching vibration at this region was also

considered. In order to confer a proper assignment, the FTIR-ATR spectrum of imidazole

functionalized membrane was also acquired by using 3-chloropropyl silica nanotube membrane

as background for the spectrum subtraction instead of air (Figure 4-8B).

Figure 4-8B compares the FTIR-ATR spectra of an imidazole functionalized membrane,

which were collected using both air and 3-chloropropyl silica nanotube membrane for

background subtraction, respectively. It was clearly observed that when the chlorosilane-

modified membrane was used for the background subtraction, the fundamental vibrational mode

from the imidazole moiety at 1118 cm-1 was more clearly discriminated. Therefore, at 1118 cm-1

the contribution of both Si-O and asymmetric C-S-C stretching vibrations should be

considered.284, 288, 289

Table 4-1.FTIR-ATR characteristic bands of imidazole functionalized nanotube membranes

Assignment Frequency (cm-)


N-H stretching -2900

N-H bending in the plane -1560

C=C and C=N ring vibration modes -1684-1630
Thioamide bands I and II, which are respectively coupled vibrations
involving stretching motion of C-S + N-H, and bending motion C-N + N-H. -1486 and 1412

C-S-C anti-symmetric and Si-O stretching vibrations. 1118
11 lo









Finally, by setting all the assumptions and writing the concentration of CH in terms of CD,

(KCD / l+aCD), a numerical solution of the partial immobilization model yields to an expression

for the permeability coefficient given by Equation 3-17.204


P =kDD 1+ lFK (3-17)
1 1 + -bCFeed
The use of the model requires the determination of both DD and F. This can be

accomplished through the experimental evaluation of the permeability coefficients; so, optimum

parameters for the diffusion coefficients can be determined.

Throughout every experimental permeation, measurements of the flux per unit area, J,

(moles cm-2 sec-1), is obtained. Therefore, if a number of permeation experiments are carried out

at different feed concentration using membranes with homogeneous thickness, L, the steady-state

permeability coefficient for a specific feed concentration can be calculated using the following

relationship showed in Equation 3-18.218 219

JL
P =- (3-18)
CFeed
This procedure would also allow comparing the experimental permeability plots with those

calculated obtained ones.

In order to explain permeation, the dual-mode model assumes that the permeation is

governed via a solution-diffusion mechanism. If the permeation of the protein in the membrane

is controlled by a solution-diffusion mechanism, the permeability can be expressed as the

product of diffusivity, D, and solubility, S as is shown in Equation 3-19.

P = DS (3-19)
Although the solubility of a permeating protein in the membrane can be also expressed in

term of the dual-mode sorption parameters by (Equation 3-20):210, 211

C Cs b
S = SD + SH = kp + (3-20)
CFeed + bCFeed









[144]Ye, M.; Jiang, X.; Feng, S.; Tian, R.; Zou, H. TrAC, Trends in Analytical Chemistry
2007, 26, 80-84.

[145]Domon, B.; Aebersold, R. Science 2006, 312, 212-217.

[146]Domon, B.; Broder, S. Journal ofProteome Research 2004, 3, 253-260.

[147]Marengo, E.; Leardi, R.; Robotti, E.; Righetti, P. G.; Antonucci, F.; Cecconi, D. Journal
ofProteome Research 2003, 2, 351-360.

[148]Barroso, B.; Lubda, D.; Bischoff, R. Journal ofProteome Research 2003, 2, 633-642.

[149]Masselon, C.; Pasa-Tolic, L.; Tolic, N.; Anderson, G. A.; Bogdanov, B.; Vilkov, A. N.;
Shen, Y.; Zhao, R.; Qian, W.-J.; Lipton, M. S.; Camp, D. G., II; Smith, R. D. Analytical
Chemistry 2005, 77, 400-406.

[150]Nedelkov, D.; Tubbs, K. A.; Niederkofler, E. E.; Kiernan, U. A.; Nelson, R. W.
Analytical Chemistry 2004, 76, 1733-1737.

[151]Zilberstein, G.; Korol, L.; Antonioli, P.; Righetti, P. G.; Bukshpan, S. Analytical
Chemistry 2007, 79, 821-827.

[152]Iritani, E.; Mukai, Y.; Murase, T. Filtration & Separation 1997, 34, 967-973.

[153]Ghosh, R.; Cui, Z. F. Journal of Membrane Science 2000, 167, 47-53.

[154]Harkensee, D.; Kokpinar, O.; Walter, J.; Kasper, C.; Beutel, S.; Reif, O. W.; Scheper, T.;
Ulber, R. Engineering in Life Sciences 2007, 7, 388-394.

[155]Guo, W.; Ruckenstein, E. Journal of Membrane Science 2003, 211, 101-111.

[156]Ramamoorthy, M.; Raju, M. D. Industrial & Engineering Chemistry Research 2001, 40,
4815-4820.

[157] Suen, S.-Y.; Liu, Y.-C.; Chang, C.-S. Journal of Chromatography, B 2003, 797, 305-319.

[158] O'Connor, A. J.; Pratt, H. R. C.; Stevens, G. W. Chemical Engineering Science 1996, 51,
3459-3477.

[159]Ho, A. K.; Perera, J. M.; Dunstan, D. E.; Stevens, G. W.; Nystrom, M. AIChE Journal
1999, 45, 1434-1450.

[160]Li, Q. Y.; Pepper, D. S. Journal of Membrane Science 1997, 136, 181-190.

[161]Nystrom, M.; Aimar, P.; Luque, S.; Kulovaara, M.; Metsamuuronen, S. Colloids and
Surfaces, A: Physicochemical and Engineering Aspects 1998, 138, 185-205.










Survey scans were taken to determine the elemental surface composition of various

elements present in the chlorosilane and imidazole modified membranes. Figure 4-13 compares

the XPS survey scan of the chlorosilane and imidazole modified membranes.

120000

100000

80000

", 60000
c Cls

40000 CI 2p

20000
a
o b

1200 1000 800 600 400 200 0
Binding Energy (eV)
Figure 4-13.XPS survey spectra for (a) chlorosilane and (b) imidazole modified membranes.

The chlorosilane modified surface shown the presence of XPS line for chlorine C12p,

which was absent in the bare alumina, which confirmed the accomplishment of the chlorosilane

modification of the silica nanotubes (spectrum a Figure 4-13). In addition, this survey scan

shown distinct peaks for A12p as well as for C Is compared to the imidazole functionalized and

bare membranes.

It was noticed that after functionalization with 2-mercaptoimidazole, an increase in C1s

peak and a diminution in the A12p peak was observed when it is compared with the chlorosilane

membrane (spectrum b Figure 4-13). Considering the surface sensitivity of XPS, the decrease of

the A12p peak in the XPS measurements for the imidazole functionalized membranes suggested

the presence of imidazole on the nanotube membrane surface.136 Additionally, the

functionalization of the membranes with the imidazole moiety was confirmed for the presence of

Nls and Sis lines in the XPS spectra.










It was also found that after equilibration the weakly binding protein (anti-H-IgG) diffuse

more rapidly than the strongly-binding protein (anti-M-IgG). Therefore, there is a trade-off

between selectivity and flux after the membranes are equilibrated. The equilibration reduces the

flux for the strongly binding protein with respect to the weakly binding one, but leads at the same

time to an increase in selectivity. Thus, it can be concluded that the M-IgG carrier without

equilibration in some extend facilitate the transport of anti-M-IgG when is compared with

transport across membranes that have been previously equilibrated.

Flux vs. Feed Concentration

A number of competitive permeation experiments were carried out at different feed

concentration, ranging from 2.65 nM to 0.5 [M using M-IgG modified membranes (Figure 2-9).



3.0 -
3.0









S1.5 0 100 200 3 40 500
1.0
L. 0.5
0.0
S 100 400 500
Feed Concentration (nM)
Figure 2-9.Flux versus feed concentration for anti-M-IgG (A) and anti-H-IgG (m) across M-IgG
modified membrane. The feed solution concentrations were equimolar in both of the
permeating proteins.

Figure 2-9 shows flux-vs.-feed plots for transport of anti-H-IgG and anti-M-IgG across M-

IgG-modified membranes. The flux for anti-H-IgG was at all times higher than the flux for anti-

M-IgG. These plots show a characteristic "Langmuirian" shapes,9' 10, 109 demonstrating that the

M-IgG attached in the membrane indeed interacts with the penetrant-antibody, acting as a carrier

during the permeation process. A straight line would be expected, if there is not interaction










through carrier-modified membranes, which is governed by both the Henry mode and the

Langmuir mode, can be modulated depending on the affinity parameters between solute-carrier.

In order to understand how the dual-mode model works, simulations of the model were

initially carried out (Figure 3-2). The simulation analysis was based on the assumption of

competitive permeation studies for two different solute species transported across carrier-

modified nanoporous membranes. With the aim of simplifying the analysis an idealized case

was compared. Thus, one of the species "the spectator" or nonbinding molecule (NBM) is

assumed to have no interaction with the carriers in membrane and therefore permeates the

membrane by only physical diffusion. On the other hand, the second specie, the binding

molecule (BM), permeates the membrane through a dual-mode model mechanism. The different

dual-mode parameters used for the simulations are listed in Table 3-1.

1.2x106
a
1.0x10 -

0 8.0x10 -
E
6.0x107

4.0x10 -
E
2.0X107

o.o F b

0.0 1.0x10"0 2.0x1010 3.0x1010 4.0x1010 5.0x1010
Feed Concentration (moles/cm3)
Figure 3-2.Permeability plot simulations for the transport of (a) BM and (b) NBM across carrier-
modified membranes. The parameters are summarized in Table 3-1.

By simple visual analysis of the simulations showed in Figure 3-2, we can notice that for a

permeation mechanism that follows the dual-mode model, the permeability is highly dependent

on the concentration, while for a mechanism based on physical diffusion only, the permeability is

totally independent of the concentration. The permeability in terms of the dual-mode model









LIST OF REFERENCES

[1] Hagg, M.-B. Separation and Purification Methods 1998, 27, 51-168.

[2] Baker, R. W. Membrane Technology and Applications, 2nd ed.; John Wiley & Sons, Ltd.,
2004.

[3] Striemer, C. C.; Gaborski, T. R.; McGrath, J. L.; Fauchet, P. M. Nature 2007, 445, 749-
753.

[4] Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655-658.

[5] Yu, S.; Lee, S. B.; Martin, C. R. Analytical Chemistry 2003, 75, 1239-1244.

[6] Yu, S.; Lee, S. B.; Kang, M.; Martin, C. R. Nano Letters 2001, 1, 495-498.

[7] Chun, K.-Y.; Stroeve, P. Langmuir 2002, 18, 4653-4658.

[8] Van Reis, R.; Zydney, A. Current Opinion in Biotechnology 2001, 12, 208-211.

[9] Lee, S. B.; Mitchell, D. T.; Trofin, L.; Nevanen, T. K.; Soederlund, H.; Martin, C. R.
Science 2002, 296, 2198-2200.

[10] Kohli, P.; Harrell, C. C.; Cao, Z.; Gasparac, R.; Tan, W.; Martin, C. R. Science 2004,
305, 984-986.

[11] Giorno, L.; De Bartolo, L.; Drioli, E. Membrane Science and Technology Series 2003, 8,
187-217.

[12] Gavalas, V. G.; Wang, J.; Bachas, L. G. Membrane Science and Technology Series 2003,
8, 379-392.

[13] Sexton, L. T.; Home, L. P.; Sherrill, S. A.; Bishop, G. W.; Baker, L. A.; Martin, C. R.
Journal of the American Chemical Society 2007, 129, 13144-13152.

[14] Baker, L. A.; Choi, Y.; Martin, C. R. Current Nanoscience 2006, 2, 243-255.

[15] Choi, Y.; Baker, L. A.; Hillebrenner, H.; Martin, C. R. Physical Chemistry Chemical
Physics 2006, 8, 4976-4988.

[16] Siwy, Z.; Trofin, L.; Kohli, P.; Baker, L. A.; Trautmann, C.; Martin, C. R. Journal of the
American Chemical Society 2005, 127, 5000-5001.

[17] Harrell, C. C.; Choi, Y.; Home, L. P.; Baker, L. A.; Siwy, Z. S.; Martin, C. R. Langmuir
2006, 22, 10837-10843.

[18] Kohli, P.; Wirtz, M.; Martin, C. R. Electroanalysis 2004, 16, 9-18.









transport concept. This facilitated transport synthetic functionalized-membranes integrate a

mediator in the membrane "the carrier".9, 10 This carrier usually interacts chemically with one of

the components in the feed to be transported across the membrane in a facilitated mode.

Therefore, in the same way that living membranes facilitated transport, the flux for this preferred

permeant is enhanced when is compared with the passive transport situation (Figure 1-10). As in

the case of facilitated transport in living membranes, the facilitated transport in synthetic

facilitated-membrane systems also displays saturation behavior. The basic mechanism for this

enhanced transport is ascribed to the sequential binding and unbinding events of the facilitated

specie with the immobilized carrier in the membrane. The concentration gradients of facilitated

process are shown schematically in Figure 1-11, in which the reaction of the facilitated specie (F)

with the immobilized carrier (C) in the membranes is given by Equation 1-1.

F+C The equilibrium constant for this reaction is characterized by Equation 1-2.

[F*C]
K =[F (1-2)
[F] x [C]
The terms in the square brackets represents the molar concentration of the chemical species

involve in the reaction.


Feed Membrane Permeate Concentration gradient
Feed Membrane Permeate for facilitated transport

C Concentration gradient
o0 for passive transport

[ F ]Feed



[ F ]Permeate
Length -----

Figure 1-11.Concentration gradients formed across the membrane during passive and facilitated
transport.










alumina membrane consisted of only one peak the C-C peak at 287 eV, which confirmed the

presence of carbon impurities in the membrane.

The immobilization of the transition metal ions in the imidazole functionalized membranes

was also verified by XPS analysis. Figure 4-16 compares the XPS survey scan of two imidazole

functionalized membranes after immobilization of Cu and Zn, respectively. For the XPS survey

scan of Cu and Zn metal-immobilized nanotube membranes, the Mg anode and the Al anode

were used respectively to stimulate photoemission.

80000
Cu 2pl
70000 a
Cu 2p3
60000 Zn 2p1

50000 Zn 2p3

S40000 -
b
S30000 Cu Auger
Zn Auger
20000
10000

0
1200 1000 800 600 400 200 0
Binding Energy (eV)
Figure 4-16. XPS survey spectra for imidazole modified membranes after immobilization of (a)
Cu+2, and (b) Zn+2

The presence of XPS spectral lines for Zn and Cu were observed in the respective

spectrum of the imidazole functionalized membranes. The XPS survey scan for Zn shown the

presence ofZn2pl at 1050 eV and Zn2p3 at 1026.5 eV peaks. Whereas the survey for Cu shown

the presence of Cu2pl and Cu2p3 peaks assigned at 954 eV and 934 eV, respectively. It should

be mentioned that analogous experiments were also carried out for the immobilization of Ni and

Co, and Auger lines for all these four metal ions were also observed. Thus, these data confirmed

the coordination and immobilization of the transition metal ions used in these studies with the

imidazole functionalized membranes.









obtained from Fulka, were used to attach the metal via coordination with the imidazole. A six-

histidine-tagged ubiquitin human recombinant protein (6His-UHRP) obtained from Sigma

Aldrich was used to investigate the transport properties of the imidazole-modified membranes.

All other chemicals were of reagent grade and used as received. Purified water, obtained by

passing house-distilled water through a Barnstead, E-pure water-purification system, was used to

prepare all solutions.

Preparation of the Silica Nanotube Membranes

Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-

based template-synthesis method described previously. Briefly, a sol-gel precursor solution was

prepared by mixing absolute ethanol, TEOS and 1 M HC1 (50:5:1 by volume), and this solution

was allowed to hydrolyze for 30 minutes. The alumina membranes were then immersed into this

solution for two minutes under sonication. The membranes were then removed and the surfaces

swabbed with ethanol to remove precursor from the alumina surface. The membranes were then

air dry for ten minutes at room temperature and cured in an oven for 12 hours at 150 C. This

yields silica nanotubes, with wall thickness of 3 nm, embedded within the pores of the alumina

membrane.9

Preparation of Imidazole Functionalized Membranes

The 3-chloropropyltrimethoxysilane was first attached to the inside walls of the silica

nanotubes. A solution that was 2% (v/v) in 3-chloropropyltrimethoxysilane was prepared in

ethanol; to this solution was added acetate buffer (50 mM, pH=5.1) to make the solution 5%

(v/v) in this buffer. This ethanol-based solution was stirred for 20 minutes and the alumina

membranes were then immersed. The soak membranes were sonicated for 2 minutes, and after

20 minutes the membranes were removed from the solution and rinsed with ethanol. To make

sure that the loose silane was removed from the surface, the membranes were also sonicated for









[19] Harrell, C. C.; Kohli, P.; Siwy, Z.; Martin, C. R. Journal of the American Chemical
Society 2004, 126, 15646-15647.

[20] Steinle, E. D.; Mitchell, D. T.; Wirtz, M.; Lee, S. B.; Young, V. Y.; Martin, C. R.
Analytical Chemistry 2002, 74, 2416-2422.

[21] Lee, S. B.; Martin, C. R. Journal of the American Chemical Society 2002, 124, 11850-
11851.

[22] Wirtz, M.; Yu, S.; Martin, C. R. Analyst 2002, 127, 871-879.

[23] Martin, C. R.; Nishizawa, M.; Jirage, K.; Kang, M. Journal of Physical Chemistry B
2001, 105, 1925-1934.

[24] Lee, S. B.; Martin, C. R. Chemistry ofMaterials 2001, 13, 3236-3244.

[25] Martin, C. R. Science 1994, 266, 1961-1966.

[26] Martin, C. R.; Mitchell, D. T. Analytical Chemistry 1998, 70, 322A-327A.

[27] Hulteen, J. C.; Martin, C. R. Journal ofMaterials Chemistry 1997, 7, 1075-1087.

[28] Cepak, V. M.; Hulteen, J. C.; Che, G.; Jirage, K. B.; Lakshmi, B. B.; Fisher, E. R.;
Martin, C. R. Journal ofMaterials Research 1998, 13, 3070-3080.

[29] Miller, S. A.; Young, V. Y.; Martin, C. R. Journal of the American Chemical Society
2001, 123, 12335-12342.

[30] Kang, M.; Trofin, L.; Mota, M. O.; Martin, C. R. Analytical Chemistry 2005, 77, 6243-
6249.

[31] Xu, F.; Wharton, J. E.; Martin, C. R. Small 2007, 3, 1718-1722.

[32] Liang, W.; Martin, C. R. Journal of the American Chemical Society 1990, 112, 9666-
9668.

[33] Van Dyke, L. S.; Martin, C. R. Langmuir 1990, 6, 1118-1123.

[34] Yamada, K.; Gasparac, R.; Martin, C. R. Journal of the Electrochemical Society 2004,
151, E14-E19.

[35] Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268, 700-702.

[36] Brumlik, C. J.; Menon, V. P.; Martin, C. R. Journal ofMaterials Research 1994, 9, 1174-
1183.









CHAPTER 6
HYBRIDIZATION STUDIES OF DNA IN A PNA-MODIFIED NANOTUBE MEMBRANE
PLATFORM

Introduction

The development of DNA-sensors devices have attracted substantial research efforts

directed to gene analysis, detection of genetic disorders, analysis of single nucleotide

polymorphism in genes, tissue matching and forensic applications among others.304- 307 Various

techniques, including quartz crystal microbalance,308 electrochemical,309 fluorescence,310

electrophoretic,311 electrochemiluminescence,312 enzymatic,313 surface plasmon resonance,314 and

atomic force microscopy,315 have been developed to detect DNA hybridization. However, the

label-free detection methods have attracted more interest than other methods because they can be

simple, rapid, highly sensitive, and inexpensive.307

In 1953, Watson and Crick proposed a model for the structure of DNA.316 This model

predicted that DNA would exist as a helix of two complementary anti-parallel strands.

Soon after the first description of the double helix by Watson and Crick, it was shown, that

the two strands could be separated by heat or treatment with alkali. The reverse process, which

underlies all the methods based on DNA renaturation or molecular hybridization, was first

described by Marmur and Doty.317 It was quickly established that the two sequences involved in

duplex formation must have some degree of sequence complementarity and that the stability of

the duplex formed depends on the extent of complementarity. These remarkable properties

suggested ways to analyze relationships between nucleic acid sequences, and analytical methods

based on molecular hybridization. Since then, these methods have been rapidly developed and

applied to a variety of biological malfunctions.

Base-pairing (A-T and G-C) or hybridization is the underlying principle of DNA-sensors,

namely the strong interaction between two complementary strands. Accordingly, such devices









Chapter 3 describes the modeling and characterization of the transport properties of

proteins across antibody-functionalized nanotube membranes by using the dual-mode model.

This work demonstrated that the dual-mode model proposes here was a suitable model for the

characterization of the transport properties of anti-H-IgG and anti-M-IgG in M-IgG modified

membranes. The model demonstrated a good agreement with the experimental system studied

here. Besides, the findings showed in these studies allow addressing the concept of modulated

facilitated transport systems using antibody-modified nanotube membranes, and how the extent

of the facilitation is determined and controlled by the binding affinity between carrier and

permeating protein molecules. These studies can provide a new direction in the development of

more effective membranes for membrane-based separation.

In chapter 4, the development of metal ion affinity nanotube membranes was demonstrated

by exploiting the concept of IMAC separations in a nanopore-membrane-platform. Silica

nanotube membranes were functionalized with imidazole through a nucleophilic substitution

reaction between 2-mercaptoimidazol and 3-chloropropyl modified silica nanotubes carried out

in an aprotic medium. It was demonstrated that transport of histidine-tagged recombinant

proteins across these imidazole-modified silica nanotube membranes is controlled by the nature

of immobilized metal ion on the imidazole surface. Besides, the extensive characterization of the

imidazole functionalized membrane allowed the understanding of how coordination chemistry

with different divalent metal ions affects the functional behavior of the membrane and transport

properties during the permeation of six-histidine recombinant proteins. Furthermore, additional

results shown that by the appropriate choice of the transition metal ion for coordination with the

imidazole ligand, the binding strength and optical properties within the functional membrane are

also modified. Hence, these studies revealed how the presence of the immobilized metal ion




Full Text

PAGE 1

BIOANALYTICAL APPLICATIONS OF AFFINITY-BASED NANOTUBE MEMBRANES FOR SENSING AND SEPARATIONS By HECTOR MARIO CAICEDO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008 1

PAGE 2

2008 Hctor Mario Caicedo 2

PAGE 3

To my parents Herlinda and Luis, my wife Jessica, and my daughter Isabela 3

PAGE 4

ACKNOWLEDGMENTS I would like to thank my Ph.D. advisor, Dr. Charles R. Martin for supporting my work, and for all his lessons on becoming a more effective scientist. I would like to express a very special gratitude to Dr. Jim Deyrup for his tremendous help and encouragement when I started pursuing my Ph.D. I would also like to thank Dr. Nicolo Omenetto for being part of my committee, for his invaluable time and discussions about my research work, and for his contagious enthusiasm for science. I would like to thank Dr. Amelia Dempere, Dr. David Powell, and Dr. Thomas Lyons for serving on my graduate committee and for their time, feedback, and willingness to proofread this dissertation. I am very grateful to Dr. Myungchan Kang, former postdoc in the Martin group, with who I had the opportunity to work with for two years, during which he always provided many helpful insights about my research work. I would like to thank my lovely wife Jessica and my beautiful daughter Isabela for their love, patience, and continuous support during the last years of my graduate studies. Finally, my greatest gratitude goes to my family, for their encouragement in pursuit of my educational goals. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 ABSTRACT ...................................................................................................................................17 CHAPTER 1 INTRODUCTION AND BACKGROUND...........................................................................19 Introduction.............................................................................................................................19 Background.............................................................................................................................20 The Concept of the Template Synthesis..........................................................................20 Description of the Nanoporous Alumina Membrane Template......................................25 Two-step anodization method..................................................................................27 Pore growth mechanism...........................................................................................30 Sol-Gel Chemistry at a Glance........................................................................................32 Silane Chemistry a Versatile Tool for Surface Functionalization...................................35 Natures Transport across Membranes: An Inspiration for Membrane-Based Separations...................................................................................................................36 Fundamentals about Proteins...........................................................................................41 Antibodies................................................................................................................42 Recombinant proteins...............................................................................................44 Synopsis and Analytical Visualization of the Instrumental Techniques.........................46 Fluorescence spectroscopy.......................................................................................46 Fluorescence quenching...........................................................................................48 FTIR-ATR spectroscopy..........................................................................................49 X-ray Photoelectron Spectroscopy (XPS)................................................................51 Scanning Electron Microscopy (SEM)....................................................................52 Impedance spectroscopy..........................................................................................53 2 TRANSPORT AND FUNCTIONAL BEHAVIOR OF ANTIBODY-FUNCTIONALIZED NANOTUBE MEMBRANES............................................................55 Introduction.............................................................................................................................55 Experimental...........................................................................................................................57 Materials..........................................................................................................................57 Preparation of the Silica Nanotube Membranes..............................................................58 Antibody Immobilization................................................................................................58 Transport Experiments....................................................................................................59 Competitive Binding Assays...........................................................................................60 5

PAGE 6

Determination of Static Protein Binding Capacity and Binding Constants through Adsorption Isotherms...................................................................................................61 Effect of Equilibration Time on Flux and Selectivity via Dynamic Binding Capacity.......................................................................................................................62 Results and Discussion...........................................................................................................62 Electron Microscopy and Calibration Data.....................................................................63 Transport Properties of the anti-IgGs through PEG-Si-modified Control Membranes...................................................................................................................64 Competitive Binding Studies...........................................................................................65 Functional Behavior of the M-IgGand H-IgG-modified Membranes...........................67 Observed Functional Behavior is not a Transient Phenomenon.....................................68 Flux vs. Feed Concentration............................................................................................70 Effect of the Pore Size on the Flux and Selectivity Coefficient......................................71 Effect of the Immobilized-Antibody Orientation on the Flux.........................................73 Separation Based on a Simultaneous Combination of Size and Affinity Selectivity......74 Binding Capacities and Binding Constants.....................................................................77 Effect of the Temperature on the Flux and Binding Constant.........................................78 Effect of the Low Ionic Strength on the Selectivity........................................................83 Conclusions.............................................................................................................................85 3 MODELING AND CHARACTERIZATION OF THE TRANSPORT PROPERTIES OF PROTEINS ACROSS ANTIBODY-FUNCTIONALIZED NANOTUBE MEMBRANES BY USING THE DUAL-MODE MODEL..................................................87 Introduction.............................................................................................................................87 Definition of Dual-Mode Model.............................................................................................89 Results and Discussion...........................................................................................................95 Simulations of Permeability vs. Feed Concentration Based on the Dual-Mode Model...........................................................................................................................95 Experimental Description of the Dual-Mode Model.....................................................100 Effect of feed concentration on the flux.................................................................100 Experimental permeabilities as function of the feed concentration.......................101 Experimental evaluation of the dual-mode model parameters...............................102 Experimental evaluation of C sites and b..................................................................103 Experimental evaluation of the partition coefficients k p ........................................107 Experimental evaluation of the diffusion coefficients and solubility....................110 Fitting Experimental Data with the Theoretical Dual-Mode Model.............................114 Validation of the Dual-Mode Model.............................................................................119 Conclusions...........................................................................................................................121 4 METAL ION AFFINITY NANOTUBE MEMBRANES: PREPARATION, CHARACTERIZATION, AND CONTROLLED TRANSPORT OF SIX-HISTIDINE-TAGGED RECOMBINANT PROTEIN..............................................................................122 Introduction...........................................................................................................................122 Experimental.........................................................................................................................124 Materials........................................................................................................................124 6

PAGE 7

Preparation of the Silica Nanotube Membranes............................................................125 Preparation of Imidazole Functionalized Membranes...................................................125 Divalent Metal Ions Complexation...............................................................................126 Instrumental Characterization of the Imidazole Functionalized Membranes................127 Transport Experiments..................................................................................................128 Results and Discussion.........................................................................................................129 Imidazole Surface Modification and Characterization..................................................129 Fluorescence microscopy characterization.............................................................130 Characterization by FTIR-ATR spectroscopy........................................................135 Surface chemical composition by XPS..................................................................141 Transport Properties of Imidazole Membranes.............................................................146 Conclusions...........................................................................................................................149 5 TRANSPORT ANALYSIS OF PROTEINS ACROSS SILICA NANOTUBE MEMBRANES THROUGH A FLUORESCENCE QUENCHING-BASED SENSING APPROACH.........................................................................................................................151 Introduction...........................................................................................................................151 Experimental.........................................................................................................................152 Materials........................................................................................................................152 Preparation of the Silica Nanotube Membranes............................................................154 Antibody Immobilization..............................................................................................154 Labeling of anti-H-IgG with QSY-7.............................................................................155 Transport Experiments..................................................................................................156 Results and Discussion.........................................................................................................157 Experimental Principle Used for the Fluorescence Quenching Detection....................157 Verification of the Fluorescence Quenching System via H-IgG-FITC and anti-H-IgG-QSY Reaction in Solid State..............................................................................159 Control Transport Studies..............................................................................................160 Effect of the Feed Solution Concentration on the Rate of Quenching..........................162 Data Analysis Based on the Stern-Volmer Plots...........................................................164 Data Analysis Based on log-log Plots of Fluorescence vs. Time..................................167 Transport across Antibody-Modified Silica Nanotube Membranes..............................169 Conclusions...........................................................................................................................170 6 HYBRIDIZATION STUDIES OF DNA IN A PNA-MODIFIED NANOTUBE MEMBRANE PLATFORM.................................................................................................171 Introduction...........................................................................................................................171 Experimental.........................................................................................................................173 Materials........................................................................................................................173 Preparation of the Silica Nanotube Membranes............................................................174 Preparation of PNA-Modified Nanotube Membranes...................................................174 Nanotube Membrane Assembly....................................................................................176 Impedance Measurement for Membrane Characterization...........................................177 Impedance Data Analysis..............................................................................................177 Results and Discussion.........................................................................................................178 7

PAGE 8

Impedance Characterization of the Unmodified Membranes........................................179 Functionalization of the Silica nanotube Membranes...................................................180 Impedance Sensing of DNA-PNA Hybridization.........................................................182 Effect of the Pore Size on R m .........................................................................................183 Stability and Reproducibility of the PNA-modified Membranes..................................185 Conclusions...........................................................................................................................187 7 CONCLUSIONS..................................................................................................................189 LIST OF REFERENCES.............................................................................................................194 BIOGRAPHICAL SKETCH.......................................................................................................212 8

PAGE 9

LIST OF TABLES Table page 2-1. Fluxes and selectivity coefficients for anti-M-IgG and anti-H-IgG across the M-IgG and H-IgG-modified membranes................................................................................67 2-2. Thermodynamic parameters of G 0 associated with the affinity interaction of anti-M-IgG and M-IgG-modified membranes as a function of the temperature.......................82 3-1. Dual-mode model parameters used for the permeability plot simulations........................97 3-2. Thermodynamic parameters from adsorption isotherms of anti-M-IgG and anti-H-IgG on M-IgG modified membranes, according to the Langmuir monolayer model......107 3-3. The k p determined by various approaches.......................................................................108 3-4. Validation of the dual-mode model parameters...............................................................121 4-1. FTIR-ATR characteristic bands of imidazole functionalized nanotube membranes.......138 4-2. XPS surface composition and binding energies for bare alumina, chlorosilane modified, and imidazole functionalized membranes.......................................................143 4-3. Fluxes and selectivity coefficients for 6His-UHRP across imidazole functionalized metal ion affinity nanotube membranes...........................................................................149 5-1. Relationship between the Stern-Volmer quenching constant (K SV ) with the feed solution concentration of anti-H-IgG-QSY.....................................................................167 5-2. Relationship between slope of the log-log plots for the fluorescence signal vs. time with the feed solution concentration of anti-H-IgG-QSY................................................168 5-3. Data for the Stern-Volmer quenching constant (K SV ) and slope of the log-log plots for antibody-modified membranes...................................................................................169 6-1. Variations of R m at the different steps of modification on a 100 nm pore diameter alumina template..............................................................................................................182 9

PAGE 10

LIST OF FIGURES Figure page 1-1. Scanning electron micrograph pointing up the honeycomb structure of nanoporous alumina...............................................................................................................................25 1-2. Structure of nanoporous alumina.......................................................................................26 1-3. Electrochemical cell setup for nanoporous alumina growth..............................................28 1-4. Scanning electron micrographs of a nanoporous alumina membrane anodized at 50 V in 5% oxalic acid, (A) before and (B) after elimination of the barrier layer.....................29 1-5. Scanning electron micrograph of a highly-ordered nanoporous alumina membrane obtained at 50 V in 5% oxalic acid....................................................................................30 1-6. Outline of the steps involved in the sol-gel synthesis........................................................32 1-7. Fluorescence spectra obtained from the biofunctionalized (a) alumina template and (b) templated silica nanotube membranes with anti-Human IgG labeled with Alexa 488......................................................................................................................................34 1.8. Reactions involved in the alkoxysilane functionalization. Scheme obtained at http://www.dowcorning.com/content/publishedlit/SILANE-GUIDE.pdf.........................35 1-9. Types of specialized transport in living membranes (A) carriers, and (B) channels.........37 1-10. Flux concentration dependence for passive and facilitated transport................................38 1-11. Concentration gradients formed across the membrane during passive and facilitated transport.............................................................................................................................39 1-12. Basic structure of the immunoglobulin molecule..............................................................42 1-13. Intermolecular attractive forces involves in the antibody-antigen binding. Scheme obtained from www.fleshandbones.com/readingroom/pdf/291.pdf..................................44 1-14. Jablonski diagram of the electronic states during the fluorescence process......................47 2-1. Immobilization of antibodies on the template synthesized silica nanotube membranes.........................................................................................................................56 2-2. Flow-through fluorescence system used to measure the flux of the permeating proteins...............................................................................................................................59 2-3. Electron micrographs of (A) branch side, (B) solution anodization side, and (C) cross-section of the alumina template membranes, and (D) silica nanotubes after removal from the membrane..............................................................................................63 10

PAGE 11

2-4. Typical plots of (A) calibration data, and (B) calibration curve for Alexa-488 labeled anti-H-IgG..........................................................................................................................64 2-5. Flux plots for anti-M-IgG () and anti-H-IgG () through PEG-Si-modified control membranes. The feed concentration was 2.65 nM (lower curves) and 21.2 nM (upper curves)....................................................................................................................65 2-6. Fluorescence spectra of (A) M-IgGand (B) H-IgG-modified membranes after equilibration with a solution that was 21.2 nM in both (a) anti-H-IgG, and (b) anti-M-IgG................................................................................................................................66 2-7. Flux plots for anti-M-IgG () and anti-H-IgG () in (A) M-IgG-modified membrane, and (B) H-IgG-modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins.......................................................................68 2-8. Dynamic binding capacity study using M-IgG-modified membranes. The flux and selectivity coefficient vs. days of equilibration are shown for the transport of anti-M-IgG and anti-H-IgG. The feed solution concentration was 21.2 nM in both of the permeating proteins............................................................................................................69 2-9. Flux versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. The feed solution concentrations were equimolar in both of the permeating proteins............................................................................................................70 2-10. Typical plot of flux versus feed concentration for anti-M-IgG or anti-H-IgG across PEG-Si modified membrane. The data plotted is an average of the fluxes for anti-M-IgG and anti-H-IgG for each specific feed concentration..................................................71 2-11. Effect of the pore size on (A) the flux, and (B) selectivity coefficients for M-IgG modified membranes. The feed solution concentrations was 2.65 nM both in anti-H-IgG and anti-M-IgG...........................................................................................................72 2-12. Flux plots for anti-M-IgG () and anti-H-IgG () in well-oriented M-IgG-modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins...............................................................................................................................73 2-13. Scanning electron micrograph of (A) the surface, and (B) cross section of a highly-ordered nanoporous alumina membrane obtained at 30 V in 5% oxalic acid...................75 2-14. Flux plots for anti-M-IgG () and anti-H-IgG () in (A) PEG-Si-modified membrane, and (B) well-oriented M-IgG -modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins.........................................75 2-15. Flux plots for anti-M-IgG () and F(ab) 2 -anti-H-IgG () in well-oriented M-IgG -modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins............................................................................................................76 11

PAGE 12

2-16. Dissociation constants and binding capacities from adsorption isotherms of anti-M-IgG and anti-H-IgG on H-IgGand M-IgG-modified membranes, according to the Langmuir monolayer model...............................................................................................78 2-17. Flux plots for anti-M-IgG () and anti-H-IgG () in well-oriented M-IgG -modified membranes showing the effect of temperature. The feed solution concentration was 21.2 nM in both of the permeating proteins.......................................................................79 2-18. Effect of temperature on the flux and selectivity coefficients...........................................79 2-19. Adsortion isotherms carried out with well-oriented M-IgG-modified membranes for anti-M-IgG at three different temperatures: 23, 32, and 42 C..........................................80 2-20. Affinity temperature dependence of anti-M-IgG in well-oriented M-IgG-modified membranes.........................................................................................................................81 2-21. Flux plots for anti-M-IgG () and anti-H-IgG () across M-IgG -modified membranes, (A) without equilibration, and (B) after equilibration. The feed solution concentration was 2.65 nM in both of the permeating proteins.........................................84 3-1. Dual-mode transport model...............................................................................................88 3-2. Permeability plot simulations for the transport of (a) BM and (b) NBM across carrier-modified membranes. The parameters are summarized in Table 3-1....................96 3-3. Permeability plot simulations for the transport of BM across carrier-modified membranes illustrating the effect of the term F on the dual-mode model. (a) F = 1, (b) F = 0.5, and (c) F = 0. The other parameters are summarized in Table 3-1................97 3-4. Permeability plot simulations for the transport of BM across carrier-modified membranes showing the effect of the binding affinity on the permeability. (a) b = 1 x 10 11 cm 3 /moles, and (b) b = 3 x 10 10 cm 3 /moles. The other parameters are summarized in Table 3-1.................................................................................................100 3-5. Flux versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. The feed solution concentration was equimolar in both of the permeating proteins..........................................................................................................101 3-6. Experimental obtained permeability versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane..................................................102 3-7. Curve of adsorption dynamics for anti-M-IgG () in a M-IgG modified membrane. Inset shows the changes of fluorescence signal with time...............................................103 3-8. Adsorption isotherms for anti-M-IgG (), and anti-H-IgG () binding on M-IgG modified membranes. Also the adsorption isotherm for anti-M-IgG () using a control membrane is shown.............................................................................................104 12

PAGE 13

3-9. Adsorption isotherm for anti-M-IgG () binding on M-IgG-PG modified membranes.......................................................................................................................106 3-10. Extrapolations of the adsorption isotherms at infinite dilute concentration for anti-M-IgG (), and anti-H-IgG () in M-IgG modified membranes........................................108 3-11. Permeability plot simulations illustrating the effect of the partition coefficient on the dual-mode model. Simulation parameters: D = 5 x 10 -8 cm 2 /sec, b = 9 x 10 10 cm 3 /moles, C sites = 3.5 x 10 -8 moles/cm 3 and k p were respectively, 1 (), 0.1 (), 0.01 (), and 0.001()....................................................................................................109 3-12. Concentration-dependence of the solubility in term of the dual-mode model for anti-M-IgG (), and anti-H-IgG () across M-IgG modified membranes............................111 3-13. Correlation plots between permeability and solubility based on the concentration dependence for (A) anti-H-IgG and (B) anti-M-IgG across M-IgG modified membranes.......................................................................................................................112 3-14. Experimental and calculated obtained permeability versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane.......................114 3-15. Experimental obtained diffusion coefficients versus feed concentration for (A) anti-H-IgG and (B) anti-M-IgG across M-IgG modified membrane......................................117 3-16. Plots for experimental and calculated obtained permeability against (1 + bC feed ) -1 for (A) anti-H-IgG, and (B) anti-M-IgG across M-IgG modified membrane.......................120 4-1. Reaction mechanism of imidazole functionalized nanotube membranes........................126 4-2. Flow-through UV-Vis detection system used to measure the flux of six-histidine-tagged ubiquitin human recombinant protein..................................................................129 4-3. Fluorescence spectra of (a) silica nanotube and (b) imidazole functionalized membranes.......................................................................................................................130 4-4. Fluorescence spectra of imidazole functionalized membranes before (upper) and after (lower) metal ion immobilization: (A) Co +2 (B) Cu +2 (C) Ni +2 and (D) Zn +2 ......131 4-5. Fluorescence quenching (F/F 0 ) for imidazole functionalized membranes in the presence of nitrate solutions prepared in ethanol that were 1 mM of the respective divalent metal ion.............................................................................................................132 4-6. Fluorescence spectra of imidazole functionalized membranes (a) before and (b) after metal ion immobilization using an ethanol solution that was 1 mM in CuCl 2 ................133 4-7. Copper concentration dependence on fluorescence enhancement for imidazole functionalized membranes...............................................................................................134 13

PAGE 14

4-8. (A) FTIR-ATR spectra of bare alumina and imidazole functionalized membranes. (B) FTIR-ATR spectra of imidazole functionalized membranes in a narrow spectral region illustrating the effect of using two different backgrounds for the spectrum subtraction: (a) air, and (b) 3-chloropropyl silica nanotube membrane...........................136 4-9. Transmission FTIR spectrum of pure 2-mercaptoimidazole in KBr...............................136 4-10. FTIR-ATR spectra of imidazole functionalized membranes showing the effect of the metal ion immobilized on the thioamide bands: (a) Zn +2 (b) Co +2 and (c) Cu +2 ...........137 4-11. Transmission FTIR spectra of the alumina template film for each of the modification steps illustrated in Figure 4-1: (a) bare alumina template, (b) silica, (c) 3-chloropropyl silane modified and (d) imidazole functionalized nanotube membranes, respectively......................................................................................................................139 4-12. XPS survey spectrum for the bare alumina template.......................................................141 4-13. XPS survey spectra for (a) chlorosilane and (b) imidazole modified membranes..........142 4-14. XPS multiplexes analysis at high resolution for the imidazole functionalized membrane.........................................................................................................................144 4-15. High-resolution XPS data of the C1s peak for (a) bare alumina template, (b) chlorosilane modified, and (c) imidazole functionalized membranes.............................144 4-16. XPS survey spectra for imidazole modified membranes after immobilization of (a) Cu +2 and (b) Zn +2 ............................................................................................................145 4-17. UV-Vis spectrum for 50 M of 6His-UHRP measured in 50 mM phosphate-buffered saline solution pH 8.0 that also contained 300 mM sodium chloride and 2 mM imidazole..........................................................................................................................146 4-18. Typical plots of (A) calibration data and (B) calibration curve for 6His-UHRP.............147 4-19. Flux plots for 6His-UHRP in imidazole functionalized metal ion affinity nanotube membranes that had immobilized different metal ions: (a) Cu+2, (b) Ni+2, (c) Co+2, and (d) Zn+2. The feed solution concentration was 1.87 M for all permeation experiments......................................................................................................................147 4-20. Effect of the divalent metal ion immobilized on the flux of 6His-UHRP across imidazole functionalized metal ion affinity nanotube membranes..................................148 5-1. Flow-through fluorescence system used to measure the changes of fluorescence intensity in the receiver solution......................................................................................157 5-2. Structures of (a) QSY-7 and (b) FITC.............................................................................158 5-3. Spectra for FITC and QSY-7...........................................................................................158 14

PAGE 15

5-4. Molecular recognition reaction used to promote the fluorescence quenching................159 5-5. Fluorescence image and spectra respectively for a H-IgG-FITC-modified membrane, (A-C) before exposure, and (B-D) after exposure to a sample solution of anti-H-IgG-QSY..................................................................................................................................160 5-6. Typical plots for (A) changes of fluorescence signal, and (B) concentration of unquenched protein in function of time for two different control transport/detection systems: (a) anti-H-IgG/H-IgG-FITC, and (b) anti-H-IgG-QSY/R-IgG-FITC...............161 5-7. Typical plots for (A) changes of fluorescence signal and (B) concentration of unquenched protein in function of time for three different feed solution concentrations: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control)........163 5-8. Stern-Volmer plots of I 0 /I against concentration of anti-H-IgG-QSY in the receiver solution transported across silica nanotube membranes. The effect of the feed solution concentrations of anti-H-IgG-QSY is shown: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control).................................................................................165 5-9. Effect of the feed concentration on the Stern-Volmer quenching constant K SV ..............166 5-10. log-log plots of the of fluorescence signal vs. time showing the effect of feed solution concentration of anti-H-IgG-QSY: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control)................................................................................................167 5-11. Linear relationship between slopes obtained from the log-log plots with the feed concentration of anti-H-IgG-QSY...................................................................................168 5-12. Typical plots for (A) changes of fluorescence signal and (B) concentration of unquenched protein in function of time illustrating the transport of anti-H-IgG-QSY across (a) H-IgG-, and (b) M-IgG-modified membranes. The feed solution concentration was 33.3 nM for both experiments............................................................169 6-1. Structures of (a) PNA, and (b) DNA...............................................................................172 6-2. Immobilization of biotinylated-PNA on templated synthesized silica nanotube membranes.......................................................................................................................175 6-3. Nanotube membrane assembly........................................................................................176 6-4. General equivalent circuit model used for impedance data analysis...............................178 6-5. Nyquist plots (Z im vs. Z re ) for impedance measurement of (a) alumina and (b) silica-nanotube assembled membranes in highly purified water...............................................179 6-7. Effect of the wetting time on the resistance impedance measurement for an assembled unmodified alumina membrane 200 nm pore diameter.................................180 15

PAGE 16

6-8. Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water: (a) alumina, (b) after synthesis of silica-nanotube, (c) after silanization, (d) after biotinylation, (e) after streptavidin immobilization, (f) after PNA immobilization, and (g) after DNA hybridization. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz..................................................181 6-9. Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water for a PNA-modified nanotube membrane: (a) before, and (b) after 10 minutes of DNA hybridization. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz.........................................................................183 6-10. Effect of the membrane pore size on the membrane resistance: (A) 200 nm, and (B) 100 nm. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz.......184 6-11. Study of the hybridization process stability with time: (A) unmodified control alumina membrane, and (B) PNA-modified membrane. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA.......................................................................................................185 6-12. Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water for a PNA-modified nanotube membrane: (a) before, (b) after adding 3 nmoles of DNA, and (c) after 6 nmoles of DNA. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz..................................................186 6-13. Reproducibility study of the PNA-modified nanotube membranes, before and after the hybridization process. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA.......................187 16

PAGE 17

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy BIOANALYTICAL APPLICATIONS OF AFFINITY-BASED NANOTUBE MEMBRANES FOR SENSING AND SEPARATIONS By Hctor Mario Caicedo August 2008 Chair: Charles R. Martin Major: Chemistry Nanotechnology has played an important role in the development of research and technology during the last two decades. The contribution of nanotechnology in different fields, along with the versatility of the constructed nanoscale materials, have made nanotechnology one of the most suitable tools to develop particular nanostructures to realize a desired function and application. A nanostructure is simply an entity at the nanometer scale with one, two or three dimensional features. Since nanotechnology covers a broad range of nanoscale materials, to simplify nanotechnology, it can be classified into two categories based on how the nanostructures are prepared: top-down and bottom-up. In the top-down methods, the nanostructures are constructed by chiseling larger bulk materials into entities of smaller size. Conversely, in the bottom-up case, small units are grown or assembled into their desired size and shape. The nanoporous materials specifically have attracted a lot of attention because they can be used for the synthesis of a variety of functional nanostructures of great usefulness in technology. These porous nanostructures usually combine many of the advantages of the top-down and bottom-up methodologies such as flexibility, size controllability, and cost. The research presented in this work utilizes nanoporous membranes to develop porous nanostructured platforms with potential applications in sensing and separations. In particular, this work is 17

PAGE 18

centered in fundamental studies for bioanalytical applications of affinity-based nanotube membranes for sensing and separations. A bottom-up methodology like the template synthesis was used to produce silica nanotubes inside of the pores of alumina membrane. The functionalization of the inside walls of these silica nanotube membranes allowed control of the functional behavior and properties of the nanostructured membrane during membrane-based separations and sensing. The general scheme of the work presented here, is distributed in seven chapters. The first chapter consists of a general description of the theory and fundamentals as background supporting evidence for the work presented here. The template synthesis method, the fabrication of the porous alumina membranes and the possibilities that silane chemistry offers in order to functionalize the nanostructured membranes is discussed. In addition, a brief overview of the fundamentals of biological production of proteins, antibodies, and recombinant proteins is described. After this introduction, the studies presented herein are centered in potential bioanalytical applications of the silica nanotube membranes. 18

PAGE 19

CHAPTER 1 INTRODUCTION AND BACKGROUND Introduction The use of membranes in industrial separations, substituting the traditional methods of separation, has become a more accepted procedure every day. Features such as compactness, simple and efficient operation, and low energy consumption have made membrane-based separations a fine alternative procedure. 1, 2 Membrane-based separations have found an extensive range of applications in technology, such as water and wastewater treatment, electrodialysis, gas separations, and fuel cell development. 2 The most important property of membranes is their ability to control the rate of permeation of different species. Traditionally, membranes have been used for sizeand charge-based separations with great success by using relatively low-resolution requirements. 3-7 However, some applications require the development of new membranes, because these rudimentary forms of selectivity are insufficient to allow membranes to compete with the traditional chromatographic methods. Current research and development efforts in membrane-based separations are directed toward drastic improvements in selectivity keeping the inherent characteristics of membranes. 8 In the search of new advanced membrane materials, the development of functionalized-membranes with a selective transport function executed through a specific molecule carrier has become in an interesting alternative to modulate the transport during membrane-based separations. 9, 10 This concept has been inspired in analogy by the strategies that biological membranes use to accomplish transport. Thus, the combined use of synthetic membranes with biological molecules such as antibodies and nucleotides has become important in the development of applications both in the biomedical and biotechnological field. 11 By using this combination, membranes have also acquired an important role in the design and development of biosensors. 12 In particular, nanotube membranes prepared by the general method 19

PAGE 20

called template synthesis to prepare nanomaterials have been extensively investigated. 4-6, 9, 10, 13-24 This method entails synthesis of the desired material within the cylindrical and monodisperse pores of a nanopore membrane or other solid. 25, 26 Both silica and gold nanotube membranes prepared by this method have demonstrated the capacity of being a novel platform for membrane-base separations and sensors. 14, 18 One of the most important features of the nanotube membranes prepared by the template synthesis when they are used in membrane-base separations, is that the deposited nanotubes are embedded within a mechanical strong support membrane, making it possible that they can act as conduits during the transport of permeating species between the solutions placed on either side of the membrane. Previous studies have shown that, by controlling the inside diameter of the nanotubes and the chemical functionalities present in the tube walls, highly-selective membranes for chemical and biochemical separations and sensors can be prepared. 46, 9,10, 13-24 This chapter provides a general description of the theory and fundamentals as background supporting evidence for the work presented here. The template synthesis method, the fabrication of the porous alumina membranes and the possibilities that silane chemistry offers in order to functionalize the nanostructured membranes it is discussed. In addition, a brief overview of the fundamentals of biological production of proteins, antibodies and recombinant proteins is described. At the end of the chapter, a brief summary of some of the instrumental analytical techniques used during the development and characterization of the applications described in this work is provided. Background The Concept of the Template Synthesis Our group has been pioneered in investigating a method called template synthesis for the fabrication of a variety of nanomaterials. 27 This template method is a general approach for preparing nanomaterials that involves the synthesis or deposition of the desired material within 20

PAGE 21

the pores or cavities of a nanopore membrane or other solid surface. Therefore, depending on the template structure and synthetic method used, a diverse array of nanostructured materials can be prepared. 28-31 Thus for example, from porous membranes is possible to prepare hollow nanotubes or solid nanowires, and by using solid spheres or porous spheres respectively, capsules or nanoporous particles can also be fashioned. Martin et al. have prepared nanotubes and nanowires composed of many types of materials, including conductive polymers, 32-34 metals, 35-37 oxides, 38 semiconductors, 39 carbon, 29, 31 and DNA. 40, 41 The variety of synthetic methodologies used for these nanofabrication include electroless plating of metals; electrodeposition of insulating polymers, conductive polymers, semiconducting oxides and metals; chemical vapor deposition; sol-gel chemistry; chemical polymerization; carbonization; and layer-by-layer template synthesis. Using either electrochemical 33, 34 or chemical 42, 43 oxidative polymerization of a equivalent monomer, Martin et al. have prepared nanowires and nanotubes out of conductive polymers such as polypyrrole, poly(3-methylthiophene) and polyaniline. Metals having nanoscopic dimensions with attractive optical properties have also been prepared by template synthesis. 44-45 For example, a macroscopic sample of pure gold has only one color, but nanoparticles of gold can show essentially all the colors in the visible region of the electromagnetic spectrum, depending on the size and the shape of the nanoparticles. 25 Hornyak et al. explored the optical properties of nanoscopic gold particles prepared by electrochemical deposition of gold within the pores of nanoporous alumina membranes. 44-45 They found that these particles achieve a plasmon resonance absorption limit that depended on the gold nanoparticle diameter. 21

PAGE 22

In the Martin group there is a commonplace methodology for the electroless deposition of gold within the pores of polycarbonate (PC) 10, 37 and polyethylene terephthalate (PET) 16, 46 template membranes. Depending on the plating time, it is possible to generate gold nanowires or nanotubes. Thus, in the nanotube situation, by controlling the plating time, the inside diameter can be varied. The electroless deposition of gold on the pore wall of these polymeric membranes is currently the best known method to precisely control the inside diameter of the template-synthesized nanotubes. 47 Martin et al. have shown that because the inside diameter of these gold nanotube membranes can be small relative to molecular dimensions, the nanotube membrane can be used to cleanly separate small molecules on the basis of molecular size. 47 By varying the rate of the plating reaction the diameter of the nanotubes was smaller in the faces of the template membrane. Thus, the flux and selectivity were determined for the bottleneck of the nanotubes. Together with size-based selectivity, they also demonstrated that generic chemical transport selectivity can be introduced into this membranes. 48, 49 More recently, Martin et al. have investigated gold nanotube membranes where the inner pore surfaces were systematically modified for biochemical functionalization to accomplish selective permeation of DNA. 10 They demonstrated a DNA-functionalized nanotube membrane for the separation of DNA by single-base mismatches selectivity via facilitated transport. 10 These DNA-functionalized nanotube membranes selectively recognize and transport the DNA strand that was complementary to the immobilized hairpin-DNA (carrier) within the pores, relative to DNA strands that are not complementary to the immobilized DNA. Another synthetic method used for the manufacture of nanotubes is the sol-gel chemistry to produce nanotubes of silica and other inorganic materials. The silica nanotubes have been used 22

PAGE 23

as paradigm to show the power of the template method for preparing nanotubes for biotechnological applications. 50 Silica nanotubes are easy to make, and they can be derivatized with an enormous variety of different chemical functional groups using simple silane chemistry with commercially available reagents. Mitchell et al. developed the concept of smart nanotubes that can be used for biological extraction and biocatalysis. 38 Nanotubes removed from the template and differentially functionalized both in the inner and outer nanotube surfaces were used for this application. Such differentially functionalized nanotube was used as smart nanophase extractor to remove specific molecules from solution. Thus, hydrophobic functionalized nanotube on the inside, with a hydrophilic nanotube exterior was used to extract lipophilic compounds from aqueous solutions. In the same work, Mitchell et al. showed that nanotube functionalized with an enantioselective antibody selectively extracts the enantiomer of a drug molecule that binds to the antibody from the racemic mixture. 38 Lee et al. demonstrated the use of silica nanotube membranes for enantioseparations. 9 They found that nanotubes embedded within the pores of the alumina template and functionalized with a Fab fragment antibody that binds the RS isomer of a drug molecule over the SR isomer, facilitated the transport of the RS enantiomer. They also found that the transport selectivity and binding affinity of RS over SR could be tuned by controlling the inside diameter of the membrane containing silica nanotube, and by addition of DMSO to the protein buffer solution, respectively. Gasparac et al. explored the template synthesized nano test tubes as an alternative payload-release strategy by using the template synthesis within the pores of nanopore alumina films. 51 These hollow nanotubes were closed on one end and open on the other. Therefore, if these nano 23

PAGE 24

test tubes could be filled with a payload and then the open end corked with a chemically labile cap, they might function as a universal delivery vehicle. More recently, Hillebrenner et al. demonstrated nano test tubes corked by chemical self-assembly. 52, 53 Buyukserin et al. have also prepared silica nano test tubes by using nanopore photoresist films as templates. 54 They used a plasma etch method, by using a nanopore alumina film as a mask, to etch a replica of the alumina pore structure into the surface of a photoresist polymer film. They found that the length of the test tubes is determined by the thickness of the porous part of the photoresist film. It is very important to control the inside diameter of the fashioned nanotubes, when the alumina membranes are used as template. In order to control the nanotube inside diameter a substitute template synthetic methodology has been considered, where the deposition of the film-forming material is based on layer-by-layer deposition. In this case, the wall thickness, and equally the inside diameter of the nanotubes would be determined by the number of layers of the material deposited along the pore wall. Using this approach, Ai et al. attempted to deposited layers of polyelectrolyte in the pores of a nanopore alumina membrane. 55 Kovtyukhova et al. have developed a method based on alternate SiCl 4 /H 2 O deposition/reaction cycles to make and control the thickness of silica nanotubes. 56 Hou et al. used Mallouks alternating ,-diorganophosphonate Zr(IV) chemistry to deposit layered nanotubes along the pore walls of an alumina template membrane. 40-41 For these nanotubes, the layers were held together by bonds between the phosphonate groups and the Zr(IV) ions, and the polyvalence of this interaction provided cross-linking that imparts the structural integrity of the layered nanotubes. 40 Hou et al. also used this synthetic strategy for the synthesis of DNA nanotubes where the bulk of the tube 24

PAGE 25

was composed of DNA, and the layers were held together by hybridization of complementary DNA strands. 40-41 The remarkable versatility offered by the material fabricated using template synthesis such as: size, shape, chemistry methodology applied, and systematic functionalization, make those nanostructured-template materials practical for particular applications. Description of the Nanoporous Alumina Membrane Template Nanoporous alumina films, which are produced by the electrochemical anodization of aluminum in suitable acidic or basic electrolytic solution, 44, 57 63 have been investigated and used as starting material for the fabrication several kinds of nanostructures, due to its very well ordered porous structure with high aspect ratio (thickness divided by the diameter). In recent years, simultaneously with the improvements in degree of ordering, these nanoporous alumina membranes have become a popular template system for the synthesis of various functional nanostructures. 9, 64 70 The porous structure appears as a honeycomb-like arrangement consisting of a close-packed array of columnar alumina units called cells, each containing a central straight pore, as shown in the scanning electron micrograph in Figure 1-1 from an alumina membrane obtained in our laboratory. Figure 1-1.Scanning electron micrograph pointing up the honeycomb structure of nanoporous alumina. 25

PAGE 26

By changing the anodization conditions such as electrolyte used, its concentration, the applied voltage and the period of anodization, the pores length, radius and the cell population density can be easily varied. The pore diameter depends on the applied voltage used for the anodization, and the thickness is a function of the anodization time. Pore diameters can be varied between 20 and 200 nm, with lengths between few tens of nanometers up to hundreds of microns. The pore size can be also controlled by the pore-widening treatment of dipping the anodized porous alumina in an appropriate acid solution after the anodization. This acidic solution is generally used to remove the barrier layer, which is a dense insulated oxide layer that separates the underlying aluminum surface and the porous oxide film (Figure 1-2). 58, 71 However before the barrier layer is eliminated, the porous alumina film has to be removed from the underlying aluminum metal in order to collect a free-standing membrane. There are two methods of removing the remaining aluminum substrate in order to obtain the free-standing membrane. The first one is by dissolution with a saturate HgCl 2 solution, and the second is a voltage-reduction method. The latter entails stepwise reduction of the potential to obtain an interfacial oxide layer that has a network of much smaller pores. 44 After formation of this highly porous layer, the alumina film is immersed into an acidic detachment solution which results in rapid dissolution of the interfacial oxide. 44 Figure 1-2.Structure of nanoporous alumina. 26

PAGE 27

There are two reported methods for the fabrication of highly ordered anodic nanoporous alumina membranes. One is the two-steps anodization method, which is the commonplace method used in our laboratory to produce the alumina membranes. Since the ordering of the pore configuration (self-organization) depends on the anodization time in the acidic electrolyte solution, two steps of anodization are required to obtain a highly ordered pore structure. 61, 62, 72 This method will be discussed in more detail in the next section. The second method is based on pretexturing of the aluminum substrate by using nanoindentation with a master mold. 73, 74 Masuda et al. used a SiC mold with a hexagonal array structure prepared by conventional electron beam lithography. 73, 74 In this pretexturing method, the depressions formed in the aluminum serve as initiation sites for the pore creation at the initial stage of the anodization. An ideally ordered nanoporous alumina membrane is formed after anodization of the aluminum substrate with the transferred textured pattern. Two-step anodization method The degree of the ordering of the pore initiation at the surface of the anodic porous alumina is low because the pores develop randomly at the initial stage of the anodization. To improve the ordering of the surface side of the anodic porous alumina, two-steps anodization is effectively adopted. 61, 62, 72 Two separated anodization processes are required for this procedure. The first anodization process consists of a long-period anodization to form the highly ordered concave nano-array of nucleation sites at the alumina/aluminum interface. The second anodization is performed after the removal of the oxide formed in the first anodization step. Thus, after the removal of the oxide, an array of highly ordered dimples is formed on the underlying aluminum substrate. These dimples act as initiation sites for the pore development during the second anodization. This process generates an ordered pore structure through the entire oxide layer. In the following paragraphs, the methodology used in our laboratory for the fabrication of alumina 27

PAGE 28

membranes will be discussed, which consists of three stages: polishing of the aluminum substrate, the two anodization steps, and the detaching step in order to have a free-standing membrane. Polishing of aluminum foils: In order to obtain a high quality alumina film, the starting aluminum material, must be very smooth. Initially, the high purity aluminum foils (99.999%) are mechanically polished with slurry of aluminum particles. If, to the naked eye, the aluminum foil shows severe scratches a mechanical polishing with fine sand paper is applied until the scratches disappear. The mechanical polishing is then followed by an electrochemical polishing. In this process, a potential difference of 15 V is applied between the anode aluminum foil and a lead plate which serve as cathode, by using a polishing electrolyte solution (95% concentrated phosphoric acid, 5% concentrated sulfuric acid and 20 g/L chromic oxide) heated to 70 C. Electropolishing steps of 5 minutes are successively repeated as many times as it is necessary, until the aluminum foil has a mirror-like surface. Two anodization steps: Electropolished aluminum is oxidized by using an electrochemical cell setup like the one presented in Figure 1-3. Figure 1-3.Electrochemical cell setup for nanoporous alumina growth. 28

PAGE 29

The first anodization step is performed in 5% oxalic acid as electrolyte for 12 hours, at 2 C, under a constant voltage. It should be pointed out that the pore diameter is controlled via anodization voltage. Thus, smaller pores require lower applied voltage and more conducted electrolytes such as sulfuric acid. 75 Large pores need larger voltages, which causes a high rate of dissolution in highly conductive electrolytes. 76 For this reason, the fabrication of larger pores requires lower conductive electrolytes, such as oxalic acid. 76 After the first anodization, the formed alumina film is removed in an etching solution that contains 0.4 M in phosphoric acid and 0.2 M in chromic oxide at 60 C. This anodization removal step leaves a uniform concave nano-array of nucleation sites in the underlying aluminum metal that correspond to each pore. The second anodization is performed at the same voltage and electrolyte conditions used in the first anodization. This second anodization step is carried out for a length of time, depending on the thickness of alumina membrane desired. In our laboratory, we obtained alumina membranes of thicknesses ranging from 0.3 to 150 m. Figure 1-4.Scanning electron micrographs of a nanoporous alumina membrane anodized at 50 V in 5% oxalic acid, (A) before and (B) after elimination of the barrier layer. Detaching step: The residual unoxidized aluminum is removed first with a saturated solution of HgCl 2 Then, the remaining barrier oxide layer is removed by etching the alumina film in a 10% phosphoric acid solution. Since the barrier layer thickness also varies with 29

PAGE 30

increasing voltage the etching step requires longer times for anodized membranes with bigger pores. Figure 1-4 shows the scanning electron micrographs of an alumina membrane obtained in our laboratory before and after elimination of the barrier layer. Thus, the resulting free-standing alumina membrane is obtained. In our laboratory, it is common to use a method with a two-step anodization to obtain highly ordered porous alumina membranes, as can be seen in Figure 1-5. Figure 1-5.Scanning electron micrograph of a highly-ordered nanoporous alumina membrane obtained at 50 V in 5% oxalic acid. Pore growth mechanism At the initial stage of anodization of aluminum, the sites of pore initiation are spread randomly over the aluminum surface. The ordering of the cell arrangement only proceeds with the anodizing time after steady state in the growth of the porous structure is established. 61, 62, 72 In addition, it is also known that the cell size and barrier-layer thickness are directly related to the anodizing potential. 58, 61, 62, 71, 72 The proposed mechanism for the formation of highly ordered porous alumina membranes has been explained as the result of two competing processes: 1) the pore initiation process due to a geometric effect, which is called field-assisted dissolution process 72 and 2) the self30

PAGE 31

organization of the pores which is considered to be triggered by mechanical stress at the alumina/aluminum interface. 76 In a first step the barrier layer is formed jointly with numerous defects in the aluminum surface for the pore initiation. Due to the decreasing current with the increase of the thickness of the barrier layer the formation of pores is initiated at defect positions. Then, assisted by the electric field, the alumina film developed at the metal-oxide interface at the pore walls are preferentially dissolved by the acidic electrolyte. For this reason, when the electric field increases by increasing the voltage, the dissolution process is accelerated and larger pores are formed. Thus, pores grow perpendicular to the surface via field-assisted oxide dissolution at the oxide/electrolyte interface in dynamic equilibrium with the oxide growth at the metal/oxide interface. 77, 78 It is believed that the anodization current is mainly related to the movement of ionic species (O 2, OH Al 3+ ) through the oxide layer at the bottom of the pore. 79 Therefore, as a consequence of the pore development, the electric field and the ionic current become concentrated in the barrier layer underneath the major pores. The oxide growth at the metal/oxide interface is due to the migration of oxygen containing ions (O -2 /OH ) from the electrolyte through the oxide layer at the pore bottom, with simultaneous ejection of Al 3+ ions drifted through the oxide layer into the solution at the oxide/electrolyte interface. 79 The fact that Al 3+ ions are lost into the electrolyte and that the atomic density of aluminum in alumina is by a factor of two lower than in metallic aluminum, have suggested a possible origin of forces between neighboring pores that contributes to the mechanical stress which is associated with the expansion during oxide formation at the metal/oxide interface. 76 Since the oxidation takes place at the entire pore bottom simultaneously, the material can only expand in the vertical direction, so that the existing pore walls are pushed upward. This somewhat explain the self-organized 31

PAGE 32

arrangement of neighboring pores in hexagonal arrays by repulsive interactions between the pores during the steady-state film formation after the pore initiation process. Sol-Gel Chemistry at a Glance The sol-gel processing of inorganic ceramic and glass materials has been known for a long time but it has been commonly used since the mid-1900s, and comprehensively studied during the past 30 years. 80, 81 In recent years, the sol-gel chemistry has evolved as a powerful approach for preparing microand nano-structured materials. 82-85 Sol-gel materials display attractive properties such as ease of preparation, low temperature synthesis, multiple structures with large open spaces of tunable porosity, controllable hydrophobicity, capacity for chemical modification of entrapment of various components, mechanical stability, and optical transparency. 86-89 The molecular precursors for preparation of sol-gel materials are essentially alkoxysilanes. The basic steps involved in the sol-process using tetraethoxysilane (TEOS) as a precursor molecule are shown in the Figure 1-6. The initiation step consists in the hydrolysis of the alkoxide precursor which leads to the formation of silanol groups (Si-OH) and ethanol. The silanols can then undergo polymerization reactions with other silanols or with other alkoxysilanes and this condensation reaction leads to the formation of siloxane groups (Si-O-Si) along with water and ethanol as additional reaction products. Figure 1-6.Outline of the steps involved in the sol-gel synthesis. 32

PAGE 33

These reactions lead initially to the formation of molecular clusters, which then grow upon polycondensation to produce a sol. This sol stage is produced when the particles formed are well dispersed in the liquid phase to form colloids. In addition, depending on the experimental conditions; this sol can be transformed in different ways. The usual case is gelation which occurs when the colloidal particles link together to become a three-dimensional network. Upon solvent extraction the gel is transformed into aerogel, 90, 91 while solvent evaporation results in the formation of xerogel. 92, 93 Those are the two most common processes of converting gels into silica. The maximum temperature for both processes can be kept under 100 C. The active silica surface with large specific surface area is of enormous utility to produce a variety of modified silica via covalent bond by using silane reagents. 94 By the insertion of organic functional groups to silica surface there is a partial conversion of surface silanol to a new organofunctional surface that acquires organophilic properties. Thus, the silane chemistry gives a set of properties to the silica surface, which differs considerably from the original matrix. 95 At the silica surface, the structure is composed of both siloxane groups (Si-O-Si) with the oxygen atom on the surface, and one of the three forms of silanol groups (Si-OH). 96 One of the types of silanol groups found in silica surface is the isolated free silanol groups, where the surface silicon atom has three bonds into the bulk structure and the fourth to the OH group. The second type corresponds to the occurrence of two vicinal free silanol groups where the OH groups are bridged by a H-bond. A third type of silanols called geminal silanol consists of two hydroxyl groups attached to one silicon atom. 97 Since the discovery of silanol groups on silica surfaces in 1936 by Kiselev, 98 studies based on theoretical calculations, physical methods and chemical methods have been explored in order to quantify the number of silanol groups on the surface. At present, it is accepted that the 33

PAGE 34

estimated number of OH groups accessible on porous silica surface is between 4 and 5 OH groups per nm 2 with an average surface area of 20.4-21.7 2 per silanol group. 94, 99 -101 In the research reported here, we took advantage of this special attribute of the higher silanol groups on silica surface for further silane functionalization, because when compared with alumina, the number of estimated hydroxyl groups accessible on alumina surface is only 0.1-2.5 OH groups per nm square. 30, 102 In order to prove the benefit of the silica nanotubes in the alumina template, some fluorescence experiments were performed. For these studies, both a templated silica nanotube membrane and alumina membrane were biofunctionalized with anti-Human IgG fluorescent labeled with Alexa 488. This anti-Human IgG was covalently attached via the Schiffs base reaction to the walls of the aldehyde functionalized nanotube membranes. 9 These results are shown in Figure 1-7, where the Alexa 488 fluorescence spectra from alumina membrane and from silica nanotube membrane are compared. The Alexa 488 fluorescence signal from silica nanotube templated membrane was approximately 6 times higher than the analogous alumina membrane. Hence, these results suggest that the deposition of silica nanotubes is essential to provide higher degree of functionalization on the membranes. Figure 1-7.Fluorescence spectra obtained from the biofunctionalized (a) alumina template and (b) templated silica nanotube membranes with anti-Human IgG labeled with Alexa 488. 34

PAGE 35

Silane Chemistry a Versatile Tool for Surface Functionalization The chemical modification of silica surfaces consist in chemical bonding of molecules using silane chemistry, which allow to modify the chemical or physical properties of the surface in a controlled way. Silane functionalization proceeds mainly via the reaction of silanol groups with organofunctional silanes, especially chloroand alkoxy-silanes. 103 The chemistry modification for alkoxysilanes is analogous to that one described for sol-gel production. They are easily prepared through the hydrolysis and condensation of organotrialkoxysilanes. The most common alkoxy groups are methoxy and ethoxy, which give methanol and ethanol respectively, as additional reaction products. The main difference between the trimethoxysilyl and triethoxysilyl monomers, when they are used in silane functionalization, is that reaction with trimethoxysilyl composite is faster. 104 The reactions involved in the alkoxysilane functionalization are illustrated schematically in Figure 1-8. Figure 1.8.Reactions involved in the alkoxysilane functionalization. Scheme obtained at http://www.dowcorning.com/content/publishedlit/SILANE-GUIDE.pdf For the chlorosilanes situation, the reaction with alcohol produces alkoxysilanes and hydrochloric acid. The extent of this reaction can be monitored by measuring the pH. 35

PAGE 36

The decision of using organosilanes with one or more than one hydrolyzable groups for silane functionalization depends on the particular set of properties desired in the new functionalized surface. Thus, when a hydrophobic surface is preferred, organosilanes with one hydrolyzable group are used. 103 If surfaces with higher degree of surface modification are desired, organosilanes with two or three hydrolyzable groups are used. 103 An alternative way of generating one monolayer of the desired silane monomer on silica substrates is by conducting silane chemistry of chlorosilanes in aprotic solvents. 105, 106 In conditions where the aprotic solvent and surface are water free, no alkoxysilane is produced, and no polymerization of the chlorosilane can take place. Therefore, the reaction must be promoted by nucleophilic substitution; where the nucleophile hydroxyl groups in the surface attack the primary halide carbon in the chlorosilane moiety. Surface modification in those cases takes longer time, usually 12-24 hours. Thereby, surface modification via silane functionalization provides a unique opportunity to regulate the interfacial properties of silica substrates while their basic structure and mechanical strength are preserved. Natures Transport across Membranes: An Inspiration for Membrane-Based Separations Biological membranes are vital components of all living systems, forming the external boundary of the living cells and the internal organelles into the cytoplasm. 107 They are constituted mainly of various lipids arranged in a bilayer that imparts a fluid character. Thus, both lipids and proteins are free to move within a bilayer. The proteins embedded in the bilayer and the carbohydrates attached to its surface, facilitate communication and transport across the membrane. In the living membranes, there are two types of membrane proteins which are involved in transport and signaling activities. Those are the peripheral and integral (or intrinsic) 36

PAGE 37

proteins. Peripheral proteins interact through electrostatic forces and hydrogen bonds with the surface of bilayers. While integral proteins are strongly associated with the bilayer itself. These functional characteristics described above, enable membranes to act as essential filters in different living processes that depend on membrane transport systems, such as the acquisition of nutrients, the elimination of waste materials, the transport of metabolites, the generation of concentration gradients vital for nerve impulse transmission and for the normal function of brain, heart, kidneys and other organs. 107 All transport processes in living membranes are mediated by the transport proteins which may function either as channels or carriers. In transport mediated by carrier proteins, only the specific binding molecule is transported. Carrier proteins cycle between conformations that allow molecules to pass to the other side of the membrane. There is never an open conduit all the way through the membrane with carrier proteins. Channels on the other hand, form open pores through the membrane that permits the free diffusion of molecules of appropriated size and charge. These two types of specialized transport systems are illustrated schematically in Figure 1-9. Figure 1-9.Types of specialized transport in living membranes (A) carriers, and (B) channels. In living membranes there are three classes of transport: passive, facilitated, and active. In passive transport, the transported specie moves across the membrane in a thermodynamically favored direction without the assistance of a specific transport system. For neutral species the driving force across the membrane is the chemical potential, while for charged species the 37

PAGE 38

driving force is the electrochemical potential. In addition, the flux of uncharged molecules across the membrane is linearly dependent on concentration. In facilitated transport, the transported species moves also according to a thermodynamically favored direction, but with the collaboration of a specific transport system the carrier. Facilitated transport systems display saturation behavior, where the flux is still dependent on concentration, but at high concentration the dependence is no longer linear. This behavior is due to the saturation of the binding carrier sites as the concentration of the transported species increases. Figure 1-10 compares the dependence on concentration of the flux for the passive and facilitated transport systems. Figure 1-10.Flux concentration dependence for passive and facilitated transport. The third class of transport in living membranes corresponds to the active transport systems. These use energy in order to transport the species against their thermodynamic potential. The most usual energy supplier is the ATP hydrolysis, but light energy and the energy stored in ion gradients are also used. 107 Thus, inspired in the strategies that biological membranes use to accomplish transport, scientists have explored the use of analog synthetic membranes with specific functionalization to selectively conduce transport. Likewise, these systems function on the base of facilitated 38

PAGE 39

transport concept. This facilitated transport synthetic functionalized-membranes integrate a mediator in the membrane the carrier. 9, 10 This carrier usually interacts chemically with one of the components in the feed to be transported across the membrane in a facilitated mode. Therefore, in the same way that living membranes facilitated transport, the flux for this preferred permeant is enhanced when is compared with the passive transport situation (Figure 1-10). As in the case of facilitated transport in living membranes, the facilitated transport in synthetic facilitated-membrane systems also displays saturation behavior. The basic mechanism for this enhanced transport is ascribed to the sequential binding and unbinding events of the facilitated specie with the immobilized carrier in the membrane. The concentration gradients of facilitated process are shown schematically in Figure 1-11, in which the reaction of the facilitated specie (F) with the immobilized carrier (C) in the membranes is given by Equation 1-1. CFCF (1-1) The equilibrium constant for this reaction is characterized by Equation 1-2. ][][][CFCFK (1-2) The terms in the square brackets represents the molar concentration of the chemical species involve in the reaction. Figure 1-11.Concentration gradients formed across the membrane during passive and facilitated transport. 39

PAGE 40

Thus, as can be seen from Figure 1-11, in the facilitated transport membranes, the presence of immobilized carriers that interact with the permeate molecules creates larger concentration gradients downstream the membrane length. Besides, the selectivity acquired with facilitated transport systems is an extra important aspect compared with the nonselective characteristic of the passive transport. Therefore, membranes with higher fluxes can be achieved by creating membranes with larger concentration gradients using immobilized carriers. This effect is also supported by Ficks first law of diffusion: xCDJ (1-3) Since according to Ficks first law, the flux (J) of a permeate molecule across a membrane is directly proportional to the concentration gradient across the membrane. The term C/x in Equation 1-3 represents the concentration gradient of the permeate specie and it is the driving force that leads to molecular movement. 108 The term D is the diffusion coefficient and it is a measure of the mobility of the individual molecules across the membrane. The negative sign indicates that the direction of flow is down the concentration gradient. The above description clearly demonstrates that carrier-facilitated transport membranes promote higher concentration gradients inside the membrane that enhance the flux of the molecule that interact selectively and reversely with the carrier. In simplest terms, the selective binding interaction enhances the local concentration gradient, and thus the flux, of the permeate molecule across the membrane. 9, 10, 109 Systems using synthetic membranes based on the concept of facilitated transport have been the subject of numerous investigations. 9, 10, 109 119 Thus, facilitated transport membrane systems 40

PAGE 41

using different type of carriers such as chemical complexing agents, 111 113 metalloproteins, 114, 115 crown ethers, 116, 117 antibodies, 9 DNA, 10 and others 118, 119 have been already demonstrated. Fundamentals about Proteins Proteins constitute the most important material that composes all living organisms. 107 They are part without exception of each cell. A brief review of the fundamentals of biological production of proteins and protein structure is described next. Proteins consist of a linear sequence of aminoacids, and their order determines the primary structure of the protein. Primary structure of a protein is encoded in the base pair sequence of DNA encoding the gene for the protein. Thus, the four different bases of nucleic acid sequences (adenine (A), cytosine (C), guanine (G), and thymine (T)) are translated into the 20 aminoacids of proteins by a genetic code and grouped into triplets of bases units called codons to code each particular aminoacid. The start codon of the gene is AUG which encodes methionine and the three stop codons are UAA, UAG, or UGA. The genetic expression is initiated when the RNA polymerase binds the start codon and transcripts a single-stranded copy of the DNA into the messenger RNA molecule (mRNA), this process takes place until the stop codon is reached. This mRNA molecule that carries now the genetic code of the DNA is then fed into the ribosome. Likewise, chemical polymerization processes, the biological synthesis of protein has three phases: initiation, elongation, and termination. In initiation, the mRNA binds in a particular complex to the transfer RNA (tRNA), which serves as a carrier of aminoacids to hybridize with the mRNA. During the elongation, the ribosome moves along the mRNA via codon-directed association. The ribosome covalently attaches the new amino acid to the end of the growing peptide chain and so forth the process continues. The termination of the peptide synthesis is activated by the appearance of a stop codon (UAA, UAG, or UGA). 41

PAGE 42

The ribosome produces an unfolded linear chain of denatured peptides. Then, the chain of aminoacids folds to form structures such as alpha-helices and beta-sheets, which are the elements of the protein secondary structure. The various secondary structures then fold into a compact tridimensional structure, which correspond to the tertiary structure of a protein. This tertiary structure, in many cases contains the functional form of the protein. In some cases, the combination of two or more subunits of tertiary structure peptides may be required for the functional form of the protein; this assembly is called the quaternary structure. Antibodies Antibodies, also called immunoglobulins, are proteins produced by the immune system for the defense of the organism against foreign species. 120 They exhibit very specific binding capacity for specific molecules. Immunoglobulins are divided in five classes: IgG IgM, IgA, IgD, and IgE. 120 They vary in size, charge, aminoacid composition and carbohydrate content. Immunoglobulins are made of two identical heavy and two identical light chains of aminoacids, forming a characteristic Y-shape (Figure 1-12). Figure 1-12.Basic structure of the immunoglobulin molecule. 42

PAGE 43

The molecular weight of the heavy chains varies with the class of antibody. They are called gamma, mu, alpha, delta and epsilon for IgG, IgM, IgA, IgD, and IgE, respectively. Based on small polypeptide structural differences, there are two types of light chains classified as kappa and lambda. This means that the polypeptides chains of immunoglobulins are encoded by three different clusters of genes. One cluster encodes the heavy chains for all the classes and subclasses of antibodies, a second one encode the kappa light chains and a third one the lambda light chains. The four glycoprotein chains are connected to one another by disulfide (S-S) bonds and noncovalent interactions. The variations in the number of disulfides bonds and the length of the hinge region determine the subclasses of antibodies. This hinge region is a flexible structure composed of aminoacids that enables the antibody to bind a couple of epitopes separated apart on an antigen. Immunoglobulins are bifunctional molecules. They have one region involved with the antigen binding which is called the Fab portions of the antibody, and a second region called the Fc portion which is involved in several biological activities. This Fc region is also used as a binding site for manipulating the orientation of the antibody during many immunochemical procedures, due to its capability to bind proteins G and A. 121 The Fab portions have a well-defined tridimensional shape specifically designed to bind the antigen, by a lock-and-key mechanism type. The binding between antigen-antibody entails the formation of multiple intermolecular noncovalent forces in the pocket of the binding site as is schematically shown in Figure 1-13. There must be appropriate atomic groups on opposite sites of the antigen and antibody to facilitate the binding between both proteins. Besides, the shape of the combining site must fit the epitope in order to hold the simultaneous formation of noncovalent bonds. Thus, the formation 43

PAGE 44

of hydrogen bonds results in hydrogen bridges between appropriated atoms. In the same way, the attractions between the opposite charged groups in both sides of the proteins contribute with the electrostatic forces, while the Van der Waals forces are created by the interaction between electrons clouds. The hydrophobic forces on the other hand, which may contribute up to half of the total strength of the antigen-antibody bond, rely in the association of non-polar hydrophobic groups, minimizing the contact with water molecules. All these attractive forces are weak if they are considered individually and compared with covalent bonds. However, if they are put together they become a large component of the total binding energy. Figure 1-13.Intermolecular attractive forces involves in the antibody-antigen binding. Scheme obtained from www.fleshandbones.com/readingroom/pdf/291.pdf Recombinant proteins Proteins produced by in vitro transcription and translation can be expressed in organisms such as Escherichia coli (E. coli). Since, the genome and structure of the E. coli have been long studied and well understood by biologists and geneticists. One of the main characteristics that make the E. coli as a very powerful and versatile system for genetic engineering and protein expression is its ability to rapidly proliferate; E. coli cells can be doubled in number under optimal conditions every five minutes. 122, 123 44

PAGE 45

Recombinant proteins are produced from clone DNA (cDNA) sequences which usually encode an enzyme or protein with a specific known function. Therefore, the process of genetic engineering begins with the isolation and extraction of the gene of interest from the clone DNA. This corresponds to the isolated DNA that is complementary to the mRNA that encodes the protein of interest. The next step is inserting the gene of interest by cloning techniques into a vector. The vector is just a small piece of DNA where foreign fragments of DNA may be inserted. Thus, a vector functions like a "molecular carrier", which will carry fragments of the gene of interest into a host cell. Those vectors are usually derived from plasmids, which are small, circular, double-stranded DNA molecules occurring naturally in the cytoplasm of bacteria. The vector thus incorporates an origin of DNA replication for the gene of interest into a host cell like E. coli. 122, 123 The production of the protein starts when these special engineered vectors are inserted into E. coli where the recombinant protein of interest is expressed. Because this E. coli duplicates rapidly, large amounts of E. coli cells containing the foreign gene can be grown in a matter of short time. In order to purify the expressed protein from the complex cell culture, it is a common practice to tag on the vector, a sequence of DNA that introduces a specific functionality into the expressed peptide to be used during protein purification. This tag or purification tool is placed in the vector to create a small extension of aminoacids added to the N-terminal of the target protein. This aminoacid sequence is usually designed to be a point of attack for a specific protease during purification. Thus, after the recombinant protein is expressed and extracted from E. coli, the N-terminal extension can be used to purify the protein and then cleavage it with a specific protease to generate an almost natural N-terminus sequence of the target peptide. 122, 123 45

PAGE 46

Frequently, the introduction of a metal binding site in the expressed recombinant peptide is placed for further purification. It has been demonstrated that an aminoacid sequence consisting of 6 or more histidine residues in a row operate as a metal binding site for recombinant proteins. 124, 125 This hexa-His sequence is called a His-Tag. By using commercially available vectors, the His-Tag sequence can be placed on the N-terminal of a target protein. This expressed His-Tag recombinant protein can be simply purified by using immobilized-metal affinity chromatography (IMAC). 126, 127 Synopsis and Analytical Visualization of the Instrumental Techniques Fluorescence spectroscopy The transport performance in membrane-based separation is normally investigated by permeation experiments of the systems in buffered-aqueous solutions of the permeant solutes to separate. One of the analytical methods used in this investigation to characterize the fluxes of the transported proteins across the membranes is fluorescence spectroscopy. This was achieved through the use of simple photophysical probes such as fluorescent-labeled antibodies. The use of fluorescence techniques for the study of proteins in solution is widespread. Fluorescence spectroscopy is a very sensitive technique, making it possible to perform experiments on protein solutions at very low concentration, minimizing the amount of molecules required for the analysis and eliminating problems associated with aggregation and low solubility. 128 130 Fluorescence is a characteristic process that occurs in certain type of molecules called fluorophores that have the property of absorbing energy and then immediately releasing this energy in the form of light. 131 The fluorescence process basically consists of three stages: excitation, lifetime of the excited state, and fluorescence emission. 46

PAGE 47

In the excitation stage, characteristic photon energy is absorbed for the flourophore molecules creating an excited electronic singlet state (S 2 ) as is illustrated in the Jablonski diagram in Figure 1-14. Figure 1-14.Jablonski diagram of the electronic states during the fluorescence process. This excited state lasts typically for a few nanoseconds (1 to 10 nanoseconds). During the excitation process, part of the excess of energy gained is dissipated by internal conversion to reach the lowest energy mode of a relaxed excited state (S 1 ). The final stage is the fluorescence emission, the electron returns to the ground state energy (S 0 ), where the remaining energy is released in the form of a photon. Due to the loss of energy during the internal conversion in the excited state, the total energy of the fluorescent emitted photons is lower than the absorbed excited photons. Thus, there is a difference between the excitation and emission wavelength which is called the Stokes shift. The ratio of the number of fluorescence photons emitted to the number of photons absorbed is called the fluorescence quantum yield, and it is a measure of the extent for the entire process. The intensity of the fluorescent emitted light, I, is described by the relationship showed in Equation 1-4. 132 bceQII10 (1-4) 47

PAGE 48

Where Q is the quantum yield, I 0 is the incident radiant power, is the molar absorptive coefficient, b is the path length of the cell, and c is the molar concentration of the fluorophore. However, for dilute concentration (bc < 0.05), the Equation 1-4 reduces to the form: bcQII 0 (1-5) Therefore, at dilute concentration of the fluorophore, the relationship between fluorescent signal and concentration should be linear, if Q, I 0 and are keeping constant. 132 Fluorescence quenching The fluorescence quenching phenomena refers to any process that decreases the fluorescence intensity of the fluorophore molecules. 131 The two general types of quenching processes encountered are the collisional and static quenching. The collisional quenching is a process that occurs during the excited state lifetime. In this case, the excited fluorophore experience contact with an atom or molecule (the quencher) that facilitate non-radiant transitions to the ground state. In the case of static quenching, the deactivation of the fluorescence intensity is due to the reaction between fluorophore and quencher that leads to formation of a stable complex in the ground state. The role of fluorescence quenching can be experimentally studied by determining the quenching rate parameters using the Stern-Volmer plots that are drawn in accordance with the Stern-Volmer equation (Equation 1-6). 131 QKIISV10 (1-6) Where [Q] is the concentration of the quencher, I 0 and I are respectively, the fluorescence intensity in the absence and presence of quencher concentration, and K SV is the Stern-Volmer quenching constant. In the cases where the relationship of Io/I with [Q] is linear the quenching process can be either collisional or static. However, those processes can be distinguished experimentally by 48

PAGE 49

carrying out fluorescence measurements at diverse temperatures. Thus, the quenching mechanism is mainly ascribed to a collisional process when at higher temperatures the system present larger values of K SV Conversely, for the cases of quenching via complex formation, higher temperatures show smaller K SV ; in this case K SV is related to the association constant of the complex (K a ). In situations, where the fluorescence ratio as a function of the quencher concentration reveals a positive deviation of the linear Stern-Volmer plot, it corresponds to a combined process of both collisional and static quenching. For such a case a modified Stern-Volmer equation is given by Equation 1-7. ][1][10QKQKIIaSV (1-7) Therefore, the upward curvature in the Stern-Volmer plot is due to the contribution of the square term (second power) of [Q] in the Equation 1-7. FTIR-ATR spectroscopy Fourier Transform Infrared Attenuated Total Reflection spectroscopy is an appropriate technique for the characterization of the surface of materials. This technique combines the capabilities of chemical structure identification offered by the infrared spectroscopy mutually with the fundamental properties of the total internal reflection techniques. The origin of the internal reflection techniques is involved with the fact that radiation propagating in an optical medium having higher refractive index (n 1 ) experiences total internal reflection at the interface of a contiguous sample medium with lower refractive index (n 2 ). This formed propagated wave, which penetrates a small portion of the sample medium, is called evanescent. 133 This phenomenon takes place only when the angle of incidence of the radiant wave exceeds a critical angle, c which can be determined by using Snells law (Equation 1-8). 1221 sin sinnn (1-8) 49

PAGE 50

It is known that in the position of the critical angle, the angle of refraction, 2 is 90. Consequently, the angle of incidence is the critical angle and the Snells equation is reduced to Equation 1-9: 12sinnnc (1-9) The critical angle, c can therefore be found simply by knowing the refractive indexes of the two mediums. It is worth mentioning that the evanescent wave is nontransverse, which allows all its vector components to interact with the dipoles in all directions, thus this wave is a very informative probe of the sample material. 134 During FTIR-ATR analysis, the infrared radiation is passed through an infrared transmitting crystal characterized by having a high refractive index, allowing the radiation to reflect within the crystal several times. Along with each reflection in the transmitting crystal, at the top surface, the evanescent wave penetrates the sample in order to explore the surface material. It should be pointed out that the depth of penetration and the total number of reflections along the crystal can be controlled either by varying the angle of incidence or by selection of crystals. According to Equation 1-10, the penetration depth of the evanescent wave in the sample material can be controlled by the selection of transmitting crystal, since each crystal posses its own refractive index. 2/121221sin2nnnd (1-10) Where is the wavelength of the IR radiation, n 2 and n 1 are the refractive index of the sample and the refractive index of the crystal respectively, and is the angle of incidence. Therefore, crystals with higher refractive index allow lower penetration depths. The depth of 50

PAGE 51

penetration of the evanescent wave, d, is defined as the distance generated at the crystal-sample interface where the intensity of the evanescent decays exponentially to 37% of its original value. 135 X-ray Photoelectron Spectroscopy (XPS) XPS also known as ESCA (Electron Spectroscopy for Chemical Analysis) alike to FTIR-ATR is a surface analysis technique, which defer in the depth of sample analysis and obviously in the incident source of excitation. For example, XPS probes only 5 to 10 nm depth whereas FTIR-ATR allows the detection of chemical species found in the first few microns of the sample material. 136 The XPS analysis is based upon the principles of the photoelectric effect, where the surface material is irradiated with monochromatic X-rays creating photoelectrons that are emitted from the sample surface. These ejected electrons come from the core level electrons of the atoms dispose on the surface material. 137 The emitted energy from the photoelectrons is then analyzed by an electron analyzer spectrometer to determine the binding energy of the electrons. From the binding energy and intensity of a photoelectron signal, it is possible to determine the elemental identity, chemical state, and quantity of an element. It should be pointed out that experimentally the kinetic energy (E k ) of the electrons is the physical parameter measured by the electron analyzer spectrometer and not its binding energy (E b ). However, it is really E b that provides the specific properties of an electron, because E k is not an intrinsic property of the analyzed material since this also depends on the photon energy of the X-rays employed during the analysis. The Equation 1-11 includes all the parameters involved in the XPS analysis and combines the relationship between E b and E k WEhEkb (1-11) 51

PAGE 52

Where h is the photon energy and W is the spectrometer work function. Thus, the binding energy of the electron is simply measured by the other three quantifiable parameters included in Equation 1-11. The XPS spectrum is composed only for the peaks of those electrons that escape the surface without energy loss. The electrons that suffer inelastic scattering and undergo energy loss contribute to the background of the spectrum. 137 As any other energy excitation process, the excess of energy in the excited state must be released or transformed in some way of energy. Thus in XPS, the ionized atom formed after the emission of a photoelectron must have some type of relaxation mechanism. This might be accomplished by X-ray fluorescence emission or by ejection of an Auger electron. This relaxation mechanism explains why during the XPS process, peaks for Auger electrons are also observed. Scanning Electron Microscopy (SEM) SEM is a microscope that uses electrons rather than light to form an image. It is created by focusing a high energy beam of electrons onto the surface. Once the focused electrons hit the surface, secondary and backscattered electrons are ejected from the specimen. 138 Those are collected in specific detectors and converted into a signal that is sent to a TV scanner where the image is processed and viewed as a digital image. It is worth mentioning that secondary electrons are low energy electrons (< 50 eV) that are originated within a few nanometers depth of the specimen. Whereas backscattered electrons are those electrons generated after incident electrons are scattered within the specimen volume and they are reflected or backscattered out of the specimen interaction volume keeping high energy. Backscattered electrons are useful to distinguish contrast between areas with different chemical composition because the image created with them depends on the mean atomic number of the 52

PAGE 53

substances that constitute the specimen. 138 Usually the SEM instruments are equipped with a set of two detectors to allow separate observations of topography (TOPO) image and composition (COMPO) image. Impedance spectroscopy Impedance spectroscopy is an effective method to probe interface properties of surface-modified electrodes. Generally there are two categories in impedance measurements: non-faradaic and faradaic impedance. Non-faradaic impedance is performed in the absence of any redox probe, which has been a method used for measuring changes in the electrical properties of synthetic membranes. 139, 140 Instead, electrochemical impedance is a faradaic impedance technique that is performed in the presence of a redox probe. Impedance is usually measured applying an AC potential to an electrochemical cell or electrical circuit and measuring the current going trough these. 141, 142 Impedance is expressed as a complex number. The complex impedance can be represented as the sum of real, Z re (), and imaginary, Z im (), components originated mainly from the resistance and capacitance of the system, respectively. The general electronic circuit, includes the ohmic resistance of the electrolyte, R s the Warburg impedance, Z W resulting from the diffusion of ions from the bulk electrolyte to the electrode, the double layer capacitance, C dl and the interfacial electron transfer resistance, R et present when a redox probe exist in the electrolyte. In the cases of non-faradic impedance, as measuring the electrical properties of synthetic membranes, this resistance can be associated with the changes in resistance across the membrane, R m The components R s and R W represent bulk properties of the electrolyte solution and diffusion features of the probe in solution. The other two components in the circuit, C dl and R et represent interfacial properties of the electrode and they are affected by immobilization of bio-organic material onto electrode surfaces. 141 Thus, analysis of C dl and R et can give important 53

PAGE 54

information about the extent of changes of the surfaces properties resulting from coupling of biomaterial to the studied interface. 142 A typical impedance data is represented in the form of Nyquist plot. The shape of this plot includes a semicircle region lying on the Z re -axis followed by an incline line. The semicircle region observed at higher frequencies corresponds to the kinetic limited process. Whereas the linear part characteristic of the lower frequency range represents diffusional limited process. In the Nyquist plot R s can be found by reading the real axis value at the high frequency intercept. While the sum of R m and R s can be found by fitting the semicircle until the real axis value at the lower frequency intercept. Nyquist plot present one shortcoming, when looking at any point on the plot, it cannot tell what frequency was used to record that point. There is another popular representation of the impedance data called the Bode plot. For this, the impedance is plotted with log frequency on the x-axis, and both the absolute value of impedance and the phase-shift on the y-axis. 141 54

PAGE 55

CHAPTER 2 TRANSPORT AND FUNCTIONAL BEHAVIOR OF ANTIBODY-FUNCTIONALIZED NANOTUBE MEMBRANES Introduction Today separation and analyses of proteins present important challenges for modern bioanalytical chemistry. It is clear that the most important frontier in the biomedical sciences and in the disease diagnostic is proteomics. The fact that mainly proteins execute and control cellular phenomena opens up the use of proteins as important molecular targets for drugs, and biomarkers for diseases. 143 Proteomics aims to understand complex biological systems by analyzing protein expression, protein function, protein modifications, and protein interactions. 144 Within the proteome-analysis strategies, separation techniques play an important role. 145 The 2D gel electrophoresis in combination with mass spectrometry is the most widely used separation/identification technology in proteomics. 145 147 However, there are a number of approaches to separate proteins based on their biochemical and biophysical features such as molecular weight, hydrophobicity, surface charge, isoelectric point, etc. that have also been developed and employed. 148 151 In addition to these chromatographic methods, synthetic membranes have been used for protein separations. 3 7, 152 164 Nevertheless, the membranes used to date typically separate proteins on the basis of size and charge, and these rudimentary forms of selectivity are insufficient to allow membranes to compete with chromatographic methods for protein separations. The use of affinity chromatography, however, has the advantage of offering a process that is specific for the protein of interest, overcoming the lack of selectivity and simplicity that many of the common methods of analysis undergo. We report here systems that exploit the advantages of affinity chromatography in a nanometer-scale membrane-based platform. In that search for new advanced membrane materials, our group has conducted fundamental studies in the 55

PAGE 56

development of functionalized-membranes with a selective transport function executed through a specific molecule carrier to modulate the transport during membrane-based separations. 9, 10 The carrier is a mediator that selectively binds to the specific chemical solute, the target compound to be separated from a sample solution. Depending on its kinetic interactions with the target compound, the carrier allows enhanced or retarded transport across the membrane. Here we describe the use of antibody-functionalized nanotube membranes for the separation of proteins. Since antibodies that selectively bind proteins are well known, 165, 166 it occurred to us that this antibody-based approach might provide a general route for preparing highly selective membranes for protein separations. To test this hypothesis, membranes containing silica nanotubes were prepared using the template synthesis method. 9, 25, 85 In the inside walls of these nanotubes, purified mouse immunoglobulin IgG (M-IgG) or human immunoglobulin IgG (H-IgG) were covalently attached, using an aldehyde-containing silane reagent (Figure 2-1). 9 Figure 2-1.Immobilization of antibodies on the template synthesized silica nanotube membranes. To study the functional behavior of the IgG-functionalized membranes, competitive transport experiments were done, using antibodies of anti-mouse IgG (anti-M-IgG) and anti-human IgG (anti-H-IgG) fluorescent labeled with Alexa 488 or Alexa 594. The studies of transport properties for these fluorescent labeled antibodies demonstrated that the nature of immobilized antibody changes the functional behavior of the functionalized membranes and the 56

PAGE 57

transport properties of the permeant proteins across them. Additional characterization of these antibody-based membranes showed a Langmuirian shape plot characteristic of a typical facilitated transport membrane when the effect of feed solution concentration on the fluxes of these fluorescent-labeled antibodies was investigated. Experimental Materials Commercially available nanopore alumina membranes (60 m thick, nominal pore diameter 100 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999% (TEOS) (Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes. The purified mouse immunoglobulin (M-IgG) immobilized within the silica nanotube membranes was obtained from Sigma Aldrich. Membranes modified with purified human immunoglobulin (H-IgG) obtained from Sigma Aldrich were also prepared. A recombinant protein G obtained from Sigma Aldrich was used for well oriented antibody-immobilization membranes. A trimethoxysilyl poly(ethylene glycol) (PEG-Si, MW 460-590 Da) obtained from Gelest (Morrisville, PA) was used to modify the silica nanotubes control membranes. A trimethoxysilyl propyl aldehyde (United Chemical Technologies) was used to attach the M-IgG or H-IgG to the nanotubes. Two whole antibodies raised in goat against M-IgG were used to investigate the transport properties of the M-IgG-modified membranes. These anti-M-IgGs were obtained from Molecular Probes (Eugene, OR) conjugated with either the dye Alexa 488 or the dye Alexa 594. The transport properties of two control IgGs, raised in goat against human IgG (H-IgG), were also investigated. These anti-H-IgGs were also obtained from Molecular Probes conjugated with either Alexa 488 or Alexa 594. It is worthy to mention that these four conjugated antibodies were also used to investigate the transport properties of the H-IgG 57

PAGE 58

modified membranes. In this case the two control IgGs were the anti-M-IgG. A F(ab) 2 fragment antibody raised in goat against human IgG (F(ab) 2 -anti-H-IgG) and labeled with FITC fluorophore was also used for transport experiments. This F(ab) 2 -anti-H-IgG antibody was obtained from Molecular Probes. SuperBlock blocking buffer solution in phosphate-buffered saline was obtained from Pierce. SuperBlock solution contains a proprietary protein for blocking binding sites for non-specific protein adsorption. Tween 20 was obtained from Sigma Aldrich. All other chemicals were of reagent grade and used as received. Purified water, obtained by passing house-distilled water through a Barnstead, E-pure water-purification system, was used to prepare all solutions. Preparation of the Silica Nanotube Membranes Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-based template-synthesis method. 9, 25, 85 Briefly, a sol-gel precursor solution was prepared by mixing absolute ethanol, TEOS and 1 M HCl (50:5:1 by volume), and this solution was allowed to hydrolyze for 30 minutes. The alumina membrane was then immersed into this solution for two minutes under sonication. The membrane was then removed and the surfaces swabbed with ethanol to remove precursor from the alumina surface. The membrane was then air dry for ten minutes at room temperature and cured in an oven for 12 hours at 150 C. This yields silica nanotubes, with wall thickness of ~3 nm, embedded within the pores of the alumina membrane. 9 Antibody Immobilization The trimethoxy aldehyde silane was first attached to the inside walls of the silica nanotubes to introduce aldehyde terminal groups onto the pores of the membrane. 9 The membrane was then immersed into a solution that was 1.0 mg per mL in either M-IgG or H-IgG dissolved in 10 mM phosphate-buffered saline with pH = 7.4 that also contained 150 mM sodium chloride and 5 mM sodium azide. The membrane was incubated overnight at 4 C, 58

PAGE 59

which allowed free amino sites on the antibody to attach via the Schiffs base reaction to the nanotube-immobilized aldehyde groups 9 After immobilization, the membranes were rinsed with buffer and then incubated for 5 hours in the SuperBlock solution that also contained 0.05% Tween 20. Transport Experiments The M-IgGor H-IgG-modified membranes were sandwiched between two pieces of Scotch tape that had 0.31 cm 2 area holes punched through them. These holes defined the area of membrane exposed to the contacting solution phases. This membrane assembly was mounted between the two halves of a U-tube permeation cell, with half-cell volumes of ~10 mL. 4, 6, 10, 22, 23 Figure 2-2.Flow-through fluorescence system used to measure the flux of the permeating proteins. The transport properties of three types of membranes were investigated membranes containing immobilized M-IgG, membranes containing immobilized H-IgG, and control membranes with attached PEG-Si. The permeating proteins, whose transport properties were investigated, were the fluorescently-labeled anti-M-IgG and anti-H-IgG. The feed solution for all experiments was equi-molar in these two proteins, dissolved in 10 mM phosphate-buffered saline pH 7.4 that also contained 150 mM sodium chloride and 5 mM sodium azide (buffer used for all studies). The receiver solution on the other side of the membrane was identical except it initially contained no protein. 59

PAGE 60

The fluorescence intensity of the receiver solution was periodically measured using two flow-through fluorescence detectors connected in series (Figure 2-2). This system consisted of an L-7485 fluorescence detector (Hitachi) and an RF-2000 fluorescence detector (Dionex). The flow stream was delivered using an L-7100 HPLC pump (Hitachi). We discovered that if the receiver solution was sent continuously through the fluorescence detectors during the entire course of a transport experiment, significant photo bleaching of the Alexa dyes occurred. To prevent this, we turned the pump on to deliver receiver solution to the detectors for only 3 minutes per hour of transport experiment. That is, we used a duty cycle of 57 minutes with pump off followed by 3 minutes with pump on (flow rate = 3 mL per min) to collect fluorescence data. The whole system was controlled using Chromeleon 6.50 chromatography management software (Dionex) through a UCI-100 universal chromatography interface (Dionex). The Hitachi detector was used exclusively to detect the Alexa-594-labeled anti-M-IgG and the Dionex detector was used for the Alexa-488-labeled anti-H-IgG. Alexa 488 was excited at 495 nm, and the emission was detected at 515 nm. Alexa 594 was excited at 590 nm and detected at 617 nm. Calibration curves were obtained with the same flow through system on solutions containing known concentrations of both of the flourophore-labeled antibodies. These calibration curves were used to obtain the concentrations of the transport proteins in the receiver solution. Competitive Binding Assays Our experimental design assumes that the permeating protein anti-M-IgG binds strongly to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to this carrier. Likewise we assume that anti-H-IgG binds strongly to the carrier H-IgG and that anti-M-IgG binds weakly to this carrier. We proved these assumptions to be true by conducting competitive 60

PAGE 61

binding assays. This was accomplished by equilibrating H-IgGand M-IgG-modified membranes to solutions that were 21.2 nM in both anti-M-IgG and anti-H-IgG, and then measuring the fluorescence intensities from the membranes. Calibration data were used to convert these fluorescence intensities to pmoles of anti-M-IgG and anti-H-IgG bound to the membrane. The standards used to obtain the calibration curves were 0.31 cm 2 membrane samples to which were applied known quantities of the desired protein. The fluorescence intensities from these standards were measured using a fluorescence microscope that was equipped with both a fluorescence detector and a CCD camera. This system combines an Axioplan 2 imaging microscope (Zeiss) with a J&M-PMT photometry system detector (SpectrAlliance), for measuring the fluorescence intensity using a BP450-490 filter for excitation and a LP515 filter for green emission of light, and a BP546/12 filter for excitation and a LP590 filter for red emission of light. Determination of Static Protein Binding Capacity and Binding Constants through Adsorption Isotherms The adsorption isotherms were carried out with M-IgG modified, H-IgG modified, and unmodified control membranes at 23 C in stirred U-tube permeation cells using the same buffer conditions used for the permeation studies, via solutions containing single concentrations of anti-M-IgG or anti-H-IgG fluorescent labeled with Alexa 488. In the case of M-IgG-modified but immobilized via protein G, the adsorption isotherms were carried out at three different temperatures: 23, 32, and 42 C. The fluorescence intensity of all the membranes was measured under the fluorescence microscope. Calibration curves were used to obtain the concentration of the Alexa 488-labeled proteins adsorbed to the membrane sample from the measured fluorescence intensity. The data were analyzed according to the monolayer Langmuir model. 167 61

PAGE 62

Effect of Equilibration Time on Flux and Selectivity via Dynamic Binding Capacity At the start of a transport experiment, all of the M-IgG or H-IgG binding sites are empty and these sites fill with anti-H-IgG or anti-M-IgG during the course of the experiment. We were concerned that binding equilibrium might not be achieved during the course of a transport experiment and that our results reflected this lack of equilibration. To explore this issue we investigated the flux and selectivity as a function of equilibration time. This was accomplished using seven membranes modified with M-IgG, which were exposed to a solution that was 21.2 nM in both of the permeating proteins, anti-H-IgG and anti-M-IgG. After one day of exposure, the first membrane was rinsed with buffer and a transport experiment was run (for 24 hours) using a feed solution that was 21.2 nM in both of the permeating proteins. After the second day of exposure, the same procedure was repeated for the second membrane, and so on, until the seventh membrane was investigated after seven days of exposure. Thus, after the binding sites available in the membrane are filled, the dynamic binding capacity is reached to conduct the transport experiments. 168 Results and Discussion The nanotube membranes investigated here contain a bound carrier, either M-IgG or H-IgG. The transport selectivity coefficient () across of the membranes was evaluated by comparing the flux of a permeating species that weakly binds to these carriers to the flux of a permeating species that binds stronger. In the case of the M-IgG-modified membrane, the strongly and weakly binding species are anti-M-IgG and anti-H-IgG, respectively. In the case of the H-IgG-modified membrane, the strongly and weakly binding species are anti-H-IgG and anti-M-IgG, respectively. These studies begin, however, with electron microscopy characterization of the alumina template and investigations of the fluxes of the two permeating proteins, anti-H-IgG and anti-M-IgG, across PEG-Si-modified control nanotube membranes. 62

PAGE 63

Electron Microscopy and Calibration Data Figure 2-3 shows electron micrographs of the surface and cross section of the alumina template membrane and of silica nanotubes deposited within the pores of such membranes. The alumina membrane has two different faces the face that was in contact with the underlying aluminum during anodization (branching side), and the face that was in contact with the electrolyte during anodization. The shape of the pores has a polygonal aspect on the branching side, while in the solution anodization side is rounded. Note that in the transport experiments the nanotubes were left embedded within the pores of the membrane, but they were removed here so that they could be imaged. The tubes have an outside diameter of 185 nm, as would be expected given the pore diameter, 188 nm in the solution anodization side and 113 nm for the branch side. The diameters for the alumina template were determined by electron microscopy based on the analysis of twenty pores in four different membranes. It is well known that the pores in these alumina templates branch at one surface due to the voltage-reduction method used to remove the membrane from the underlying Al surface. 44 Figure 2-3.Electron micrographs of (A) branch side, (B) solution anodization side, and (C) cross-section of the alumina template membranes, and (D) silica nanotubes after removal from the membrane. Evidence for this branching can be seen at the ends of some of the silica nanotubes (Figure 2-3D). The length of the pores with larger size cover most of the 60 m thickness of the alumina template, while the smaller pores cover less than 200 nm of the membrane. 63

PAGE 64

Figure 2-4 shows typical calibration data and a typical calibration curve. The calibration data (Figure 2-4A) were obtained by pumping the indicated concentrations of Alexa 488-labeled anti-H-IgG through the Dionex detector (Figure 2-2). The calibration curve obtained from these data (Figure 2-4B) showed excellent linearity (correlation coefficient = 0.996) over the range from 20 pM to 1280 pM. Analogous data were obtained for Alexa 594-labeled anti-M-IgG using the Hitachi detector. Figure2-4.Typical plots of (A) calibration data, and (B) calibration curve for Alexa-488 labeled anti-H-IgG. Transport Properties of the anti-IgGs through PEG-Si-modified Control Membranes The transport data were processed via flux plots 9, 10 moles of permeating protein transported across the membrane vs. time. Flux plots for feed solutions containing equi-molar concentrations of anti-M-IgG and anti-H-IgG across PEG-Si-modified control membranes are shown in Figure 2-5; data for two different feed concentrations are shown. These plots were linear (correlation coefficients >0.998) indicating that Ficks first law is obeyed and that over the time window investigated, the feed solution was not appreciably depleted of the transporting proteins. Figure 2-5 shows clearly that the fluxes of anti-M-IgG and anti-H-IgG are the same in the PEG-Si-modified control membranes. This means that the diffusion coefficients for these two 64

PAGE 65

proteins are the same, which in turn means that the proteins are identical in size, as would be expected for two IgG molecules. Therefore, the transport properties across these control membranes are the same for these two antibodies. These points are further reinforced by the concentration dependence of the flux. The concentration used for the lower two curves was eight-times lower than the concentration used for the upper two curves. Correspondingly, the flux at the high concentration (1.45 pmoles h -1 cm -2 ) is eight times higher than the flux at the low concentration (0.18 pmoles h -1 cm -2 ). After the permeation experiments, the membranes were observed under the fluorescence microscope, but no fluorescence was detected. This indicates PEG-Si-modification suppresses non-specific protein adsorption to the silica nanotubes. Figure 2-5. Flux plots for anti-M-IgG () and anti-H-IgG () through PEG-Si-modified control membranes. The feed concentration was 2.65 nM (lower curves) and 21.2 nM (upper curves). Competitive Binding Studies As noted above, our experimental design assumes that the permeating protein anti-M-IgG binds strongly to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to this carrier. Likewise we assume that anti-H-IgG binds strongly to the carrier H-IgG and that 65

PAGE 66

anti-M-IgG binds weakly to this carrier. We proved these assumptions to be true by conducting competitive binding assays. Figure 2-6A shows the fluorescence spectra of the permeating proteins bound to a M-IgG-modified membrane after equilibration with a solution that was 21.2 nM in both anti-M-IgG and anti-H-IgG. The membrane shows strong fluorescence from anti-M-IgG and very weak fluorescence from anti-H-IgG. Calibration data allowed us to convert these fluorescence intensities to moles of each protein bound to the surface. In agreement with our design paradigm, 10-times more anti-M-IgG was bound to the M-IgG-modified membrane than anti-H-IgG. The small amount of anti-H-IgG bound is due to antigen cross reactivity, as a result of the polyclonal characteristic of the antibodies. 169 172 Similarly, 14-times more anti-H-IgG was bound to the H-IgG-modified membrane than anti-M-IgG (Figure 2-6B). b a b a Figure 2-6.Fluorescence spectra of (A) M-IgGand (B) H-IgG-modified membranes after equilibration with a solution that was 21.2 nM in both (a) anti-H-IgG, and (b) anti-M-IgG. These experiments also confirmed the success of the chemistry immobilization and the capability of the antibodies immobilized in the nanotube membranes to preserve their bioactive properties on the porous surface. 66

PAGE 67

Functional Behavior of the M-IgGand H-IgG-modified Membranes In these studies, we have been exploring the transport properties, and separation of different anti-IgGs through antibody-modified nanotube membranes, and how the type of immobilized antibody alters the functional behavior of these functionalized membranes. Although, porous membranes have been used for size-based protein separation with great success, separation of proteins with similar size is more difficult. 7 This process is even more challenging, when the separation is between proteins that have practically the same molecular structure and size as in the case of different anti-IgGs. Flux plots for the transport of anti-M-IgG and anti-H-IgG across the M-IgG-modified nanotube membrane (Figure 2-7A) are linear, as are the flux plots for these permeating proteins for a H-IgG-modified membrane (Figure 2-7B). The fluxes obtained from the slopes of these plots are shown in Table 2-1. As mentioned above, the selectivity coefficient () across of the membranes was evaluated by comparing the flux of a permeating species that binds weakly to the carrier to the flux of a permeating species that binds more strongly. Table 2-1.Fluxes and selectivity coefficients for anti-M-IgG and anti-H-IgG across the M-IgG and H-IgG-modified membranes IgG-attached to the membrane Flux of anti-M-IgG (pmoles h -1 cm -2 ) Flux of anti-H-IgG (pmoles h -1 cm -2 ) M-IgG 0.60 1.31 2.2 H-IgG 1.33 0.48 2.8 It was observed that for both membranes the flux of the species that binds strongly to the carrier (e.g., anti M-IgG for the M-IgG-modified membrane) is lower than the flux of species that binds weakly to the carrier. This is a clear indication that the transport properties for these 67

PAGE 68

two permeating proteins are different, which in this situation, depends on the functional behavior of the IgG-functionalized nanotube membrane. It is worth mentioning that under steady state conditions observed in all the transport experiments, only an averaged transport rate can be determined because proteins being transported do not enter membrane pores at the same time so that the measured signal is a spatially and temporally averaged one. Figure 2-7.Flux plots for anti-M-IgG () and anti-H-IgG () in (A) M-IgG-modified membrane, and (B) H-IgG-modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins. Observed Functional Behavior is not a Transient Phenomenon One way to interpret the results in Figure 2-7 and Table 2-1 is that they might reflect a transient phenomenon. That is, one could hypothesize that during the permeation experiments summarized in Figure 2-7, the flux of the species that bound strongly to the carrier were suppressed because all the sites for binding were initially empty. According to this hypothesis the flux for the strongly binding species would remain low until all sites were filled and breakthrough occurred. It is difficult to reconcile this hypothesis with the constant flux observed for the strongly binding species over these nearly one-daylong transport experiments (Figure 2-7). One would 68

PAGE 69

anticipate instead, very low rates of transport at short times and higher rates at longer times as the sites fill. Nevertheless, we designed a set of experiments to test this hypothesis (Figure 2-8). We found that the flux of the weakly-binding species (in this case anti-H-IgG) remained higher than that for the strongly binding species (anti-M-IgG) even after 1 week of equilibration with a solution equi-molar in both proteins. Interestingly, the selectivity coefficient () improved slightly over the time course of this experiment. The other important point to glean from Figure 2-8 is that the data indicate that the membrane-bound M-IgG remains bioactive for the weak-long period summarized in the figure. Figure 2-8.Dynamic binding capacity study using M-IgG-modified membranes. The flux and selectivity coefficient vs. days of equilibration are shown for the transport of anti-M-IgG and anti-H-IgG. The feed solution concentration was 21.2 nM in both of the permeating proteins. It was also observed that some of the protein transmission was lost across membranes with more than 4 days of saturation. This effect was ascribed to the fact that of protein bound to the M-IgG-modified membranes decreased the effective pore size for the remaining diffusing protein molecules. Besides, such diminution in flux is not uncommon with protein-transporting membranes and results from gradual fouling of the membrane. 162, 173 176 69

PAGE 70

It was also found that after equilibration the weakly binding protein (anti-H-IgG) diffuse more rapidly than the strongly-binding protein (anti-M-IgG). Therefore, there is a trade-off between selectivity and flux after the membranes are equilibrated. The equilibration reduces the flux for the strongly binding protein with respect to the weakly binding one, but leads at the same time to an increase in selectivity. Thus, it can be concluded that the M-IgG carrier without equilibration in some extend facilitate the transport of anti-M-IgG when is compared with transport across membranes that have been previously equilibrated. Flux vs. Feed Concentration A number of competitive permeation experiments were carried out at different feed concentration, ranging from 2.65 nM to 0.5 M using M-IgG modified membranes (Figure 2-9). Figure 2-9.Flux versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. The feed solution concentrations were equimolar in both of the permeating proteins. Figure 2-9 shows flux-vs.-feed plots for transport of anti-H-IgG and anti-M-IgG across M-IgG-modified membranes. The flux for anti-H-IgG was at all times higher than the flux for anti-M-IgG. These plots show a characteristic Langmuirian shapes, 9, 10, 109 demonstrating that the M-IgG attached in the membrane indeed interacts with the penetrant-antibody, acting as a carrier during the permeation process. A straight line would be expected, if there is not interaction 70

PAGE 71

between carrier and permeating antibodies at the very low feed concentration used during these studies. A key way to prove that transport across a membrane is facilitated is to plot the flux vs. the feed concentration. 9, 10, 109 For facilitated-transport membranes, such plots are linear at low feed concentration but flatten (Langmuirian shape) 9, 10, 109 at high concentrations. Data for both of the permeating proteins show the characteristic Langmuirian shape of facilitated transport. Membranes that contained attached PEG-Si, instead of attached M-IgG showed linear plots as is shown in Figure 2-10. Therefore, the antibody-functionalized membranes showed here exhibit a type of behavior for facilitated-transport membranes during protein separations. Figure 2-10.Typical plot of flux versus feed concentration for anti-M-IgG or anti-H-IgG across PEG-Si modified membrane. The data plotted is an average of the fluxes for anti-M-IgG and anti-H-IgG for each specific feed concentration. Effect of the Pore Size on the Flux and Selectivity Coefficient The influence of the pore size on the flux and selectivity coefficient was explored by using three different sizes of commercially available (Whatman) nanopore alumina membranes having 60 m thickness, and nominal pore diameters respectively of 200, 100, and 20 nm. To study this effect, nanotube membranes of the specified pore size were modified with M-IgG. For the 71

PAGE 72

transport experiments, the feed solution concentration was in all cases 2.65 nM in both of the permeating proteins anti-H-IgG and anti-M-IgG (Figure 2-11A). Figure 2-11.Effect of the pore size on (A) the flux, and (B) selectivity coefficients for M-IgG modified membranes. The feed solution concentrations was 2.65 nM both in anti-H-IgG and anti-M-IgG. At this specific feed concentration (2.65 nM) not appreciable differences for the flux of anti-M-IgG across these M-IgG modified membranes were observed. However, lower fluxes were obtained for anti-H-IgG as the pore size decreases (Figure 2-11A). This reduction on flux is due mainly to the resistance that the transported antibodies experience in the pores that arises from the hydrodynamic interaction of the protein molecules with the wall of the pore, which include far field and near field interactions. 177, 178 Interestingly, the flux for the 200 nm membrane decreases slightly compared with the 100 nm membrane. This reduction in flux was ascribed to the fact that 200 nm pore size membranes are much easier of functionalize than 100 nm one. Therefore, the data suggests that the number of binding carrier sites immobilized in this membrane is possibly higher than the 100 nm, which in turn decreases the effective pore size of the membrane. Additional evidence for this conclusion was obtained from the comparison of fluorescence spectra of the membranes after transport experiments, which showed higher fluorescence intensities for the 200 nm M-IgG modified membrane. 72

PAGE 73

It was observed that the selectivity coefficient () increases as the pore diameter increases (Figure 2-11B). The increase in selectivity with bigger pores was attributed to the availability of much more room to diffusive for the protein that weakly binds the M-IgG carrier, which in this specific case corresponds to anti-H-IgG. In addition, this effect was observed in the case of the 20/200 nm membranes when the transport was carried out placing the feed solution on the 200 nm side of the membrane (Figure 2-11A). Effect of the Immobilized-Antibody Orientation on the Flux This study was accomplished by taking advantage of the specific affinity that protein G has for the Fc region of the IgG molecule. 121, 170, 179, 180 Thus, it was possible to manipulate the orientation of the immobilized antibody on the walls of the nanotubes. Figure 2-12.Flux plots for anti-M-IgG () and anti-H-IgG () in well-oriented M-IgG-modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins. It is known that due to steric hindrance and random orientation of the antibody molecules on the solid-phase surface, the binding activity of antibody immobilized directly on the solid-phase surface is usually less than that of soluble antibody. 179, 180 In order to overcome this problem, protein G was used to modify silica nanotube membrane for antibody immobilization. Thus, the well oriented immobilization of M-IgG was carried out on nanotube membranes 73

PAGE 74

modified with protein G that was previously immobilized via aldehyde silane chemistry on the nanotube walls. The results for the transport experiments of anti-M-IgG and anti-H-IgG across this well-oriented M-IgG modified-membrane are shown in Figure 2-12. The feed solution concentration was 21.2 nM in both of the permeating proteins. The fluxes of anti-H-IgG and anti-M-IgG across this well-oriented M-IgG modified membrane were 0.64 and 0.20 pmoles h -1 cm -2 respectively. As the M-IgG-modified membranes immobilized by direct immobilization on the nanotube surface, this membranes also preferentially transport the protein that binds weakly to the immobilized carrier. The fluxes for anti-H-IgG and anti-M-IgG were respectively two times and almost three times lower compared with those fluxes for membranes modified with M-IgG but where the immobilization was directly attached to the silane aldehyde surfaces in the membrane. The reason for these lower fluxes was attributed to the decrease in the effective pore size in the membrane with the addition of protein-G and the possibility of having all the M-IgG molecules in a well-oriented immobilization which also contributes to the decrease in the effective pore size. Besides, the well-oriented immobilization contributes with a stronger binding affinity, because this helps to preserve the binding activity of the antibody, which at the same time lowers the diffusion coefficients of the permeating proteins. Separation Based on a Simultaneous Combination of Size and Affinity Selectivity In order to prove this concept of a simultaneous combination of sizeand affinity-based separation of proteins, a nanotube membranes templated in a monodisperse pore size alumina of 40 nm crosswise the whole 60 m thickness was used. Scanning electron micrographs of the surface and cross section of this membrane are shown in Figure 2-13. 74

PAGE 75

It should be pointed out that the 40 nm pore size of this alumina template is about 3 times the hydrodynamic diameter for the IgG whole molecule (14 nm). 3 Therefore, after M-IgG immobilization the effective pore size of the nanotube membranes is reduced to a magnitude that allows a closer interaction of the permeating proteins both with the M-IgG immobilized carrier and the wall of the pores. Figure 2-13.Scanning electron micrograph of (A) the surface, and (B) cross section of a highly-ordered nanoporous alumina membrane obtained at 30 V in 5% oxalic acid. The next set of experiments demonstrated the capacity of these antibody-modified membranes to be used with a combination of both size and affinity selectivity. First, the transport of anti-H-IgG and anti-M-IgG across a control PEG-Si modified membrane and a well-oriented M-IgG modified membrane are compared (Figure 2-14). A B Figure 2-14.Flux plots for anti-M-IgG () and anti-H-IgG () in (A) PEG-Si-modified membrane, and (B) well-oriented M-IgG -modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins. 75

PAGE 76

The permeation process through membranes can be divided into its transient and steady state components. 181, 182 The transient component can be characterized by the time lag. This time lag observed phenomenon correspond to the time at which the permeating specie enters the membrane to the time that the flow rate of diffusing species into the close volume reaches a steady state of permeation. 181, 182 Figure 2-14A shows the transport of anti-M-IgG and anti-H-IgG across a PEG-Si modified control membrane. It is clearly observed a time lag event for both of the permeating proteins. This effect was even more noticeable for the antibody-modified membrane as can be seen in Figure 2-14B. By just comparing the time in which anti-M-IgG reaches a steady state of permeation across PEG-Si control and M-IgG-modified membranes, ten and twelve hour respectively, we can see that antibody immobilization has a significant effect in the functional behavior of the membrane. Therefore, by taking advantage of this time lag effect, transport of antibodies with different sizes were carried out in order to prove the concept of a simultaneous combination of sizeand affinity-based separation of proteins (Figure 2-15). Figure 2-15. Flux plots for anti-M-IgG () and F(ab) 2 -anti-H-IgG () in well-oriented M-IgG -modified membrane. The feed solution concentration was 21.2 nM in both of the permeating proteins. 76

PAGE 77

The difference between the experiment in Figure 2-14B and Figure 2-15 is that for the experiments in Figure 2-15 a F(ab) 2 -anti-H-IgG labeled with FITC was transported instead of the anti-H-IgG whole antibody labeled with Alexa 488. It can be seen from Figure 2-15 that the time lag is less evident for transport of the Fab fragment antibody than for the whole antibody IgG as a result of the smaller size of this protein. The other important point to glean from Figure 2-15 is that during the initial three hours of the experiment the flux of the fragment antibody seems lower that the whole antibody in Figure 2-14B. The reason for this lower flux was ascribed to the fact that the sensitivity reach with Alexa 488 is much higher than with the FITC fluorophore. Therefore, these experiments demonstrated the capacity of these antibody-modified nanotube membranes to be used with a combination of both size and affinity selectivity for protein separation. Binding Capacities and Binding Constants Adsorption isotherms were carried out to evaluate the protein binding capacity and the binding constants for the permeating proteins to the antibody carriers. The data were analyzed using a monolayer Langmuir model and nonlinear regression methods for the fitting. A summary of the results obtained from the adsorption isotherms carried out with M-IgG modified, and H-IgG modified membranes, using the 100 nm pore size commercially alumina templates is shown in Figure 2-16. By observing the binding capacities in Figure 2-16 obtained from each isotherm, it is obvious, that these antibody-modified membranes preferentially bound the protein that binds stronger to the carrier on the membrane as was demonstrated above. It was also observed that a small amount of the weakly bounding proteins were bound to the carriers, which was due to antigen cross reactivity. Likewise, the results for the dissociation constant showed the same trend observed for the binding capacities. Lower dissociation constants (higher binding constant) were 77

PAGE 78

obtained by the proteins that strongly bind to the carrier. The binding constant errors found in Figure 2-16 for the proteins that weakly bind the carriers may result not only from the possible lack of equilibrium due to the weak affinity with the carrier, but also from the random orientation effect of the immobilized antibody. It was also demonstrated that by controlling the orientation of the immobilized antibody using protein G the potency affinity of the antibodies is effectively preserved (bars in the right side of Figure 2-16). Figure 2-16.Dissociation constants and binding capacities from adsorption isotherms of anti-M-IgG and anti-H-IgG on H-IgGand M-IgG-modified membranes, according to the Langmuir monolayer model. Effect of the Temperature on the Flux and Binding Constant For these studies, nanotube membranes templated on the 100 nm pore size commercially alumina were modified with M-IgG well-oriented immobilized via the protein G immobilization technique. Transport experiments (Figure 2-17), and adsorption isotherms (Figure 2-19) were carried out at three different temperatures: 23, 32, and 42 C. Figure 2-17 shows the flux plots for the transport experiments carried out at the three studied temperatures. It was observed that as the temperature increases the flux for anti-M-IgG 78

PAGE 79

also increases, which in turn decreases the selectivity coefficient between for these two permeating proteins. It should be pointed out that the flux of anti-H-IgG also increased at higher temperatures, but in a less extent compared with the fluxes for anti-M-IgG. A much clearer illustration of this flux and selectivity behavior with the changes in temperature is shown in Figure 2-18. Figure 2-17.Flux plots for anti-M-IgG () and anti-H-IgG () in well-oriented M-IgG -modified membranes showing the effect of temperature. The feed solution concentration was 21.2 nM in both of the permeating proteins. Figure 2-18.Effect of temperature on the flux and selectivity coefficients. The effect of the temperature in the affinity was study by carrying out adsorption isotherms for anti-M-IgG with well-oriented immobilized M-IgG at 23, 32, and 42 C (Figure 2-19). As 79

PAGE 80

should be expected adsorption isotherms perform to higher temperatures displayed a more linear character than the typical Langmuirian shape obtained at room temperature, indicating a lower affinity for anti-M-IgG. Figure 2-19.Adsortion isotherms carried out with well-oriented M-IgG-modified membranes for anti-M-IgG at three different temperatures: 23, 32, and 42 C. The results of the data analyzed according to the Langmuir monolayer model are summarized in the Figure 2-20. Higher dissociation constants (Lower binding constant) were obtained as the temperature is raised. Interestingly, it is the fact that the binding capacity of the antibody-modified nanotube membranes remains almost constant with the change in temperature, which is a clearly indication that the functional behavior of these antibody-modified membranes is governed for the immobilized carrier in the membrane. It is worth mentioning that previous studies reported in literature have demonstrated that alumina membranes have low thermal expansion and good stability at high temperatures. 183 In addition, temperature studies ranging from 20 to 60 C have demonstrated that the permeability of solutes across these porous membranes increases slightly with temperature compared with the diffusivity in the bulk solution. 184 Therefore, it can be concluded that the enhancement in the fluxes observed in Figure 2-17, as the temperature was raised, is mainly due to the lower affinity between the permeating protein and the antibody carrier immobilized on the membrane. An 80

PAGE 81

additional evidence for this conclusion is that the fluxes at higher temperatures for anti-M-IgG were increased in a more extent compared with the fluxes for anti-H-IgG. Figure 2-20.Affinity temperature dependence of anti-M-IgG in well-oriented M-IgG-modified membranes The data obtained in Figure 2-19 and plotted in Figure 2-20, can be used to determine thermodynamic parameters such as G, H, and S of the system in study. Thermodynamic costs for a system indicate whether processes are likely to occur and they are conveniently quantified in terms of free energy G. 185 Therefore, the feasibility of the protein and antibody to interact (at a constant pressure P, and temperature T) is determined by the change in free energy of the system, which for a reversible process can be written as (Equation 2-1): 186 000STHG (2-1) Where H 0 and S 0 are the enthalpic and entropic changes experienced during the biomolecular reaction. The enthalpic part is measure of the average potential energy of interaction between molecules, and the entropic part is a measure of the order or intermolecular correlations. 185 By assuming that the biomolecular reaction is a reversible process, the relationship between binding affinity constant (K a ), and free energy change can be expressed by Equation 2-2. 186 (2-2) aKRTGln0 81

PAGE 82

Since the values K a T and R the gas constant (8.314 Joules moles -1 K -1 ) are known, by using Equation 2-2 was possible to calculate G 0 The results of these energetic calculations are summarized in Table 2-2. Table 2-2.Thermodynamic parameters of G 0 associated with the affinity interaction of anti-M-IgG and M-IgG-modified membranes as a function of the temperature Temperature ( o C) Temperature ( K ) K a ( M -1 ) G o ( KJ / moles ) 23 296 3.76 x 10 8 48.59 32 305 7.81 x 10 7 46.08 42 315 3.19 x 10 7 45.24 As can be seen from Table 2-2, the resulting G 0 values were negative, which indicate that the process is exothermic and proceed spontaneously toward the formation of the complex protein-antibody. In addition, we can see how small changes in free energy generate significant changes in the binding affinity constant. Therefore, these changes in energy can be associated with the variations in the fluxes observed in Figure 2-17 when the temperature was changed. Now that that the changes of free energy as a function of the temperature were established, from a plot of G 0 vs. T the values of H 0 and S 0 can be extracted from Equation 2-1, by assuming that H 0 is independent of temperature. From the slope and intercept of the obtained plot with a linear correlation of 0.918 the estimated values for S 0 and H 0 were 0.17 0.06 KJ/mol and -99.95 17.12 KJ/mol, respectively. Those estimated values confirm the spontaneity of this biomolecular reaction toward the formation of the protein-antibody complex. Interesting is that S 0 is still positive even though the loss of rotational-translational degrees of freedom after complex formation. This effect has been extensively studied and had been associated with 82

PAGE 83

partial desolvation and amount of water excluded during the formation of multiple intermolecular noncovalent forces in the pocket of the binding site. 187 -190 Thus, these studies confirm that the functional behavior of these antibody-modified membranes is governed for the changes in free energy before and after interaction of the permeating proteins with the carrier antibody, which can be attributed to the changes in enthalpy and entropy of the system as was demonstrated here. Effect of the Low Ionic Strength on the Selectivity One method to improve selectivity in membrane-based separations is to shut down the transport of one of the permeating species. It was found that by using transport solution with low ionic strength it is possible to contemplate this option with antibody-modified membranes. To test this option, we studied the rate and selectivity of transport of anti-H-IgG and anti-M-IgG in only water solution across M-IgG modified nanotube membranes templated in commercial alumina with 100-nm-diameter pores (Figure 2-21). It was observed that after two hours of transport in low ionic strength conditions (only water), the fluxes of both the anti-H-IgG and anti-M-IgG were depleted (Figure 2-21A). The selectivity for these two permeating proteins through M-IgG membranes without equilibration was only 2.7. Alternatively, it was found that transport across M-IgG-modified membranes that were previously equilibrated for six hours in a water solution that was 2.65 nM in both of the permeating proteins (anti-H-IgG and anti-M-IgG), the selectivity coefficient was increased to 17.3. This effect is shown in Figure 2-21B. The effect of low ionic strength in the selectivity was attributed to the fact that IgG has no unique isoelectric point, since each specific antibody molecule has its own variable region containing different charged residues. 191 Thus, for transport experiments carried out at low ionic strength the thickness of the electric double-layer has a considerable influence in the mobility of the permeating proteins across the antibody-modified membranes. 186, 192 Consequently, for 83

PAGE 84

transport experiments carried out at low ionic strength in antibody-modified membranes that has been previously equilibrate with a protein that strongly binds the antibody carrier, both the modified membrane and this permeating protein bearing similar electric double-layer that creates double-layer repulsions when they approach each other. Besides, at low ionic strength the overlapping between the double-layers both of the membrane pore wall, and that one of the permeating protein affects the mobility across the membrane. 193, 194 Therefore, the repulsion effects together with the double-layer overlapping are the responsible factors of the enhanced selectivity at low ionic strength observed in Figure 2-21B. Then, the high selectivity obtained under these conditions might be attributed to the electrostatic exclusion of the charged anti-M-IgG molecules from the membrane pores that were equilibrated with the same protein (anti-M-IgG) having identical isoelectric point. 162, 193 Figure 2-21.Flux plots for anti-M-IgG () and anti-H-IgG () across M-IgG -modified membranes, (A) without equilibration, and (B) after equilibration. The feed solution concentration was 2.65 nM in both of the permeating proteins. These data suggest that this antibody-modified nanotube-membranes at low ionic strength can be used for situations were the transport of an unwanted protein needs to be shut down, in order to improve selectivity. One potential application that this approach might be useful is in the sample complexity reduction in proteomic analysis. Since the analysis and characterization of 84

PAGE 85

complex mixtures of proteins are the central aims of proteomics. 195 Thus, it would make possible to deplete the flux of high abundance proteins across the antibody-modified membranes while at the same time it would allow improving the separation and detection of low abundance proteins by subsequent analytical techniques. Conclusions A detailed experimental study on antibody-functionalized nanotube membranes has been carried out, including the effect of surface modification, pore size, feed concentration, temperature, and ionic strength, both on the transport as on the functional behavior of the membranes. It was demonstrated that the type of immobilized antibody changes the functional behavior of the antibody-modified membranes and the transport properties of the permeating proteins across them. These membranes showed a Langmuirian shape plot characteristic of a typical behavior of facilitated transport membranes, when the effect of feed solution concentration on the fluxes of the permeating proteins was investigated. Therefore, the antibody-functionalized membranes showed here exhibit a type of behavior for facilitated-transport membranes during protein separations. Important parameters such as the binding capacities of the antibody-modified membranes and the binding constants for the permeating proteins to the antibody carriers confirmed that the permeating protein anti-M-IgG binds strongly to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to this carrier, likewise, the permeating protein anti-H-IgG binds strongly to the carrier H-IgG and that anti-M-IgG binds weakly to this carrier. It was also demonstrated that by controlling the orientation of the immobilized antibody using protein G the potency affinity of the antibodies is effectively preserved. Through these studies was also demonstrated that by selecting a template membrane with an appropriated pore size diameters these antibody-modified nanotube membranes have the ability to be used with a combination of both size and affinity selectivity for protein separation. 85

PAGE 86

Additional studies on the effect of temperature on the fluxes and the binding constant demonstrated that the functional behavior of these antibody-modified membranes is governed for the changes in free energy before and after interaction of the permeating proteins with the carrier antibody, which was attributed to the changes in enthalpy and entropy of the system. Other important finding during this investigation was that by controlling the ionic strength of the transport solution, these antibody-modified nanotube membranes can be used for situations were the transport of an unwanted protein needs to be shut down, in order to improve selectivity. This latter approach might be useful in the sample complexity reduction in proteomic analysis. Since the analysis and characterization of complex mixtures of proteins are the central aims of proteomics. It is also noteworthy to mention the low pM detection limits achieved during these studies. In general, the ability to alter the surface biochemistries and thus tailor the membrane for specific analytes, promises the use of these membranes for the analysis of a wide range of proteins. 86

PAGE 87

CHAPTER 3 MODELING AND CHARACTERIZATION OF THE TRANSPORT PROPERTIES OF PROTEINS ACROSS ANTIBODY-FUNCTIONALIZED NANOTUBE MEMBRANES BY USING THE DUAL-MODE MODEL Introduction In the design and development of more effective membranes for membrane-based separations, the value of the transport mechanism must be considered in order to understand the different interactions that govern transport through the pores of the membranes. In addition, the transport mechanism, supported by mathematical models and experimental evidences provides a grater understanding of the underlying transport phenomena. The separation properties of molecules through membranes are based on the difference of permeation. Therefore, it is important to understand permeation behavior, not only to understand the permeation mechanism but also to elucidate the separation properties. In order to elucidate the transport properties across selective membranes a wide variety of models have been developed. 196 202 However, a model that explains the sorption and permeation mechanism is of fundamental importance to fully understand the transport properties in membrane separation. The well-known dual-mode model allows evaluation of both sorption and permeation mechanism, for those cases where the model applies. The dual-mode model was initially developed by Vieth and Sladek 203 and later modified by Paul and Koros 204 to describe the sorption and permeation of gas molecules in glassy polymeric films. In literature there are various modified dual-mode models that have also been explored. 182, 205 208 The model assumes that the membrane contains two distinct regions in which sorption can occur. 204 One region exhibits solubility of the molecule based on Henrys law and the second 87

PAGE 88

region has a Langmuir sorption isotherm. For permeation, it is assumed that each region is equilibrated and permeation is governed by a solution-diffusion mechanism. Figure 3-1.Dual-mode transport model. The dual-mode model has also been useful to explain membranes with higher transport selectivity obtained via a methodology called facilitated transport. 114, 115, 209 These models typically draw an analogy between the facilitated transport process and the dual-mode model for gas transport in glassy polymers. During facilitated transport, the enhanced transport of a specific permeate is mediated by a carrier immobilized within the membrane, plus the normal physical permeation pathway, since this carrier in the membrane reversibly binds the specific permeate molecule. In Figure 3-1 is shown a schematic illustration of the dual-mode model presented here. This chapter presents the results on the characterization of the transport properties of anti-M-IgG and anti-H-IgG across M-IgG modified membranes by using the dual-mode model. We have studied the effect of systematic chemical modification on the transport properties of silica nanotube membranes by functionalization with covalently attached-carrier-antibodies as a means of selectively improving transport properties of proteins. It is discussed how the presence of this carrier in the nanotube-functionalized membranes enhance the permeability of the transported anti-IgG. This enhancement in permeability at low concentration is explained with the dual-mode model and it is caused by a contribution of the diffusion due to the Langmuir mode 88

PAGE 89

species. Based on these results, the concept of modulated facilitated transport using antibody-modified nanotube membranes is proposed. Thus, depending on the affinity parameters, the transport in the carrier situation can be modulated in comparison to the situation without carrier modification where the rate of transport is only controlled by physical diffusion. The present study test the dual-mode model under circumstance where is possible to have modulated facilitated transport, and simulations and experimental results are discuses to validate this concept. Definition of Dual-Mode Model The dual-mode model is widely used to describe the sorption of gas molecules on glassy polymers 210, 211 Depending on their internal structure, the polymers are characterized as either glassy or rubbery. Sorption of gases to the rubbery state occurs by dissolution, while sorption to glassy state occurs by concurrent dissolution and hole-filling mechanisms. 212, 213 The sorption mechanism of gas molecules in glassy polymers has been understood in term of the dual-mode model. 203 208 The model is based on two types of sorption sites. One obeys the Henrys law dissolution and the second a Langmuir-type sorption. According to the dual-mode model, total sorption to surface membrane is the sum of sorption in the dissolution domain and sorption in the Langmuir-type sorption domain (Equation 3-1): 214 bpbpCpkCCCsitesDHD1 (3-1) By analogy but making a couple of changes, we have used the dual-mode model used for gas-transport to interpret transport of proteins in aqueous solution through antibody-modified nanotube membranes. Instead of reporting the concentration of penetrant in the membrane for a given gas pressure, the concentration is reported as a function of the feed protein concentration. The Henrys law coefficient, k D, was also adjusted and it was given in terms of partition 89

PAGE 90

coefficient k p Since, the sorption coefficient given in the Equation 3-1 is a gas-phase coefficient; in solution therefore the sorption coefficient should be considered as a liquid-phase coefficient. The adapted dual-mode model is shown in Equation 3-2. feedfeedsitesfeedpHDbCbCCCkCCC1 (3-2) Where C is the total protein concentration in the membrane, C D is the protein concentration based on Henrys law sorption, C H is the protein concentration based on Langmuir sorption. Whereas, k p is defined as the partition coefficient, b and C sites are the Langmuir affinity constant and maximum capacity parameters, respectively. The Langmuir affinity constant, b, characterizes the sorption affinity for a particular protein-antibody membrane system, and C sites is used to measure the amount of antibody immobilized in the membrane. The three dual-mode sorption parameters, k p C Sites and b, can be determined through the experimental evaluation of protein-adsorption isotherms. The estimation for C Sites and b, sorption parameters can be accomplished by analyzing the data according to the monolayer Langmuir model (Equation 3-3): 167 FreeFreeSitesBoundbCbCCC1 (3-3) Where C Bound is the specific concentration of protein bound to the carriers in the membrane, which is related to each specific concentration of free protein (C free ) in the equilibration solution. The other of the three sorption parameters contained in the dual-mode model is the partition coefficient. Through the analysis of the boundary conditions of the Langmuir-mode model, an optimum parameter value for the partition coefficient could be estimated. 90

PAGE 91

Thus, at the high concentration range, the Langmuir model approaches to a constant value that is equal to C Sites so no more sorption will take place when all the sites are filled. On the other hand, at very low concentrations, the Langmuir model obeys the correct thermodynamic boundary condition of Henrys law over infinitely dilute concentration range. 215, 216 In the limit of zero concentration the Langmuir model leads to the following equation which satisfies the condition of a linear isotherm (Equation 3-4): FeedDFeedSitesCCkbCCC 0lim (3-4) Hence, a good estimation of Henrys law coefficient, k D, may be written as the product of C Sites and b. Consequently, this k D parameter can be used for the calculation of the partition coefficient (k p ), which is defined as the product of k D times the ratio of volumes between volume inside the membrane (V Membrane ) and volume in the free equilibration solution (V Free Solution ) (Equation 3-5). SolutionFreeMembraneDpVVkk (3-5) It should be pointed out that for linear isotherms, the Henrys law coefficient, k D, can be determined from the slope, while for dual-mode adsorption isotherms, it can be determined from the initial slope at infinite dilute concentration. The partition coefficients can also be calculated using Equation 3-6, which expresses the ratio of the intramembrane concentration to external concentration at equilibrium. SolutionFreeSolutionFreeMembraneMembranepCVCVk (3-6) All this dual-mode sorption model theory explained until here can be validated using linearization methods for the Equation 3-2. Several linear plots of the dual-mode sorption model have been studied, 217 and for instance, the equation can be rearranged and transformed into the Equation 3-7. 91

PAGE 92

FeedSitesSitesFeedpFeedCCbCCkCC11 (3-7) A plot of C Feed / (C k p C Feed ) against C Feed is expected to be linear, where the Langmuir sorption parameters, C sites and b, can be evaluated from the slope and intercept of the straight line, respectively. 217 Now, if the solute-adsorption isotherms are represented by the dual-mode model, as shown above, the permeation mechanism can also be described in terms of the dual-mode model. The permeation mechanism using the dual-mode model assumes partial immobilization of the Langmuir-mode species. 204 Therefore, independent dual diffusion coefficients are given for each sorption mode. One diffusion coefficient (D D ) is used for the fraction of the solute dissolved in the membrane according to Henrys law expression and a second (D H ) for the fraction of the solute contained in the Langmuir-type sorption mode. The expression for the permeation flux through the membrane is derived by Ficks law, and it can be written in the form given in Equation 3-8. xCCFKDxCDxCDJDDDHHDD211 (3-8) F corresponds to the mobile fraction of the Langmuir species (Equation 3-9). DHDDF (3-9) Where and K are constants that contain the three sorption parameters (Equation 3-10 and Equation 3-11). pSiteskbCK (3-10); pkb (3-11) The mathematical solution that Koros and Paul found for the Equation 3-8 is known as the partial immobilization model. 204 The model assumes that there is always local equilibrium between the two sorption modes, thus, C H can be written in terms of C D In addition, the 92

PAGE 93

diffusion coefficient is assumed constant so it does not depend on concentration. Besides, the model splits the total concentration into two parts, one a mobile part with a diffusion coefficient D D and concentration C m and the second remainder part (C C m ), corresponds to the part that is totally immobilized. In other words, the concentration associated with C D as well as a fraction F of that associated with C H has finite mobility while the remaining fraction, 1-F, of C H has zero mobility. Therefore, the concentration for the mobile part can be written as (Equation 3-12): HDmFCCC (3-12) The flux for this mobile part may be written as is shown in Equation 3-13. xCDJmD (3-13) Whereas the equation of continuity (Ficks second law), represented by the dual-mode model for a solute transported through the membrane is given by the Equation 3-14. xJtCCtCHD)( (3-14) The equation of continuity is nothing more that the mass balance for the system, which relates the accumulation of solute in the porous membrane to the changes in flux at any time and any direction of the flux. In other words it is the rate of solute balance. Thus, the difference between the rate of solute that comes in to the membrane, and the rate of solute out, is equal to the rate of solute accumulation. The insertion of the Equation 3-13 into the Equation 3-14, results in Equation 3-15. xxCDtCCmDHD)()( (3-15) By assuming that the diffusion coefficient does not depend on concentration, and inserting the Equation 3-12 into the mass balance equation given by the Equation 3-15, the equation becomes (Equation 3-16) 22)()( x FCCDtCCHDDHD (3-16) 93

PAGE 94

Finally, by setting all the assumptions and writing the concentration of C H in terms of C D (KC D / 1+C D ), a numerical solution of the partial immobilization model yields to an expression for the permeability coefficient given by Equation 3-17. 204 FeedDpbCFKDkP11 (3-17) The use of the model requires the determination of both D D and F. This can be accomplished through the experimental evaluation of the permeability coefficients; so, optimum parameters for the diffusion coefficients can be determined. Throughout every experimental permeation, measurements of the flux per unit area, J, (moles cm -2 sec -1 ), is obtained. Therefore, if a number of permeation experiments are carried out at different feed concentration using membranes with homogeneous thickness, L, the steady-state permeability coefficient for a specific feed concentration can be calculated using the following relationship showed in Equation 3-18. 218, 219 FeedCJLP (3-18) This procedure would also allow comparing the experimental permeability plots with those calculated obtained ones. In order to explain permeation, the dual-mode model assumes that the permeation is governed via a solution-diffusion mechanism. If the permeation of the protein in the membrane is controlled by a solution-diffusion mechanism, the permeability can be expressed as the product of diffusivity, D, and solubility, S as is shown in Equation 3-19. DSP (3-19) Although the solubility of a permeating protein in the membrane can be also expressed in term of the dual-mode sorption parameters by (Equation 3-20): 210, 211 FeedSitespHDFeedbCbCkSSCCS1 (3-20) 94

PAGE 95

Where S is the solubility of permeating protein, and the terms S D and S H corresponds to the solubility values based on Henrys law and Langmuir-type sorption, respectively. If the diffusion coefficient does not depend on concentration, a plot of permeability against solubility is expected to be linear, with a slope equals to the diffusion coefficient (Equation 3-19). Through this correlation between P and S experimentally the diffusion coefficients can be estimated. With the aim of determining the F parameter, the experimental permeability data can be plotted against (1 + bC Feed ) -1 A value for F can be estimated from the slope of this line. 211 Additionally, the diffusion coefficient D D might be evaluated from the intercept as well, but only if the plot is straight line with a very high linear correlation coefficient (Equation 3-21). FeedDSitesDpbCbFDCDkP1 (3-21) In fact, the dual-mode model makes feasible the evaluation of the sorption and permeation mechanism. This is of fundamental importance to fully understand the transport properties in membrane separation. Hence, using the dual-mode model, a comprehensible explanation of the transport mechanism might be accomplished through a detailed elucidation of the transport properties such as sorption, diffusion and permeability. Results and Discussion Simulations of Permeability vs. Feed Concentration Based on the Dual-Mode Model The permeability is the product between diffusivity and solubility as is described in Equation 3-19. The permeability dependence in terms of the dual-mode model is not only related with the concentration of solute molecules in the feed solution, but also with the different properties and interactions between solute-membrane. Thus, the permeation of solute molecules 95

PAGE 96

through carrier-modified membranes, which is governed by both the Henry mode and the Langmuir mode, can be modulated depending on the affinity parameters between solute-carrier. In order to understand how the dual-mode model works, simulations of the model were initially carried out (Figure 3-2). The simulation analysis was based on the assumption of competitive permeation studies for two different solute species transported across carrier-modified nanoporous membranes. With the aim of simplifying the analysis an idealized case was compared. Thus, one of the species the spectator or nonbinding molecule (NBM) is assumed to have no interaction with the carriers in membrane and therefore permeates the membrane by only physical diffusion. On the other hand, the second specie, the binding molecule (BM), permeates the membrane through a dual-mode model mechanism. The different dual-mode parameters used for the simulations are listed in Table 3-1. Figure 3-2.Permeability plot simulations for the transport of (a) BM and (b) NBM across carrier-modified membranes. The parameters are summarized in Table 3-1. By simple visual analysis of the simulations showed in Figure 3-2, we can notice that for a permeation mechanism that follows the dual-mode model, the permeability is highly dependent on the concentration, while for a mechanism based on physical diffusion only, the permeability is totally independent of the concentration. The permeability in terms of the dual-mode model 96

PAGE 97

decreases with increasing feed concentration, which is concave to the concentration axis as can be seen in the plot (a) in Figure 3-2. Table 3-1.Dual-mode model parameters used for the permeability plot simulations Permeant specie k p D (cm 2 /sec) C sites (moles/cm 3 ) b (cm 3 /moles) Nonbinding Molecule (NBM) 6 x 10 -3 9 x 10 -8 Binding Molecule (BM) 6 x 10 -3 9 x 10 -8 1.5 x 10 -10 1 x 10 11 Figure 3-3 shows simulations for the permeability versus feed concentration, illustrating the effect of the F parameter in dual-mode model mechanism. Figure 3-3 compares three simulation plots for the case of the BM situation, which correspond to three different values of F. The term F is associated with the mobile fraction of the Langmuir species as was stated before, and its magnitude is equal to unity or zero, when the model assumes no immobilization or total immobilization of the Langmuir species, respectively. Figure 3-3.Permeability plot simulations for the transport of BM across carrier-modified membranes illustrating the effect of the term F on the dual-mode model. (a) F = 1, (b) F = 0.5, and (c) F = 0. The other parameters are summarized in Table 3-1. 97

PAGE 98

Hence, the permeability for a solute molecule that permeates a membrane through a dual-mode mechanism depends also on the magnitude of F. For the BM situation, the permeability is highly dependent on the feed concentration as F gets closer to unity (plot a, in Figure 3-3). Consequently, as the value of F decreases or the degree of immobilization increases, the permeability becomes less dependent on the feed concentration (plot b, in Figure 3-3). When there is total immobilization, when F is equal to zero, the permeability is independent of the concentration, and so the BM dissolves and diffuses through the membrane via Henrys mode dissolution only (plot c, in Figure 3-3). It should be pointed out that regardless of the value of F, the permeability decreases gradually with the increase of feed concentration until a steady level that reach the value of k p D D The above simulations in Figures 3-2 and 3-3 suggested that when no immobilization of the Langmuir species is achieved, the permeability across carrier-modified membranes is always enhanced compared with the physical diffusion case. Therefore, the transport through carrier-modified membranes that fulfill the dual-mode model mechanism is always facilitated when the Langmuir species no experience immobilization. The concentration dependence of the permeability for the dual-mode model can also be analyzed through the evaluation of the limit boundaries conditions in concentration for the equation of permeability (Equation 3-17). At the high limit concentration range the equation for permeability becomes equal to the product between k p times D, this is due to the behavior of the factor [1 + FK / (1 + b C Feed )], which becomes closer to unity as the concentration increases. At the low limit concentration, the highest permeability value is reached when the denominator term, (1 + b C Feed ), becomes unity. As the latter gets closer to one, there is a range in concentrations where the variations in permeability are less noticeable, and the permeability 98

PAGE 99

becomes less dependent on the concentration. This region resembles to the time lag region for gases in glassy polymers. 181, 182 In general, as stated by Koros and Paul the limits of the dual-mode model are confined to two situations. 204 One is the limit where total immobilization occurs, and only the Henrys mode species contributes to the permeability. The other limit is a situation with a nonlinear isotherm with no distinctions for sorption mechanism or transport behavior between any of the adsorbed solutes. However, the concentration-dependence of the permeability is a consequence of the shape of the adsorption isotherm, and therefore a result of the binding affinity strength between Langmuir permeating species and Langmuir binding sites. In order to demonstrate the concentration dependence on the permeability as consequence of the shape of the adsorption isotherm, permeability plot simulations were carried out comparing two different sorption mechanisms via two different binding affinities (Figure 3-4). The plot (a) in Figure 3-4 clearly indicates that a carrier-modified membrane with stronger affinity for the solute molecules displays higher concentration-dependence as a result of the sorption isotherm. So, the concentration dependence of permeability is determined by the strength of the binding affinity for the solute-carrier complex in the membrane. However, the total permeability is highly dependent on the diffusion coefficient. It is worth mentioning that by using simulations was verified that the order of influence of the dual-mode model parameters in the permeability was D > b > C sites >>> k p respectively. According to these results, the dual-mode model permeation analysis is a useful model to explain transport across carrier-modified membranes. The analysis showed until here demonstrated how the fluctuations in permeability are a result of the strength of the sorption mechanism. Therefore, the differences in permeability and the differences in permeability 99

PAGE 100

concentration-dependence can be used to elucidate the separation properties for the solute molecules that permeate a carrier-modified membrane. For the reason that differences in permeability are a consequence of the sorption mechanism, the separation properties of one solute might differ from that of another, depending on the affinity solute-carrier interaction. Figure 3-4.Permeability plot simulations for the transport of BM across carrier-modified membranes showing the effect of the binding affinity on the permeability. (a) b = 1 x 10 11 cm 3 /moles, and (b) b = 3 x 10 10 cm 3 /moles. The other parameters are summarized in Table 3-1. Experimental Description of the Dual-Mode Model Effect of feed concentration on the flux A number of competitive permeation experiments were carried out at different feed concentration, ranging from 2.65 x 10 -12 to 5.0 x 10 -10 moles cm -3 using M-IgG modified membranes (Figure 3-5). Throughout every experimental permeation measurements, the flux per unit area (moles cm -2 sec -1 ) was obtained, using 0.31 cm 2 as the area of membrane exposed to the contacting solution phases. Figure 3-5 shows the studies of the effect of feed concentration on the flux during competitive permeation experiments of anti-H-IgG and anti-M-IgG across M-IgG-modified membranes. The flux for anti-H-IgG was at all times higher than the flux for anti-M-IgG. The linear dynamic range for anti-M-IgG was 3 times lower than for anti-H-IgG, as a consequence of 100

PAGE 101

a stronger binding affinity with the M-IgG carrier attached in the membrane. These plots show a characteristic Langmuirian shapes, demonstrating that the immobilized M-IgG indeed interacts with the permeating proteins, acting as a carrier during the permeation process. 9, 10, 109 Figure 3-5.Flux versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. The feed solution concentration was equimolar in both of the permeating proteins. Experimental permeabilities as function of the feed concentration In order to conduct fundamental studies of any membrane process, a well defined membrane system is required. One of the worthy characteristics of the alumina template membranes used in these studies is their well defined and characterized structure. Thus, it is possible compare theoretical models with the experimental results as is demonstrated in this dissertation. Because the thickness of the membranes and feed concentration for each permeation experiment were known, the Equation 3-18 for permeability was used to calculate the experimental permeability coefficient of anti-M-IgG and anti-H-IgG across M-IgG modified membranes (Figure 3-6). Figure 3-6 shows the plots for the experimental permeability versus feed concentration. It was observed that the permeability decreases as the concentration increases both for anti-H-IgG and for anti-M-IgG across M-IgG modified membranes. This diminution in permeability may be understood in terms of the 101

PAGE 102

dual-mode model, 204 206, 211, 220 since the permeability in terms of the dual-mode model decreases with increasing feed concentration. 221 Figure 3-6. Experimental obtained permeability versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. At first sight from the data of concentration dependence of the experimental permeability in Figure 3-6, we might anticipate that the transport mechanism for these two proteins across M-IgG modified membranes approaches to the dual mode model mechanism. In order to prove this transport mechanism for these two proteins across the M-IgG modified membranes, an evaluation of the sorption and permeation parameters were carried out, and they are discussed next. Experimental evaluation of the dual-mode model parameters The mathematical solution that Koros and Paul found for the permeability coefficients based on the dual mode model is known as the partial immobilization model (Equation 3-17). 204 The use of the model requires the determination of both D D and F, which were evaluated experimentally by using the permeability coefficients in Figure 3-6. Additionally, to use the model is a prerequisite to have the information of the three dual-mode sorption parameters, k p C Sites and b, which were also determined experimentally via protein-adsorption isotherms. 102

PAGE 103

Experimental evaluation of C sites and b These Langmuir sorption parameters were determined experimentally using adsorption isotherms. The relationship that exists between concentrations of protein in the porous membrane isothermally in equilibrium with those in the feed solution is called adsorption isotherm. In order to assess the equilibration time for the sorption experiments, a sorption dynamic study was initially carried out (Figure 3-7). For this, a solution that was 21.2 nM of anti-M-IgG labeled with Alexa 488 was presented to M-IgG modified membranes. The amount of anti-M-IgG adsorbed was investigated observing the changes in fluorescence intensity as a function of time. Figure 3-7.Curve of adsorption dynamics for anti-M-IgG () in a M-IgG modified membrane. Inset shows the changes of fluorescence signal with time. As can be seen from Figure 3-7, the amount of anti-M-IgG adsorbed in the M-IgG modified membrane increases rapidly during the first 3 hours. After that, the rate of adsorption decreases and the system reaches a plateau of apparently equilibrium. All the adsorption isotherms were carried out with M-IgG modified membranes at 23 C in stirred U-tube permeation cells using the same buffer conditions used for the permeation studies, via solutions containing single concentrations of anti-M-IgG or anti-H-IgG fluorescent labeled 103

PAGE 104

with Alexa 488. Data were analyzed according to the monolayer Langmuir model (Equation 3-3). 167 By using the experimental values of C free and C bound in moles per cm 3 in Equation 3-3 the parameter C sites was calculated in moles per cm 3 while b in cm 3 per moles. Figure 3-8 shows and compares adsorption isotherms for anti-M-IgG and anti-H-IgG carried out with M-IgG modified membranes; also the adsorption isotherm for anti-M-IgG using unmodified nanotube membranes is showed. Table 3-2 summarizes the corresponding binding constant and protein binding capacities, as calculated by fitting the data using nonlinear regression methods. Figure 3-8. Adsorption isotherms for anti-M-IgG (), and anti-H-IgG () binding on M-IgG modified membranes. Also the adsorption isotherm for anti-M-IgG () using a control membrane is shown. Figure 3-8 shows the adsorption isotherms obtained from the fluorescence intensity of the wet membranes exposed to different protein concentrations. For the unmodified silica nanotube membranes the adsorption of anti-M-IgG labeled with Alexa 488 was negligible in comparison with the antibody-modified membranes. It was also observed, that the adsorption for the unmodified membranes increases proportionally with the concentration of fluorescent-labeled 104

PAGE 105

antibody, in the entire range of concentration used for the saturation binding experiments (from 1.0 x 10 -12 to 2.5 x 10 -11 moles cm -3 ), obeying clearly a Henrys linear isotherm. The linear shape of the unmodified isotherm also indicates a very low affinity for the anti-M-IgG. Thus, these results also suggested that the silica nanotube membranes exhibit low protein binding (non-specific antibody adsorption) to the surface membrane, and minimal autofluorescence. On the other hand, the M-IgG modified membranes exhibited clearly nonlinear isotherms and they bound more protein than the unmodified ones. When anti-H-IgG labeled with Alexa 488 was presented to M-IgG modified membranes the adsorption isotherm was nonlinear. This effect is due mainly to the existence of cross-reactivity between the permeating anti-H-IgG and the attached M-IgG. That kind of cross-reactivity of IgG has been also seen in the literature. 169 172 By fitting the experimental adsorption isotherm for anti-H-IgG, the existence of one binding site for the antibody was demonstrated. A binding affinity constant equal to (2.90 1.35) x 10 10 cm 3 moles -1 was obtained. The amount of protein bound to the membranes exposed to anti-H-IgG was almost twice compared to the unmodified silica nanotubes ones (Table 3-2). In contrast to the anti-H-IgG sorption isotherm case, when anti-M-IgG labeled with Alexa 488 was presented to the M-IgG modified membranes the protein bound to the membrane was approximately 20 times higher than the anti-H-IgG situation. These data confirmed the successful modification of the membranes, and the remarkable selective attribute of the antibody-modified membranes. The fitting process for the M-IgG modified sorption isotherms exposed to anti-M-IgG indicated the existence of two binding sites in the membrane for the permeating protein (Figure 3-8). This is probably due to the spatial orientation of the M-IgG molecules immobilized on the 105

PAGE 106

pore surface of the membrane. Therefore, the binding site with the stronger binding constant, (4.46 0.91) x 10 11 cm 3 moles -1 was attributed to the concentration of well-oriented immobilized IgG molecules. The other site with an affinity constant of (5.27 1.58) x 10 10 cm 3 moles -1 was ascribed to the random-oriented concentration. Thus, anti-M-IgG first binds to the site of well-oriented immobilized M-IgG. After a fixed amount of anti-M-IgG is adsorbed the random immobilized M-IgG become available as cooperative binding sites. Figure 3-9.Adsorption isotherm for anti-M-IgG () binding on M-IgG-PG modified membranes. Additional confirmation for this conclusion about two binding sites with different spatial orientation was obtained through adsorption isotherms of anti-M-IgG using M-IgG well-oriented antibody-modified membranes via protein-G (PG) immobilization (Figure 3-9). The well-oriented immobilization of M-IgG was achieved through silica nanotube membranes already modified with PG. This PG was covalently attached via the Schiffs base reaction to the walls of aldehyde modified nanotube membranes. The adsorption isotherm with M-IgG-PG modified membranes revealed the existence of only one binding site by using this well-oriented antibody-immobilization (Figure 3-9). For this specific case, the value of the binding affinity constant was (3.77 0.53) x 10 11 cm 3 moles -1 (Table 3-2). 106

PAGE 107

It should be pointed out that the antibodies used in these studies were all polyclonal antibodies. Therefore, the found binding affinity constants are indeed an average of the pool of the different clonally related subpopulations of antibodies existing in the sample solution. 222, 223 Table 3-2.Thermodynamic parameters from adsorption isotherms of anti-M-IgG and anti-H-IgG on M-IgG modified membranes, according to the Langmuir monolayer model Adsorption Isotherms C sites1 (moles/cm 3 ) b 1 (cm 3 /moles) C sites2 (moles/cm 3 ) b 2 (cm 3 /moles) R anti-H-IgG / M-IgG (7.24 1.33) x 10 -11 (2.90 1.35) x 10 10 0.983 anti-M-IgG / M-IgG (1.16 0.19) x 10 -10 (4.46 0.91) x 10 11 (1.30 0.39) x 10 -9 (5.27 1.58) x 10 10 0.928 anti-M-IgG /M-IgG-PG (3.49 0.14) x 10 -10 (3.77 0.53) x 10 11 0.992 Experimental evaluation of the partition coefficients k p For definition, partition coefficients provide a thermodynamic measurement of the solubility of the permeating proteins in the membrane. Therefore, considering this thermodynamic fact, proteins with higher diffusion coefficient should have lower partition coefficients. Contrarily, an increase of the partition coefficient increases the resident time of the permeating proteins in the membrane. The partition coefficient is one the other three sorption parameters contained in the dual-mode model. The experimental evaluation of the partition coefficients across M-IgG modified membranes for the proteins studied here were investigated using three different approaches. Table 3-3 compiles the results for these studies. The first approach for the analysis of k p was based on the examination of the boundary conditions of the Langmuir-mode in the limit of zero concentration, that determine that the Henrys law coefficient (k D ) may be written as the product of C Sites times b (Equation 3-4). 215, 216 107

PAGE 108

Since the values of C Sites and b were already estimated for both of the permeating proteins studied here as was shown above. The partition coefficients calculated using the Equation 3-5 were 7.24 x 10 -3 and 2.94 x 10 -4 for anti-M-IgG and for anti-H-IgG, respectively. Table 3-3.The k p determined by various approaches Method anti-M-IgG Anti-H-IgG Equations 3-4 and 3-5 7.24 x 10 -3 2.94 x 10 -4 Slopes Figure 3-10 6.14 x 10 -3 2.77 x 10 -4 Equation 3-6 (5.84 1.06) x 10 -3 (2.79 0.69) x 10 -4 It is important to mention that for linear isotherms, the Henrys law coefficient, k D, can be determined from the slope of the isotherm plot, while for dual-mode sorption isotherms, it can be determined from the initial slope at infinite dilute concentration (Figure 3-10). To prove this assertion the fitting plots of the adsorption isotherms for anti-M-IgG and anti-H-IgG were extrapolated to a very dilute concentration range (from 0 to 6 x 10 -13 moles cm -3 ), and the respective slopes from the linear plots were estimated, these results are shown in Figure 3-10. Figure 3-10. Extrapolations of the adsorption isotherms at infinite dilute concentration for anti-M-IgG (), and anti-H-IgG () in M-IgG modified membranes. 108

PAGE 109

The k D obtained for anti-M-IgG and anti-H-IgG from the slopes in Figure 3-10 were 43.85 and 1.98, respectively. Likewise, using Equation 3-5 a partition coefficient of 6.14 x 10 -3 for anti-M-IgG and 2.77 x 10 -4 for anti-H-IgG were found. The partition coefficients were also calculated using Equation 3-6, which expresses the ratio of the intramembrane concentration to external concentration at equilibrium. The partition coefficients estimated with this equation were comparable to those calculated by using the direct method via k D As is shown in Table 3-3, the partition coefficients in this case were (5.84 1.06) x 10 -3 and (2.79 0.69) x 10 -4 for anti-M-IgG and anti-H-IgG, respectively. In order to validate these partition coefficients, dual-mode model permeability plot simulations were carried out (Figure 3-11). For these simulations, four different values for the partition coefficients were assigned by keeping the rest of the dual-mode model parameters constant. The results shown in Figure 3-11 demonstrated that variations in the partition coefficients have a negligible effect in the permeability coefficients when the binding affinity is fairly strong as the studies showed here. Figure 3-11.Permeability plot simulations illustrating the effect of the partition coefficient on the dual-mode model. Simulation parameters: D = 5 x 10 -8 cm 2 /sec, b = 9 x 10 10 cm 3 /moles, C sites = 3.5 x 10 -8 moles/cm 3 and k p were respectively, 1 (), 0.1 (), 0.01 (), and 0.001(). 109

PAGE 110

Experimental evaluation of the diffusion coefficients and solubility The most direct method of experimental determination of S and D requires steady-state permeation data, complemented with equilibrium sorption data. 199 Therefore, effective values of S, D, and hence P characterized for a membrane-penetrant system that has been restricted to equilibrium sorption and steady-state permeation conditions represent appropriate averages of the transport parameter of the system. 224 To use the dual-mode model requires knowing D and S. This was accomplished experimentally through an evaluation of the permeability and solubility in terms of the dual-mode model as will be shown next. As was state before, the solubility for a penetrant protein in the membrane can be expressed in term of the dual-mode sorption parameters by the Equation 3-20. 210 For the reason that the three sorption parameters required for this calculation were already established, an experimental study of the concentration dependence of the solubility was conducted (Figure 3-12). Figure 3-12 shows the concentration-dependence of the solubility in term of the dual-mode model for anti-M-IgG and anti-H-IgG in M-IgG modified membranes. The concentration-dependence of the solubility in term of the dual model is a consequence of the shape of the adsorption isotherm. Hence, from Figure 3-12 M-IgG modified membranes possess higher affinity for anti-M-IgG than for anti-H-IgG, as was demonstrated before (Table 3-2). These data also indicate where a protein molecule prefers to dissolve, whether in the Langmuir binding sites or in the membrane matrix based on Henrys law. According to Figure 3-12, anti-M-IgG molecules would rather dissolve in the Langmuir sorption mode (upper plot), while anti-H-IgG molecules in the Henrys mode (lower plot). It was also observed that regardless of the protein at low concentrations the Langmuir mode contributes the most to the total solubility, which is probably due to the strong affinity exerted for that fraction of well-oriented immobilized carriers 110

PAGE 111

in the membrane. Therefore, at low concentration the solubility of the proteins was highly increased by the presence of the M-IgG carriers, so those can be used as facilitated partition domains of the transport. In addition, the data revealed that as the concentration increases, the solubility decreases and reaches a plateau where the solubility in the membrane is only controlled by the solubility in the Henrys mode. Figure 3-12. Concentration-dependence of the solubility in term of the dual-mode model for anti-M-IgG (), and anti-H-IgG () across M-IgG modified membranes. As was described before in Equation 3-19, the permeability is the product between diffusion and solubility. Since both the concentration dependence of the permeability and solubility were experimentally determined, they were used to estimate the effective values for the diffusion coefficients of anti-M-IgG and anti-H-IgG across M-IgG modified membranes. By using this correlation method a plot of permeability against solubility is expected to be linear, with a slope equals to the diffusion coefficient (Equation 3-19). Figure 3-13 shows the correlation plots of permeability and solubility based on the concentration dependence for the anti-H-IgG (Figure 3-13A) and the anti-M-IgG (Figure 3-13B) across M-IgG modified membranes. These plots in Figure 3-13 were carried out by using the experimental concentration-dependence data from Figure 3-6 and Figure 3-12 both for the 111

PAGE 112

permeability and solubility of anti-H-IgG and anti-M-IgG in M-IgG modified membranes. The plots in Figure 3-13 demonstrated a good correlation between permeability and solubility. The slopes of these correlation plots will provide the diffusion coefficients of the protein transported across the M-IgG modified membranes, if the diffusion coefficients are concentration-independent as the dual-mode model assumes. Figure 3-13.Correlation plots between permeability and solubility based on the concentration dependence for (A) anti-H-IgG and (B) anti-M-IgG across M-IgG modified membranes. However, the lines in Figure 3-13 were not linear in the total range of concentration, but instead they showed two straight-line slope segments. The segment with negative slope corresponds to the region at very low feed concentration. As mentioned above, in this region (low feed concentration) the Langmuir sorption sites contributed the most to the solubility of these proteins as was demonstrated in Figure 3-12. Therefore, the lower diffusion coefficients at low concentration are attributed to the presence of slow-moving adsorbed proteins in this Langmuir sorption mode. A more detailed analysis of this conclusion will be demonstrated below in the discussion of fitting experimental data with the theoretical dual-mode model. On the other hand, the segment with positive slope was used to estimate the diffusion coefficients. By fitting of the positive linear segments in 112

PAGE 113

Figures 3-13A and 3-13B, diffusion coefficients of (7.61 0.27) x 10 -8 cm 2 sec -1 and (1.02 0.06) x 10 -9 cm 2 sec -1 were found for anti-H-IgG and anti-M-IgG, respectively. These obtained values compare favorably with other IgG diffusion coefficients found in the literature using different porous materials. 225 229 A lower diffusion coefficient of 4.09 x 10 -10 cm 2 sec -1 was reported for a bare nanoporous alumina with a pore size of 75 nm and membrane thickness of 70 nm. 228 A diffusion coefficient of 2.71 x 10 -9 cm 2 sec -1 across nanoporous silicon with a pore size of 49 nm and membrane thickness of 6 m was also reported. 229 It should be mentioned that the most general equation used to obtain a rough estimate of the diffusion coefficients in liquids is the Stokes-Einstein equation. 230 233 This equation is derived for a sphere with hydrodynamic radius R H moving in a fluid (Equation 3-22). HBRTkD60 (3-22) Where, k B is the Boltzmann constant, T the temperature, the solvent viscosity, and R H the hydrodynamic radius of the molecule, respectively. This Stokes-Einstein relationship described that the diffusion process is basically controlled by the size of the molecules when they are in free solution and in absence of any interactions with other molecules. However, it is known that not only the size but also the intermolecular interaction and the molecular shape affect the magnitude of D 0 234, 235 Therefore, we can concluded that the differences in the diffusion coefficients for the anti-H-IgG and anti-M-IgG across M-IgG modified membranes are mainly due to the changes in the molecular energy of the proteins as the molecular recognition interaction takes place. It should be pointed out that the estimated diffusion coefficients were significantly lower than the reported free diffusion coefficient for IgG in water (3.85 x 10 -7 cm 2 sec -1 ). 228, 229 The reasons for the decreased diffusion coefficients found in these studies compared 113

PAGE 114

with those in free solution are probably due to the intrinsic membrane resistance itself, together with the changes in molecular energy as was explained above. The above results show clearly that the correlation analysis between permeability and solubility is a direct method to estimate diffusion coefficients when equilibrium sorption and steady-state permeation data are experimentally available. In addition, the two straight-line slope segments observed through this correlation made also possible to distinguish the heterogeneity in the spatial orientation of the immobilized carrier in the membrane which demonstrated to have minor effects in the functional behavior of the membrane (Figure 3-4). Fitting Experimental Data with the Theoretical Dual-Mode Model Figure 3-14 compares the experimental and calculated concentration dependence of permeability for anti-M-IgG and anti-H-IgG across M-IgG modified membranes. Figure 3-14. Experimental and calculated obtained permeability versus feed concentration for anti-M-IgG () and anti-H-IgG () across M-IgG modified membrane. 114

PAGE 115

At first glance observing the plots in Figure 3-14, one can conclude that according to the dual-mode model there is not total immobilization on the Langmuir-mode sites for these two permeating proteins across M-IgG modified membranes. If total immobilization occurs in the Langmuir-mode sites (F=0) as the dual-mode model explain, the permeability become independent of the concentration. These studies demonstrated a high concentration dependence on permeability both for anti-H-IgG and anti-M-IgG. These also demonstrated that the M-IgG carriers in the membrane indeed enhance the permeability at low concentration for both of the permeating proteins as shown in Figure 3-14. It was found that the permeability enhancement for these two proteins was only up to a certain threshold of the binding affinity and appeared to depend strongly on the concentration. This enhancement in permeability at low concentration is caused by a contribution of the diffusion due to the Langmuir mode species. 204, 211 As a consequence of its lower binding affinity, the permeabilities for anti-H-IgG were higher than those for anti-M-IgG. At the beginning we set a hypothesis that the differences in permeation for anti-H-IgG and anti-M-IgG observed could be due to a sort of antifacilitated transport. This was initially considered based on our previous works, 9, 10 which showed that higher fluxes were always observed for the molecule that binds stronger to the carrier in the membrane. The mechanism for the facile kinetic conquered, in our previous works, was ascribed by the sequential binding and unbinding events of the transported permeating species with the carrier in the membrane. This behavior might be mathematically rationalized in a more detailed manner by the model described here. As indicated in Figure 3-14, the experimental and calculated data were in good agreement at concentrations higher than 2.0 x 10 -11 moles per cm 3 Nevertheless, at very low concentration 115

PAGE 116

that agreement was missing. The explanation for this behavior was that the dual-mode model assumes a constant diffusion coefficient, which in practice does not happen. Therefore, the model is not able to discriminate a different regimen for the concentration dependence on the diffusion coefficient as the case at low concentration observed here. As was said above, at low concentration the adsorption of these two proteins in M-IgG modified membranes is dominated basically for the binding sites with stronger binding affinity (Figure 3-12). Then, these Langmuir-mode sites impart higher sorption rates for the permeating proteins, slowing down the diffusion coefficients. Consequently this is also reflected in the permeability. Thus, at low concentration the experimental permeabilities were lower than the calculated permeabilities. This reduction in permeability is mostly due to a combined effect of both the lower diffusion coefficients, and concentration-dependence of diffusion coefficients at this lower region in concentration (Figure 3-14). Additional evidence for this conclusion was also demonstrated by observing the concentration dependence on the experimentally obtained diffusion coefficients in Figure 3-15. As can be seen from Figure 3-15, lower diffusion coefficients were obtained at very low concentration. Thus, these results demonstrate highly concentration-dependent diffusion coefficients at low concentration as a result of the two types of binding sites present in the M-IgG modified membranes (well-oriented and random). Lower permeabilities were also observed for anti-H-IgG at the low concentration region (Figure 3-14). This effect was explained taking into consideration the differences in diffusion coefficients between anti-H-IgG and anti-M-IgG. Due to anti-H-IgG diffuses faster and thereby reaches the binding sites inside the pores faster; the concentration of anti-H-IgG is initially greater in the membrane than that of anti-M-IgG. Thus, anti-H-IgG initially at low concentration 116

PAGE 117

binds more sites but those are displaced when the anti-M-IgG molecules catch up to them. Additional confirmation for this conclusion is obtained by comparing the binding capacities for these two proteins (Table 3-2). Figure 3-15.Experimental obtained diffusion coefficients versus feed concentration for (A) anti-H-IgG and (B) anti-M-IgG across M-IgG modified membrane. At the region of low limit concentration it was also observed that as the concentration increases the permeability increases as well (Figure 3-14). It was deduced that between the two set of binding sites (well-oriented and random) present in M-IgG membranes, there is a transition stage in concentration before the random binding sites with a higher binding capacity become in a positive cooperative binding sites. This transition involves a mutual cooperation of these two binding sites in the membrane. This effect was observed at concentrations lower than 2.0 x 10 -11 moles per cm 3 (Figure 3-14). These results are in favorable agreement with the positive cooperative contribution observed in the adsorption isotherms in Figure 3-8. Besides as mentioned previously, the Langmuir binding sites were the dominant factor in determining the total solubility of anti-H-IgG and ant-M-IgG in the M-IgG modified membranes. And regardless of the protein these Langmuir sorption sites contributed the most to the solubility at low concentration, as was demonstrated in Figure 3-12. 117

PAGE 118

The above results demonstrated an enhancement of permeability both for anti-H-IgG and anti-M-IgG across M-IgG modified membranes. It was also confirmed that the extent of this enhancement and the linear dynamic range for each system can be controlled by the strength of the binding affinity between permeant-protein and the antibody-carrier immobilized in the membrane. Thus, we have demonstrated that antibody-modified nanotube membranes can be used for the transport of proteins at a rate determined by the affinity between carrier antibody and transported protein. Hence, it demonstrated that it is possible to develop systems of antibody-functionalized membranes with different dynamic range and controlled rate of transport. The result described herein clearly demonstrated a suitable principle for the concept of controlled facilitated transport systems for proteins using antibody-modified nanotube membranes. Of course, each system will require its own development of an appropriate affinity for the carrier/protein complex, in order to reach the desired rate of diffusion. Since the useful dynamic range of separation with these antibody-functionalized membranes depends on the binding affinity, the gamma of affinity constants found with antibodies makes those as proper elements for this principle. Besides, with the current scientific development both in genetically engineered and cloned binding proteins, the system proposed here may also be extended to other system by using proteins different than antibodies as carriers. Additional research into redesigned proteins may eventually permit the production of binding proteins against any desired analyte, just as antibodies can be produced against any antigen. 236 The concept of binding proteins genetically fused as affinity tags to other proteins without the loss of their structural and functional characteristics has been already demonstrated. 237 It has been also demonstrated that a common 118

PAGE 119

way to tune the affinity of a binding protein is to perform rational mutations at the binding site 238 240 Thus, depending on the mutations performed in the binding pocket of binding proteins, mutants with different binding affinities can be obtained; 240 this approach would allow the development of membrane-based separation systems with diverse dynamic ranges. Moreover, if this tuning affinity can be integrated with a system of membranes where the number of carriers can be augmented in a way that the permeability of the desired permeating protein can be selectively facilitated with respect to other present proteins in the separation sample, a selective and amplified separation system of membranes can be developed. This can be accomplished by taking advantage of the enhancement in permeability at low concentration caused by a contribution of the diffusion due to the Langmuir mode species as was explained above. 204, 211 Thus, this system would permit to the large number of accessible binding sites to promote higher concentration gradients inside the membrane that would enhance the flux of the selected target proteins, which imply a notable key aspect in designing more effective membranes for membrane-based separation. Validation of the Dual-Mode Model The validity of the above results can be assessed by the plot of experimental obtained permeabilities against (1 + bC feed ) -1 From the slope of the straight line in this plot, the Langmuir-mode diffusion coefficient (D H ) can be evaluated and consequently the mobile fraction of the Langmuir species F. 211 The results for these data are shown and summarized in Figure 3-16 and Table 3-4, respectively. Figure 3-16 shows and compares the experimentally and calculated obtained plots of anti-H-IgG and anti-M-IgG permeabilities vs. (1 + bC feed ) -1 Taking into consideration the cooperation of the two binding sites at low concentration, the examination of the linear relationship of these plots demonstrates that the concentration dependence of the permeability for these two proteins 119

PAGE 120

agrees fairly well with the dual-mode model (Equation 3-17). It worth mentioning, that this validation method requires the use of only one binding affinity. Therefore, in the case of anti-M-IgG that has two binding sites in the membrane; an average affinity was defined as being the sum of the free energy changes that occur in the first and second binding site. Thus, in order to keep the consistency with the conventional units the average binding affinity constant was given by the square root of the product between the stronger and weaker binding constants (b = 1.53 x 10 11 cm 3 moles -1 ). It should be pointed out that comparable results were also obtained for the experimental concentration dependence on permeability for anti-M-IgG when they were fitted using this nominal average binding affinity constant. Figure 3-16. Plots for experimental and calculated obtained permeability against (1 + bC feed ) -1 for (A) anti-H-IgG, and (B) anti-M-IgG across M-IgG modified membrane. The above results demonstrated that the permeation mechanism for anti-H-IgG and anti-M-IgG across M-IgG modified membranes obeys the dual-mode model proposed in these studies. The differences in dual-mode model parameters for these two proteins suggested the existence of a distribution of binding sites inside of these antibody-modified nanotube membranes that are accessed for each permeating protein in a particular manner. In fact the permeability plots in Figure 3-14 and Figure 3-16 demonstrated the differentiation of specific transport mechanisms for each one of the permeating proteins across the M-IgG modified membranes. The higher 120

PAGE 121

permeability for anti-H-IgG, and the higher concentration-dependence in permeability for the anti-M-IgG are clearly evidences that the permeation mechanisms across M-IgG modified membranes for each one of these proteins are different and highly depending on the affinity interactions with the immobilized carrier in the membrane. The closeness of the two obtained diffusion coefficients D D and D H also confirmed the partial immobilization of the Langmuir species (F=1) on the Langmuir sorption mode (Table 3-4). This also confirms the contribution of the Langmuir species were the responsible factor of the enhancement of permeability observed in Figure 3-14. Table 3-4.Validation of the dual-mode model parameters Permeating Protein D D (cm 2 /sec) D H (cm 2 /sec) F anti-H-IgG 7.61 x 10 -8 8.11 x 10 -8 1.06 anti-M-IgG 1.02 x 10 -9 1.02 x 10 -9 1.00 Conclusions With this work, we have demonstrated the dual-mode model proposes here is suitable model for the characterization of the transport properties of anti-H-IgG and anti-M-IgG in M-IgG modified membranes. The model demonstrated a good agreement with the experimental system studied here. Besides, the findings showed in these studies allow addressing the concept of modulated facilitated transport systems using antibody-modified nanotube membranes, and how the extent of the facilitation is determined and controlled by the binding affinity between carrier and permeating protein molecules. These studies can provide a new direction in the development of more effective membranes for membrane-based separation. 121

PAGE 122

CHAPTER 4 METAL ION AFFINITY NANOTUBE MEMBRANES: PREPARATION, CHARACTERIZATION, AND CONTROLLED TRANSPORT OF SIX-HISTIDINE-TAGGED RECOMBINANT PROTEIN Introduction Since the introduction of Immobilized Metal Ion Affinity (IMAC) by Porath and coworkers 126 in 1975, IMAC has become in one of the most widely used methods for protein purification. 241 250 IMAC exploits the fact that many transition metal ions, such as, copper, nickel, cobalt and zinc, can be coordinated to amino acid sites exposed on the surface of the proteins. It has also been proposed that metal ions in IMAC can be classified into three groups: hard, intermediate and soft. 251, 252 This classification has been based on the preferred reactivity of the metal ions toward their nucleophiles. Thus, metal ions such as Fe +3 Ca +2 and Al +3 which show preference for oxygen belongs to the hard metal ion group. Those ions that prefer sulfur (Cu + Hg +2 Ag + etc.) correspond to soft metal ions. Then, the Cu +2 Ni +2 Co +2 and Zn +2 ions, which coordinate nitrogen, oxygen and sulfur, compose the intermediate ions. Given that the number of cysteine residues on the surface of proteins is limited, makes the histidine residues the main targets for intermediate metal ions. For example, a binding constant of 4.5 x 10 3 M -1 has been determined for an average protein with a single histidyl residue. 253 Thus, multipoint histidyl interactions can be used to control the strength of binding of a protein in IMAC applications. The advanced understanding of which aminoacid site chains on the surface of proteins are responsible of the interaction with the metal ions, has been one of the most important contributions for the development of tagged peptides containing multiple histidine residues. 254, 255 In particular, the hexahistidine tags attached to the 5 or 3 end of the target gene that encodes the protein are usually used as purification tags in recombinant proteins. 256, 257 122

PAGE 123

Today, recombinant protein purification represents one of the principal protein purification markets, and many systems have been developed in recent years for the purification of these proteins. 241 250, 257 The system proposed here, exploits the advantages of IMAC separations, but in a nanopore-membraneplatform. The use of imidazole-modified silica nanotube membranes for the controlled transport of histidine-tagged recombinant proteins is described. Imidazole is a functional aromatic nitrogen heterocycle important in ligand coordination chemistry. The structure of imidazole is a five-membered ring with two different nitrogen atoms. One of the nitrogen is bound to hydrogen, having two lone pair of electrons in an unhybridized p-orbital donated to the aromatic ring, and it is structurally related with the pyrrole nitrogen. The second nitrogen structurally related with the pyridine nitrogen, donates its single electron in an unhybridized p-orbital to the ring, containing also a free lone pair of electrons in a hybrid orbital. 258 In aqueous solution imidazole behaves like a weak organic base with pK a = 6.99. 259 Therefore, the lone pair of electrons in the pyridine-like nitrogen atom, are only involved in coordination when the pH of aqueous solutions is close or higher than 6.99. It is worth mentioning, that the mode of coordination of imidazole can be also influenced by the nature of the metal. In this investigation, the silica nanotube membrane was prepared by sol-gel template synthesis of silica nanotubes within the pores of a nanopore alumina template. Imidazole was attached to the inside walls of chloropropyl-modified silica nanotubes by nucleophilic substitution reaction with 2-mercaptoimidazole in anhydrous tetrahydrofuran. The modified membrane was then exposed to solutions of the desired divalent metal ion (copper, nickel, cobalt or zinc) to attach the metal via coordination with the imidazole. The histidine-tag on the 123

PAGE 124

recombinant protein reversibly coordinates with the metal ion sites, and via this chemistry the flux of the protein across the membrane could be modulated. In these studies the concept of IMAC separations was exploited in a nanopore-membrane-platform. The transport of histidine-tagged recombinant proteins in metal ion affinity nanotube membranes is described. It was demonstrated that transport of histidine-tagged recombinant proteins across imidazole-modified silica nanotube membranes can be controlled by the nature of the immobilized metal ions on the imidazole surface. The imidazole functionalized membrane was extensively characterized using three different analytical techniques in order to identify how the coordination chemistry with different divalent metal ions affects the functional behavior and transport properties during the permeation of six-histidine recombinant proteins. Furthermore, additional results shown that by the appropriate choice of the transition metal ion for coordination with the imidazole ligand, the binding strength and optical properties within the functional membrane can also be modified. These studies demonstrated an alternative way of immobilized metal ion affinity by using nanotube membranes. Experimental Materials Commercially available nanopore alumina membranes (60 m thick, nominal pore diameter 100 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999% (TEOS) (Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes. A 3-chloropropyltrimethoxysilane (United Chemical Technologies) was used for silane modification of the silica nanotubes. 2-mercaptoimidazole from Sigma Aldrich was used as attacking nucleophile to substitute the chlorine atoms in the 3-chloropropyl silica nanotube functionalized membrane. As an aprotic solvent anhydrous tetrahydrofuran >99.9%, inhibitor free from Sigma Aldrich was used. Divalent salts of copper, nickel, cobalt and zinc nitrate 124

PAGE 125

obtained from Fulka, were used to attach the metal via coordination with the imidazole. A six-histidine-tagged ubiquitin human recombinant protein (6His-UHRP) obtained from Sigma Aldrich was used to investigate the transport properties of the imidazole-modified membranes. All other chemicals were of reagent grade and used as received. Purified water, obtained by passing house-distilled water through a Barnstead, E-pure water-purification system, was used to prepare all solutions. Preparation of the Silica Nanotube Membranes Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-based template-synthesis method described previously. Briefly, a sol-gel precursor solution was prepared by mixing absolute ethanol, TEOS and 1 M HCl (50:5:1 by volume), and this solution was allowed to hydrolyze for 30 minutes. The alumina membranes were then immersed into this solution for two minutes under sonication. The membranes were then removed and the surfaces swabbed with ethanol to remove precursor from the alumina surface. The membranes were then air dry for ten minutes at room temperature and cured in an oven for 12 hours at 150 C. This yields silica nanotubes, with wall thickness of ~3 nm, embedded within the pores of the alumina membrane. 9 Preparation of Imidazole Functionalized Membranes The 3-chloropropyltrimethoxysilane was first attached to the inside walls of the silica nanotubes. A solution that was 2% (v/v) in 3-chloropropyltrimethoxysilane was prepared in ethanol; to this solution was added acetate buffer (50 mM, pH=5.1) to make the solution 5% (v/v) in this buffer. This ethanol-based solution was stirred for 20 minutes and the alumina membranes were then immersed. The soak membranes were sonicated for 2 minutes, and after 20 minutes the membranes were removed from the solution and rinsed with ethanol. To make sure that the loose silane was removed from the surface, the membranes were also sonicated for 125

PAGE 126

2 minute in ethanol. The membranes were then dry with nitrogen and placed under vacuum for 12 hours. The resulting 3-chloropropyl silica nanotube functionalized membranes were then immersed into different vials containing a solution that was 0.5 M in 2-mercaptoimidazole dissolved in anhydrous THF at 45 C. This reaction was kept at this temperature for a period of 1 hour, with periodically additions of THF to continually replace the evaporated THF. Then, the lids of the vials were tightly closed and they were wrapped with paraffin film, and the membranes were incubated in the dark for 24 hours at 25 C to allow the completion of the reaction. After imidazole functionalization, the membranes were rinsed copiously with ethanol and sonicated for 2 minutes to remove the unreacted mercaptoimidazole. A schematic illustration of the reaction mechanism for the preparation of imidazole functionalized membranes is shown in Figure 4-1. Figure 4-1.Reaction mechanism of imidazole functionalized nanotube membranes. Divalent Metal Ions Complexation The imidazole modified membranes were then exposed to solutions of the desired divalent metal ion (copper, nickel, cobalt or zinc) to attach the metal ion via coordination with the 126

PAGE 127

imidazole. Briefly, the imidazole functionalized membranes were immersed in different vials containing solutions prepared in ethanol that were 1 mM in Cu +2 Ni +2 Co +2 and Zn +2 respectively. The vials were placed in an orbital shaker platform and the membranes were equilibrated for 12 hours at 25 C to affirm the complexation of the metal ions on the membrane. After equilibration the membranes were removed and rinsed copiously with ethanol. Instrumental Characterization of the Imidazole Functionalized Membranes To characterize the imidazole functionalized nanotube membranes fluorescence microscopy, Fourier transformed infrared attenuated total reflectance spectroscopy (FTIR-ATR), and X-ray photoelectron spectroscopy (XPS) measurement were taken. We discovered that the imidazole membranes are fluorescent after functionalization. Therefore, the observation of these imidazole functionalized membranes under the fluorescence microscope was used as the first control to confirm the success of the chemistry modification. The fluorescence intensity was measured using a fluorescence microscope which was equipped with both a fluorescence detector and CCD camera. This system combines an Axioplan 2 imaging microscope (Zeiss) with a J&M-PMT photometry system detector (SpectrAlliance), for measuring the fluorescence intensity using a BP450-490 filter for excitation and a LP515 filter for emission of light. The FTIR-ATR was employed to examine the presence of characteristic bands for 2-mercaptoimidazole in the functionalized membranes. The FTIR-ATR spectra were recorded on a Thermo Nicolet Nexus 670 FTIR spectrometer equipped with a GATRTM as the ATR crystal. All spectra were run at 4000cm-1 to 400cm-1, 512 scans, 4 cm-1 resolution, and gain equal to 1. The spectrometer aperture was set to 100 for all runs. A background spectrum (air) was taken prior to each sample spectrum, which was used for background subtraction using the OMNIC ATR software for the ATR correction. 127

PAGE 128

The surface composition of imidazole functionalized membranes and the presence of the immobilized transition metal ions were also determined through XPS analysis. The analysis was performed using a XPS/ESCA Perkin-Elmer PHI 5100 ESCA system. For the XPS measurement a Mg K (1253.6 eV) or Al K (1486.6 eV) monochromatized X-ray source was used to stimulate photoemission. Survey and high resolution spectra were obtained with a hemispherical energy analyzer operated in the constant analyzer transmission mode to measure the binding energies of emitted photoelectrons. Transport Experiments The imidazole-modified membranes were sandwiched between two pieces of Scotch tape that had 0.13 cm 2 area holes punched through them. These holes defined the area of membrane exposed to the contacting solution phases. This membrane assembly was mounted between the two halves of a U-tube permeation cell, with half-cell volumes of 2.5 mL. The feed solution for the recombinant protein experiments was 1.87 M dissolved in 50 mM phosphate-buffered saline pH 8.0 that also contained 300 mM sodium chloride and 2 mM imidazole. The receiver solution on the other side of the membrane was identical except it initially contained no protein. The concentrations of the transporting protein in the feed solution were in all cases the same. It is important to mention that prior to starting each transport experiment the metal-immobilized imidazole membranes were equilibrated in the PBS buffer solution for two hours, and that the metal ion immobilization for each divalent cation was completed right before this equilibration. The rate of protein transport across the membrane was determined by periodically checking the absorbance at 276 nm in the receiver solution using a flow-through UV-Vis detector (Figure 4-2). 128

PAGE 129

Figure 4-2.Flow-through UV-Vis detection system used to measure the flux of six-histidine-tagged ubiquitin human recombinant protein. This system consisted of a L-7455 UV-Vis detector (Hitachi), where the flow stream was delivered using an L-7100 HPLC pump (Hitachi) using a flow rate of 3 mL per min. The whole system was controlled using Chromeleon 6.50 chromatography management software (Dionex) through a UCI-100 universal chromatography interface (Dionex). Calibration curve was obtained with the same flow through system on solutions containing known concentrations of the transport recombinant protein. This calibration curve was used to obtain the concentrations of the transport proteins in the receiver solution. Results and Discussion Imidazole Surface Modification and Characterization The imidazole functionalized nanotube membranes were prepared by covalent immobilization of imidazole inside of the silica nanotubes. This coupling between silica nanotube surface and imidazole was accomplished via simple silane chemistry using 3-chloropropyltrimethoxysilane. Silica surfaces offer the positive attribute that they can be derivatized with an enormous variety of chemical functional groups using simple silane chemistry 95 Here, the covalent immobilization of imidazole was achieved by nucleophilic substitution of chlorine atoms in the resulting silane modified 3-chloropropyl silica nanotube 129

PAGE 130

membranes. It has been demonstrated that mercapto compounds are appropriated modification reagents for this strategy. 106, 260 263 Furthermore, it has been confirmed that the use of aromatic thiols leads to degrees of modification which are much higher than those achieved with aliphatic ones. In addition, when aromatic thiols are used the possibilities of elimination as a side reaction are eradicated. 261 In order to promote the substitution nucleophilic bimolecular mechanism (S N 2) the imidazole modification was carried out in an aprotic solvent (THF). Since aprotic solvents do not solvate the nucleophiles, the reactivity of the 2-mercaptoimidazole can be enhanced making it more accessible to approach to the primary halide carbon as the conditions in 3-chloropropyl silica nanotube modified membranes. A high concentration of 2-mercaptoimidazole was also used to promote the S N 2. It is worth mentioning that some indications that the reaction is taking place were the sulfur smell during the reaction and the lower pH of the solution at the end of it. Fluorescence microscopy characterization We found that after imidazole immobilization the nanotube membranes were fluorescent, and this observation become in our first control to monitor the success of the modification chemistry. Figure 4-3.Fluorescence spectra of (a) silica nanotube and (b) imidazole functionalized membranes. 130

PAGE 131

Figure 4-3 shows the fluorescence spectra of the nanotube membranes before and after imidazole functionalization. It is clearly seen from these data that the fluorescence intensity for the imidazole functionalized membrane was nine-fold higher compared to the bare silica nanotube membranes. This fluorescence signal was attributed to the imidazole moiety in the functional membrane. The origin of this fluorescence signal was also confirmed by conducting fluorescence experiments of pure 2-mercaptoimidazole in methanol. Additionally, we found that by the choice of the appropriate transition metal ion for the complexation with the imidazole, the optical properties within the functional membrane can be modified (Figures 4-4 and 4-5). Figure 4-4.Fluorescence spectra of imidazole functionalized membranes before (upper) and after (lower) metal ion immobilization: (A) Co +2 (B) Cu +2 (C) Ni +2 and (D) Zn +2 Figure 4-4 shows the fluorescence spectra for a set of four imidazole functionalized membranes that were exposed respectively to four different nitrate salts solutions of Cu +2 Ni +2 Co +2 and Zn +2 (vide infra). It was observed that after immobilization of these transition metal ions the fluorescence intensity of the functional membranes was quenched. Figure 4-5 reports the fluorescence quenching ratio F/F 0 where F 0 is the fluorescence of imidazole functionalized membrane and F is the corresponding fluorescence in the presence of 131

PAGE 132

the metal ions. It is evident that fluorescence quenching was much higher for Cu +2 and Co +2 ions, while it was weaker in the case of Ni +2 and very low for Zn +2 Figure 4-5.Fluorescence quenching (F/F 0 ) for imidazole functionalized membranes in the presence of nitrate solutions prepared in ethanol that were 1 mM of the respective divalent metal ion. Fluorescence quenching by transition metal ions is a common phenomenon. 264 The fluorescence quenching by transition-metal ions and their coordination complexes is a process which has been known for a long time and is still the subject of intense investigations. 265 267 It is now well recognized that electron transfer and electronic energy transfer are considered as the two main deactivation pathways responsible for efficient fluorescence inhibition. 264 267 These metallosystems based on fluorescence detection have found applications in many areas such as metal ion sensing, 268, 269 artificial photosynthesis, 270 and optical sensing of DNA. 271 273 On the other hand, when the imidazole functionalized membrane was exposed to solutions of CuCl 2 instead of Cu(NO 3 ) 2 the optical behavior for this metal ion immobilized membrane was different than in the case of nitrate salts. Under this condition, a fluorescence enhancement instead of quenching was observed in the presence of the metal ion (Figure 4-6). 132

PAGE 133

Figure 4-6 compares the fluorescence spectra for imidazole functionalized membranes before and after coordination with CuCl 2 Simultaneously with the fluorescence enhancement, a small bathochromic shift of the maximum emission wavelength was observed in the fluorescence spectra of the copper immobilized membrane from 443 nm to 550 nm. Figure 4-6. Fluorescence spectra of imidazole functionalized membranes (a) before and (b) after metal ion immobilization using an ethanol solution that was 1 mM in CuCl 2 Besides, a dependence on the copper concentration for the fluorescence enhancement was found (Figure 4-7). As can be seen from Figure 4-7 the addition of the chloride salt of copper leads to an increase in fluorescence intensity of the functional membrane and at a certain concentration the fluorescence intensity remains constant. Such an increase in fluorescence enhancement can result from conformational restriction induced upon binding. 274 It is beyond the scope of the present work to discuss in detail the mechanism of interaction of the functional imidazole membrane with these metal ions. Nevertheless, the observed fluorescence intensity changes can be reasonably explained on the basis of the complex structure formed after coordination takes place. It was ascribed that for the situation of the imidazole membranes coordinated with the nitrate salts, the nitrate anions in [M(Im) n ](NO 3 ) 2 complexes are strictly counterions, hence they are no part of the sphere of coordination. Therefore, the 133

PAGE 134

fluorescence quenching through an electron transfer mechanism can be accredited. 264, 265 The great -acceptor properties of the imidazole permit it to accept electronic charge from d orbitals on the metal ion. 258 These charge transfer transitions correspond roughly to the transfer of an electron from the metal ion to a ligand. More precisely, an electron is transferred from a bonding orbital of mostly metal ion character to an empty antibonding orbital of mostly ligand character. 275, 276 Figure 4-7. Copper concentration dependence on fluorescence enhancement for imidazole functionalized membranes. Conversely, in the situation using chloride salts, the chloride anions can be part of the sphere of coordination. Therefore, in the [M(Im) n (Cl) 2 ] complex both the imidazole and chloride ligands can be coordinated to the metal ion. Furthermore, for that reason that chloride is known as a -trans-directing ligand, the enhancement in fluorescence can be ascribed to a -trans-influence, so that a trans--acceptor ligand as chloride, can form bonds with the metal and withdraw electron density away from the other imidazole ligand. 275, 276 Then, the -trans bonding have a cooperative effect allowing more delocalization of the electrons. 277 The geometrical restraint required for having this interaction, support that fact that the increase in fluorescence enhancement can result from conformational restriction induced upon binding. 269, 274 McFarland and Finney269, 274 demonstrated that metal ion binding could restrict the rotational excited-state of 134

PAGE 135

a fluorophore suppressing some of the fluorescence deactivation pathways, producing fluorescence enhancement. Thus, those changes in fluorescence intensity showed how the optical properties of the functied tion by FTIR-ATR spectroscopy ube membrane with the imidazole moiety was aa nce. To he differences observed in the spectra of bare alumina and imidazole-modified memb 2-n the is known that hydrogen bonds with water molecules contribute to broaden peak areas.280, 281 onal imidazole membranes changes depending on the transition metal ion immobilized. Besides, these results confirm the accomplishment in the preparation of imidazole functionalizmembranes. Characteriza The successful modification of the silica nanot lso demonstrated by FTIR-ATR spectroscopy. Figure 4-8 compares the FTIR-ATR spectrfor bare alumina and imidazole functionalized membranes. It should be pointed out that all spectra were corrected for spectral distortion using the OMNIC ATR software. However, byusing this ATR correction it is possible to lose spectral information of the fundamental vibrational modes of a molecule, overall when the entire spectral region is examined at oavoid this, it is a common method to analyze the data in limited spectral regions in order to give a more detailed assignment of the characteristic bands. The Figure 4-8B illustrates this issue, anda narrow spectral region, from 2000 to 600 cm -1 for the imidazole functionalized membrane is used. T ranes clearly demonstrated the modification of the nanotube membranes by usingmercaptoimidazole. For the FTIR-ATR spectrum of bare alumina two broad bands appear i3900-3200 cm -1 and 3000-1600 cm -1 regions, which are attributed mainly to the O-H stretching modes and O-H deformation modes, respectively. 278, 279 Although, for these two broad bands the contribution of water (at about 3600 cm -1 and about 1600 cm -1 ) also has to be considered, since it 135

PAGE 136

Figure 4-8. (A) FTIR-ATR spectra of bare alumina and imidazole functionalized membranes. (B) FTIR-ATR spectra of imidazole functionalized membranes in a narrow spectral region illustrating the effect of using two different backgrounds for the spectrum subtraction: (a) air, and (b) 3-chloropropyl silica nanotube membrane. For tn of their freque in KBr (FiguFigure 4-9.TransmThe FTIR-ATR spectrurane showed clearly three tion modes for C=C he case of imidazole membranes, the observed bands were assigned by comparisoencies with the transmission FTIR spectrum of pure 2-mercaptoimidazol re 4-9). 263, 282 285 ission FTIR spectrum of pure 2-mercaptoimidazole in KBr. m of the imidazole functionalized memb of the characteristic vibrations for 2-mercaptoimidazole. One was the ring vibra and C=N bonds in the region ~1684 1630 cm -1 A second band, observed at ~1486 1400 cm -1 region that was assigned to the thioamide bands I and II, which are coupled vibrations 136

PAGE 137

involving stretching motion of C-S + N-H and bending motion C-N + N-H, respectively. 286 The third band was the very strong band at 2890 cm -1 assigned to N-H stretching mode. This imidazole N-H stretching appeared as a broad band in a wide range from ~3400 to 2200 cm 1 When the characteristic band of N-H was compared with the band of pure 2-mercaptoimishift from 3145 cm dazole a zed -1 to 2890 cm -1 was observed. This shifting might be due to the solid state characteristic of the sample, but also to the formation of hydrogen bonds between imidazole moieties. All the assigned bands were also observed in metal-immobilized imidazole functionalized membranes, and some of them were enhanced by the presence of the immobilimetal ion as is shown in Figure 4-10. Figure 4-10.FTIR-ATR spectra of imidazole functionalized membranes showing the effect of the metal ion immobilized on the thioamide bands: (a) Zn+2, (b) Co+2, and (c) Cu+2. Figure 4-10 compares the results for three different imidazole functionalized membranes which werely. It was for e exposed to solutions of divalent metal ions of copper, cobalt and zinc, respectiv found that the enhancement of the band intensity is related to the strength of the binding constant between the metal ions with the functional membrane. 258 In addition, it was observed that those bands shifted toward lower frequencies as is demonstrated in Figure 4-10. These changes of the band intensity may be attributed to changes of the electric dipole moment after coordination with the metal ions. 287 It may be possible that IR energy is preferential absorbs 137

PAGE 138

some favored orientations of the imidazole moiety in the functional membrane after coordination. The observed band at 1110 cm was also assigned, and it is clearly illustrated i8B. This band n Figure 4-was labeled to the asymmetric C-S-C stretching vibrations.284, 288 For the correct assignne embrane for backgnal mode 8 cm-1 branes Frequency (cm-1) ment of this band, the contribution of Si-O stretching vibration at this region was also considered. In order to confer a proper assignment, the FTIR-ATR spectrum of imidazole functionalized membrane was also acquired by using 3-chloropropyl silica nanotube membraas background for the spectrum subtraction instead of air (Figure 4-8B). Figure 4-8B compares the FTIR-ATR spectra of an imidazole functionalized membrane, which were collected using both air and 3-chloropropyl silica nanotube m round subtraction, respectively. It was clearly observed that when the chlorosilane-modified membrane was used for the background subtraction, the fundamental vibratiofrom the imidazole moiety at 1118 cm -1 was more clearly discriminated. Therefore, at 111the contribution of both Si-O and asymmetric C-S-C stretching vibrations should be considered. 284, 288, 289 Table 4-1.FTIR-ATR characteristic bands of imidazole functionalized nanotube memAssignment N-H stretching ~2900 N-H bending in the plane ing vibration modes ~1684-1630 which are respectively coupled vibrations n of C-S + N-H, and bending motion C-N + N-H. ~1486 and 1412 hing vibrations. 1118 ~1560 C=C and C=N r Thioamide bands I and II, involving stretching motio C-S-C anti-symmetric and Si-O stretc 138

PAGE 139

139 Table 4-1 shows the assignment of the characteristic bands for the imidazole functionalized embranes. In some cases the overlapping between bands made this assignment difficult. This was particularly more apparent in the region for N-H bending vibration and the coupled vibrations involving C-S and N-H of the thioamide band I. Transmission FTIR spectra for the alumina template, silica nanotube, and nanotube modified membrane films were also studied. Figure 4-11 shows the transmission FTIR spectra of the alumina template film for each of the modification steps illustrated in Figure 4-1. Figure 4-11. Transm for each of the modificationplate, (b) silica, (c) By ovariations ofith each chemistry modification. However, it was possible to distinguish through the different modification steps by monitoring the changes in transmittance in each transmission FTIR spectrum. It was clearly observed that the absorption of stretching vibrations band for the O-H groups in the region between 3000 and 3600 cm changes throughout the different modification steps. Likewise, differences in the region 1500 and 1600 cm-1 for the bending vibration of O-H were m 263 ission FTIR spectra of the alumina template film steps illustrated in Figure 4-1: (a) bare alumina tem3-chloropropyl silane modified and (d) imidazole functionalized nanotube membranes, respectively. bserving the transmission FTIR spectra in Figure 4-11, there were no obvious the characteristic bands related w -1 139

PAGE 140

also od rmed the success of the silanization reactied e C-S sf om characteristic of its cohilic the infrared absorption spectral region. Thus, the shorter wavelengths can penetrate less the bserved. In the latter region, the differences between the bending vibrations of the O-H for the alumina membrane and the silica nanotube membrane were not significant (spectra a, and b in Figure 4-11). This was reasonable because part of the OH groups for the alumina are replacefor the silanols OH groups in silica; besides silica surface introduce higher silanol groups on surface for further silane functionalization. 30, 94, 99 102 In the case of chlorosilane modification (spectrum c in figure 4-11), lower absorption for the silane-modified membrane was observed in the bending O-H region compared with alumina and silica. This difference in transmittance clearly confi on where the O-H free groups in the alumina matrix and the silanol groups in the silica nanotubes were replaced by the condensation reaction with 3-chloropropyltrimethoxysilane. The differences in transmittance were even more evident for the imidazole functionalizmembrane (spectrum d in Figure 4-11). Additionally this exhibited two observable absorptionbands at 1118 cm -1 and 474 cm -1 assigned respectively to the anti-symmetric C-S-C, and to th tretching vibrations, which confirmed the imidazole functionalization. These differences in the intensity of the bands might be attributed once more to changes othe electric dipole moment after each chemistry modification. 287 Besides, each chemistry modification introduces a variety of vibrational motions, and degrees of freed mponent atoms that should produce a net change in the dipole moment. It should be also considered the fact that may be after the modifications the membranes become less hydropthan the bare alumina template, thereby the contribution of water on surface must be less significant, even though all the membranes were dried at the same conditions. These transmission FTIR data in Figure 4-11 may perhaps be interpreted on the basis of 140

PAGE 141

alumina matrix and transmit more IR energy through the porous structure of the membraninduce vibrational excitation of the covalently bonded atoms on the nanotube su e to rface. On the other re metal azole membranes. This analysis begins, however, with the XPS rane (Figure 4-12). Figure 4-12.The survey sped analysis showed the presposition peaks for Al, O, P, and C. The the surface; this hand, longer wavelengths should have deeper penetration on the alumina matrix, allowingless transmission of the IR energy. Therefore, the latter region at the same time should be mosensitive to changes associated with electric dipole moment, and more information may be obtained in this region. Surface chemical composition by XPS XPS analysis was performed not only to assure the presence of chlorine and imidazole groups in the modified membranes, but also to confirm the immobilization of the transition ions on the functionalized imid survey studies for the bare alumina memb XPS survey spectrum for the bare alumina template. ectrum for the alumina template is shown in Figure 4-12. The performence of elemental chemical comunexpected peak for C1s at 289.3 eV was attributed mainly to impurities on lained below in more detail. carbon impurity is exp 141

PAGE 142

Survey scans were taken to determine the elemental surface composition of various elements present in the chlorosilane and imidazole modified membranes. Figure 4-13 compares the XPS survey scan of the chlorosilane and imidazole modified membranes. XPS survey spectra for (a) chlorosilane and (b) imidazole modified modified surface shown the presence of XPS line for china, which confirmed the accompmodification of the silica nanotubes (spectrum a Figure 4-13). In addition, Figure 4-13.embranes. The chlorosilane mlorine Cl2p, which was absent in the bare alumlishment of the chlorosilane this survey scan d bare mlane e XPS measurements for the imidazole functionalized membranes suggested the pr shown distinct peaks for Al2p as well as for C1s compared to the imidazole functionalized an embranes. It was noticed that after functionalization with 2-mercaptoimidazole, an increase in C1s peak and a diminution in the Al2p peak was observed when it is compared with the chlorosimembrane (spectrum b Figure 4-13). Considering the surface sensitivity of XPS, the decrease ofthe Al2p peak in th esence of imidazole on the nanotube membrane surface. 136 Additionally, the functionalization of the membranes with the imidazole moiety was confirmed for the presence of N1s and S1s lines in the XPS spectra. 142

PAGE 143

The chemical composition and binding energies data from such spectra are summarized in Table 4-2. It is worth mentioning that owing to the insulator characteristics of the aidentification of the specific peaks was demanding. Moreover, due to the surface charging a positive shift toward higher binding en lumina, the ergy was observed for some peaks. Tablemodified, and imidazole functionalized membranes 4-2.XPS surface composition and binding energies for bare alumina, chlorosilane Bare Alumina Chlorosilane Modified Imidazole Functionalized XPS Signal % BE(eV) % BE(eV) % BE(eV) Na 1s 4.4 1075.6 1 .0 1074.5 O 1s 60.3 536.2 52.5 534.5 49.8 534.0 N 1s 3.9 9.3 88.0 201.8 104.5 402.0 C 1s 7.4 28 11.9 2 .4 287.0 Cl 2p 1.4 203.0 S 2p 1.5 171.0 P 2p 2.8 138.4 2.3 137.0 2.1 136.0 Si 2p 1.4 104.0 Al 2p 29.5 78.4 25.7 77.5 19.9 76.5 Furthermore, the XPS multiplexes analysis aigh resoluor the izole mes (Figure 4-14) confirmed the imidazctionaln. An that was based only on the relative areas of the XPS peaks at hition demonstrated an elemental compositional ratio f 6:2:1 for C: N: S. However, after the individual peak areas were divided by their respective atomi t h tion f mida embran ole fun izatio analysis gh resolu o c sensitivity factor (ASF) the elemental compositional ratio for C: N: S respectively was 14:2:1. This increase in C signal was attributed to impurities on the surface of the alumina membranes. These carbon impurities were corroborated by observing the high-resolution scan of the C1s peak (Figure 4-15). 143

PAGE 144

Figure 4-14.XPS multiplexes analysis at high resolution for the imidazole functionalized membrane. Figure 4-15 compares the high-resolution C1s peaks for the bare alumina membrane, the chlorosilane modified membrane and the imidazole functionalized membrane, respectively. plate, (b) chlorosmbranes consisted both of energy binding where thnd, the C1s peak for which may b Figure 4-15.High-resolution XPS data of the C1s peak for (a) bare alumina temilane modified, and (c) imidazole functionalized membranes. It was clearly seen that the C1s peak for Imidazole modified mea well defined peak at 287 eV which was ascribed to the C-C peak and a shoulder at higher e C-N, C-S peaks are overlapped. On the other ha chlorosilane modified membranes revealed a better indication of a second peak at 291 eV e assigned to the C-Cl environment. The high-resolution scan of C1s peak for the bare 144

PAGE 145

alumina membrane consisted of only one peak the C-C peak at 287 eV, which confirmed the presence of carbon impurities in the membrane. The immobilization of the transition metal ions in the imidazole functionalized membranwas also verified by XPS analysis. Figure 4-16 compares the XPS survey scan of two imidazofunctionalized membranes after immobilization o es le f Cu and Zn, respectively. For the XPS survey scan oFigure 4-16. Xmobilization of (a) Cu+2spectrum of the impresence of Zn2p1 at 1050 eV and Zn2p3 at 1026.5 eV peaks. Whereas the survey for Cu shown the prhould f Cu and Zn metal-immobilized nanotube membranes, the Mg anode and the Al anode were used respectively to stimulate photoemission. PS survey spectra for imidazole modified membranes after im, and (b) Zn+2. The presence of XPS spectral lines for Zn and Cu were observed in the respective idazole functionalized membranes. The XPS survey scan for Zn shown the esence of Cu2p1 and Cu2p3 peaks assigned at 954 eV and 934 eV, respectively. It sbe mentioned that analogous experiments were also carried out for the immobilization of Ni andCo, and Auger lines for all these four metal ions were also observed. Thus, these data confirmed the coordination and immobilization of the transition metal ions used in these studies with the imidazole functionalized membranes. 145

PAGE 146

Transport Properties of Imidazole Membranes The transport properties of the imidazole functionalized membranes were explored by monitoring the flux of six-histidine tagged ubiquitin human recombinant protein (6His-UHRP) in function of time. For the transport experiments, the rate of 6His-UHRP across the membrane was determthe he absorption maximum occurs in 276 nm. Figure 4-17.UV-Vis spectrummM phosphate-buffered loride and 2 mM imidazole. Calibration curve at 276 nm showed in Figure used to obtain the concentration of the transport protein in the receiver solution. Figure 4-18 show ined by periodically checking the absorbance at 276 nm in the receiver solution using flow-through UV-Vis detector illustrated in Figure 4-2. Figure 4-17 shows the absorption spectrum of 50 M solution of 6His-UHRP that was dissolved in 50 mM phosphate-buffered saline pH 8.0 that also contained 300 mM sodium chloride and 2 mM imidazole. It is clearly observed that t for 50 M of 6His-UHRP measured in 50 saline solution pH 8.0 that also contained 300 mM sodium ch was obtained with the flow through system 4-2 on solutions containing known concentrations of 6His-UHRP. This calibration curve was s the typical calibration data (Figure 4-18A) and typical calibration curve (Figure 4-18B) 146

PAGE 147

for 6His-UHRP. The calibration curve shown an excellent linear correlation (r=0.999) withinrange from 0.374 to 1.496 M in protein concentration. a and (B) calibration curve for 6His-UHRP. binant protein transported across the metal Figure 4-18.Typical plots of (A) calibration dataFigure 4-19 shows the nanomoles of the recomperme ion affinity nanotube membranes versus permeation time. In addition, the results of the ation experiments for four different metal-immobilized imidazole membranes are compared. Figure 4-19.Flux plots for 6His-UHRP in imidazole functionalized metal ion affinity nanotube membranes that had immobilized different metal ions: (a) Cu+2, (b) Ni+2, (c) Co+2, and (d) Zn+2. The feed solution concentration was 1.87 M for all permeation experiments. 147

PAGE 148

We found that the flux of 6His-UHRP through modified membranes can be controlled by riate divalent metal ion (copper, nickel, cobalt or zinc) immobilized in the memb the approprane via coordination with the imidazole, as is shown in Figure 4-19. The data clearly demonstrated that the functional transport of the imidazole functionalized membrane depends on the appropriate immobilized divalent metal to coordinate with the imidazole ligands in the membrane. When the Zn ion was immobilized in the membrane the flux was considerably higher than the flux for the other three divalent metal ions (Co, Ni, and Cu). The flux for the Zn situation was not linear after 3 hours of permeation due mainly to the rapid equilibration reached between the feed and receiver solutions. Thus, the coordinated Zn in the imidazole functional membrane acts as binding sites to facilitate the transport of the 6His-UHRP. Figure 4-20.Effect of the divalent metal ion immobilized on the flux of 6His-UHRP across imidazole functionalized metal ion affinity nanotube membranes. Analogous permeation experiments were obtained for 6His-UHRP with the other three quite linear and their fluxes were always lower than the Zn immobilized situation. The slope of the flux plots for the four different immobilized metal ions was determined by using transport data divalent metal ions. The transport plots of 6His-UHRP with these three metal ions were not 148

PAGE 149

only during the initial three hours of permeation, period where fluxes were apparently linear for all four metal-immobilized membranes. These results are summarized in Figure 4-20. In order to quantify the effect of the different metal ions immobilized in the trans port of 6His-g the ionalized metal ion affinity nanotube membranes moles h-1 cm-2) UHRP across the metal ion affinity nanotube membranes, an enhancement selective coefficient, was defined as the calculated fluxes during three hours of transport for each divalent metal ions divided by the flux of the Cu situation. Thus, it was observed that durinthree initial hours of permeation, the fluxes across the metal ion affinity nanotube membranes with Ni, Co and Zn immobilized ions are enhanced respectively in ~2, ~3 and ~9 times, compared with the Cu immobilized membrane. These results are shown in table 4-3. Table 4-3.Fluxes and selectivity coefficients for 6His-UHRP across imidazole funct 2 Metal Ion Immobilized Flux of 6His-UHRP (nano 2 Cu F lux MFlux Cu +2 0.43 1.0 Ni +2 0.77 1.8 Co +2 1.47 3.4 Zn +2 4.05 9.4 Thus, these data demonstrated that by the selection of the appropriate transition metal ion to cooConclusions The development of metal ion affinexploe rdinate the imidazole, the transport properties within the functional membrane can be controlled. ity nanotube membranes has been demonstrated by iting the concept of IMAC separations in a nanopore-membrane-platform. Silica nanotubmembranes were functionalized with imidazole through a nucleophilic substitution reaction between 2-mercaptoimidazol and 3-chloropropyl modified silica nanotubes carried out in an 149

PAGE 150

aprotic medium. It was demonstrated that transport of histidine-tagged recombinant proteins across these imidazole-modified silica nanotube membranes is controlled by the nature of immobilized metal ion on the imidazole surface. Besides, the extensive characterization of imidazole functionalized membrane allowed the understanding of how coordination chemistry with different divalent metal ions affects the functional behavior of the membrane and transportproperties during the permeation of six-histidine recombinant proteins. Furthermore, additional results shown that by the appropriate choice of the transition metal ion for coordination with the imidazole ligand, the binding strength and optical properties within the functional membrane are also modified. Hence, these studies revealed how the presence of the immobilized metal ion affects the functionality of the imidazole membrane and its transport properties. The metal ioaffinity nanotube membrane described here has the potential to be a versatile analytical platformwhere different metal ions can be immobilized on the same prevailing conditions, additionally offering the possibilities of regeneration without apparent loss of the metal coordination properties of the functional membrane. These studies demonstrate an alternative way of immobilized metal ion affinity by using nanotube membranes. the n 150

PAGE 151

CHAPTER 5 TRANSPORT ANALYSIS OF PROTEINS ACROSS SILICA NANOTUBE MEMBRANES THROUGH A FLUORESCENCE QUENCHING-BASED SENSING APPROACH Introduction Molecular recognition is a very important and commonly used tool in both clinical diagnostics and pharmacology. The practical methods based on recognition depend fundamentally on specific biomolecular recognition reactions. In some cases, these biomolecular recognition reactions confer such extremely high specificity that it has allowed the development of analytical assays for the determination of many chemicals and biochemicals at very low concentrations. 290 293 This is important because in this era of proteomics in particular, the quantitative determination and identification of proteins at low concentrations are essential requirements. Many very sensitive methods have been developed for the detection of proteins and other biological molecules on the basis and capacity of fluorescent compounds to transfer energy absorbed from photon radiation to close molecules by taking advantage of the molecular recognition. 294 297 This called fluorescence resonant energy transfer (FRET) phenomenon takes place when the energy in the excited state of a donor fluorophore is transferred to a close acceptor molecule via an appropriate orientation of the dipole-dipole interaction between these donor-acceptor molecules. 298, 299 One of the basic conditions for the FRET process to occur is that the emission spectrum of the donor must overlap the absorption spectrum of the acceptor. Therefore, FRET efficiency is a distance-dependent interaction process. 131, 298 This distance dependence of FRET has been extensively used to study both structure and dynamics of proteins and nucleic acids, as well for the detection and visualization of intermolecular associations, and for the development of biomolecular devices based on intermolecular binding interaction. 298, 300, 301 151

PAGE 152

If the donor and acceptor molecules are brought close together within the range of 20 to 100 the fluorescence intensity of the donor fluorophore is reduced (quenched) as a result of the FRET process. However, if the donor and acceptor molecules are brought within an intimate distance, most of the absorbed energy is dissipated as heat and only a small amount of the energy is emitted as fluorescence; this phenomenon is commonly referred to as static or contact quenching. 131 The rate of fluorescence quenching can be studied experimentally by calculating the Stern-Volmer quench (K SV ) constant. 131 In the present study, a new approach is described to study the transport of proteins across silica nanotube membranes by using a fluorescence quenching-based analytical method. This analysis was accomplished by simply observing the fluorescence quenching of a donor-fluorescent labeled protein through a ligand-receptor binding mechanism. Thus, the flux of anti-Human IgG conjugated with the quencher QSY-7 (anti-H-IgG-QSY) was monitored by using a known concentration of Human IgG FITC-labeled (H-IgG-FITC) as fluorescence emitter and donor in the permeation compartment. In order to preserve the chemical potential as the actual driving force for the flux, a much higher concentration of protein in the feed compartment (anti-H-IgG-QSY) was intentionally kept with respect to the concentration of emitter-labeled protein in the permeation compartment. It is described that the rate of fluorescence quenching depends on the flux of anti-H-IgG-QSY across the nanotube membrane, and that the rate of fluorescence quenching can be used to determine the concentration of the analyzed protein on the feed solution. The results of this investigation are reported here. Experimental Materials Commercially available nanopore alumina membranes (60 m thick, nominal pore diameter 100 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999% (TEOS) 152

PAGE 153

(Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes. The purified mouse immunoglobulin (M-IgG) immobilized within the silica nanotube membranes was obtained from Sigma Aldrich. Membranes modified with purified human immunoglobulin (H-IgG) obtained from Sigma Aldrich were also prepared. A trimethoxysilyl propyl aldehyde (United Chemical Technologies) was used to attach the M-IgG or H-IgG to the nanotubes. A purified human immunoglobulin conjugated with the FITC dye (H-IgG-FITC) obtained from Sigma Aldrich was used as emitter during the fluorescence quenching detection. A purified rabbit immunoglobulin conjugated with the FITC dye (R-IgG-FITC) obtained from Sigma Aldrich was used as control emitter during the fluorescence quenching detection. To investigate the transport properties across silica nanotube membranes the whole antibody raised in goat against H-IgG was used was used as permeating protein. This anti-human IgG (anti-H-IgG) with specific affinity for the F ab fragments of the H-IgG was obtained from Sigma Aldrich. In order to observe the fluorescence quenching of the donor-fluorescent labeled protein through a molecular recognition reaction, the anti-H-IgG permeating protein was conjugated with the QSY-7 quencher. This protein labeling was accomplished via acylation reaction of an N-Hydroxysuccinimidyl (NHS) ester of QSY-7 to the primary amine residues of the protein to form a stable amide bond. This NHS ester of QSY-7 was purchased from Molecular Probes (Eugene, OR). The BioGel P-30 resin provided in the amine-reactive protein labeling kit obtained from Molecular Probes was used for purification of the anti-H-IgG conjugated with QSY-7. SuperBlock blocking buffer solution in phosphate-buffered saline was obtained from Pierce. SuperBlock solution contains a proprietary protein for blocking binding sites for non-specific protein adsorption. Tween 20 was obtained from Sigma Aldrich. All other chemicals were of 153

PAGE 154

reagent grade and used as received. Purified water, obtained by passing house-distilled water through a Barnstead, E-pure water-purification system, was used to prepare all solutions. Preparation of the Silica Nanotube Membranes Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-based template-synthesis method described previously. Briefly, a sol-gel precursor solution was prepared by mixing absolute ethanol, TEOS and 1 M HCl (50:5:1 by volume), and this solution was allowed to hydrolyze for 30 minutes. The alumina membranes were then immersed into this solution for two minutes under sonication. The membranes were then removed and the surfaces swabbed with ethanol to remove precursor from the alumina surface. The membranes were then air dry for ten minutes at room temperature and cured in an oven for 12 hours at 150 C. This yields silica nanotubes, with wall thickness of ~3 nm, embedded within the pores of the alumina membrane. 9 Antibody Immobilization The immobilization chemistry has also been described previously. The aldehyde silane was first attached to the inside walls of the silica nanotubes. The membrane was then immersed into a solution that was 1.0 mg per mL either in M-IgG or in H-IgG dissolved in 10 mM phosphate-buffered saline with pH = 7.4 that also contained 150 mM sodium chloride and 5 mM sodium azide. The membrane was incubated overnight at 4 C, which allowed free amino sites on the antibody to attach via the Schiffs base reaction to the nanotube-immobilized aldehyde groups. After immobilization, the membranes were rinsed with buffer and then incubated for 5 hours in the SuperBlock solution that also contained 0.05% Tween 20. It is worth mentioning that the unmodified silica nanotube membranes were also incubated for 5 hours in the SuperBlock solution containing 0.05% Tween 20 before the transport experiments. 154

PAGE 155

Labeling of anti-H-IgG with QSY-7 Forty microliters of 6 mM QSY-7 solution prepared in absolute DMSO was added to a 1 mL of 13.3 M of anti-H-IgG solution prepared in 10 mM phosphate-buffered saline with pH = 7.4 that also contained 150 mM sodium chloride. This solution was mixed thoroughly by gentle vortexing. The reaction was incubated at room temperature for 2 hours on an orbital shaker platform. Unconjugated QSY-7 was separated from the protein-quencher conjugated by size exclusion chromatographic using BioGel P-30 resin equilibrated with 10 mM phosphate-buffered saline with pH = 7.4 that also contained 150 mM sodium chloride. The same buffer was also used as eluent. It should be pointed out that in order to have a separation of the elute protein with minimal sample dilution, the dimensions of the column separation, which was provided in the amine-reactive protein labeling kit, were approximately 20 cm in length and 0.5 cm in diameter. The final concentration and degree of labeling of the purified anti-H-IgG-QSY was determined spectrophotometrically by measuring the absorbance of the protein at 280 nm and the absorbance of the QSY-7 at 560 nm using a sample dilution factor of 10 times of the original purified fraction. The protein concentration was calculated by the Beer-Lambert law using a correction factor of 0.22 for the absorbance of the QSY-7 at 280 nm, 302 and a molar extinction coefficient ( IgG ) for the IgG of 210,000 M -1 cm -1 in the Equations 5-1 and 5.2. 302 )22.0(560280AAAproteinconjugated (5-1) IgGIgGproteinconjugatedCbA (5-2) The degree of labeling in the protein was also determined by using the Equation 5-3. )()(][]7[11711560cmMAcmMAIgGQSYQSYproteinconjugatedIgG (5-3) It was found a degree of labeling of 4.6 by using 90,000 M -1 cm -1 as the molar extinction coefficient ( QSY-7 ) for the QSY-7 quencher molecule. 302 155

PAGE 156

Transport Experiments The bare silica nanotube membranes, M-IgGor H-IgG-modified membranes were sandwiched between two pieces of Scotch tape that had 0.31 cm 2 area holes punched through them. These holes defined the area of membrane exposed to the contacting solution phases. This membrane assembly was mounted between the two halves of a U-tube permeation cell, with half-cell volumes of 10 mL. The feed solution contained the quencher conjugated antibody (anti-H-IgG-QSY) dissolved in 10 mM phosphate-buffered saline pH 7.4 that also contained 150 mM sodium chloride and 5 mM sodium azide (buffer used for all studies). The receiver solution on the other side of the membrane contained the buffered solution of the fluorescence emitter labeled protein the H-IgG-FITC. The concentration of the transporting protein in the feed solution was in all cases higher than the emitter-donor-protein in the receiver side. Thus, the chemical potential was preferentially preserved from the feed to receiver solutions as the actual driving force for the flux. A protocol was established to carry out all the transport experiments. First, the membrane mounted in the U-tube permeation cell was equilibrate for one hour in the buffer solution, and the receiver solution were frequently measured using a flow-through fluorescence detector until a stable base line for the fluorescence signal was reached. After that, a known concentration of anti-H-IgG-QSY was placed in the feed solution, and it was permitted to equilibrate for another fifteen minutes. Then, a known concentration of H-IgG-FITC was put in the receiver solution, after five minutes when the fluorescence signal was stable the experimental data for the transport experiment started being recorded. The changes of fluorescence intensity in the receiver solution were periodically measured using a flow-through fluorescence detector (Figure 5-1). 156

PAGE 157

Figure 5-1.Flow-through fluorescence system used to measure the changes of fluorescence intensity in the receiver solution. This system consisted of an L-7485 fluorescence detector (Hitachi). The flow stream was delivered using an L-7100 HPLC pump (Hitachi). We discovered that if the receiver solution was sent continuously through the fluorescence detectors during the entire course of a transport experiment, significant photobleaching of the FITC dye occurred. To prevent this, we turned the pump on to deliver receiver solution to the detectors for only 3 minutes per hour of transport experiment. That is, we used a duty cycle of 57 minutes with pump off followed by 3 minutes with pump on (flow rate = 3 mL per min) to collect fluorescence data. The whole system was controlled using Chromeleon 6.50 chromatography management software (Dionex) through a UCI-100 universal chromatography interface (Dionex). FITC was excited at 495 nm, and the emission was detected at 515 nm. Calibration curve was obtained with the same flow through system on solutions containing known concentrations of the flourophore-labeled antibody. The calibration curve was used to obtain the correct concentration of the unquenched fluorescent labeled protein in solution. Results and Discussion Experimental Principle Used for the Fluorescence Quenching Detection For these studies, the transport analysis design of the anti-H-IgG-QSY across silica nanotubes was mainly based on the FRET process. Therefore, the analysis required the use of a 157

PAGE 158

fluorescence donor (FITC) and fluorescence acceptor (QSY-7) molecules, which their structures are shown in Figure 5-2. Figure 5-2.Structures of (a) QSY-7 and (b) FITC One of the basic conditions for the FRET process to occur is that the emission spectrum of the donor must overlap the absorption spectrum of the acceptor. Figure 5-3 illustrates both the excitation and emission spectra for FITC and the absorption spectrum for QSY-7. Figure 5-3.Spectra for FITC and QSY-7 It is clearly seen how the emission spectrum of the FITC fluorophore overlaps significantly the absorption spectrum of the QSY-7 dye. It demonstrated that these selected pair of donor and acceptor molecules fulfills the overlapping condition for the FRET process required for the studies described here. The QSY-7 quencher is a nonfluorescent acceptor fashioned to quench specifically the fluorescence within the range of 500 to 600 nm. 303 The FRET efficiency is a distance-dependent interaction process, and the FRET efficiency is related to the Frster radius 158

PAGE 159

(R o ), and this radius corresponds to the distance at which fluorescence resonance energy transfer from the donor dye to the acceptor dye is 50% efficient (Equation 5-4). 131, 298 quencher and efluorophorbetween distance r 1160RrE (5-4) The magnitude of R o is dependent on the spectral properties of the donor and acceptor dyes and the typical value reported for the FITC and QSY-7 pair is 61 303 The suitable proximity for the fluorescence quenching process between these fluorescence donor and acceptor molecules was accomplished through a molecular recognition reaction using two antibodies with specific affinity for each other. This antigen-antibody immunoassay is illustrated in Figure 5-4. Figure 5-4.Molecular recognition reaction used to promote the fluorescence quenching. Thus, the flux of anti-H-IgG-QSY was monitored by using a known concentration of H-IgG-FITC as fluorescence emitter in the permeation compartment. Verification of the Fluorescence Quenching System via H-IgG-FITC and anti-H-IgG-QSY Reaction in Solid State In order to verify that the molecular recognition reaction between H-IgG-FITC and anti-H-IgG-QSY leads to an inhibition of the fluorescence process, a silica nanotube membrane was modified with H-IgG-FITC. The fluorescence signal for this modified membrane was observed under microscope before and after six hours of exposure to a solution of anti-H-IgG-QSY that 159

PAGE 160

was 100 nM in protein. The results of the fluorescence images and fluorescence spectra of the H-IgG-FITC-modified membrane are shown in Figure 5-5. Figure 5-5.Fluorescence image and spectra respectively for a H-IgG-FITC-modified membrane, (A-C) before exposure, and (B-D) after exposure to a sample solution of anti-H-IgG-QSY. It was observed that the fluorescence intensity of the H-IgG-FITC-modified membrane is completed inhibited through the immunoassay with anti-H-IgG-QSY. These results confirmed that the fluorescence signal of H-IgG-FITC can be quenched by anti-H-IgG-QSY by taking advantage of the antibody-antigen molecular recognition reaction, which situates the fluorescence donor and acceptor molecules within the intimate distance required for the quenching process. Control Transport Studies In order to demonstrate that the transport of proteins across silica nanotube membranes can be measured using fluorescence quenching detection through a biomolecular recognition reaction 160

PAGE 161

between a fluorescence donor fluorescentlabeled antibody and a fluorescence acceptor quencherconjugated antibody, two different types of controls were carried out. In the first control experiment (control 1), the fluorescence of a 3.35 nM solution of R-IgG-FITC was measured in the buffered receiver solution, whereas a 33.3 nM solution of the quencher-conjugated protein (anti-H-IgG-QSY) was placed in the buffered feed solution. For the second control (control 2), the fluorescence of a 3.35 nM solution of H-IgG-FITC was measured in the buffered receiver solution, but in this case, the protein place in the buffered feed solution was the anti-H-IgG without QSY-7 conjugation. The same concentration of protein, a 33.3 nM of anti-H-IgG was also used in the buffered feed solution during the control 2. The results for these control experiments are shown in the Figure 5-6. Figure 5-6.Typical plots for (A) changes of fluorescence signal, and (B) concentration of unquenched protein in function of time for two different control transport/detection systems: (a) anti-H-IgG/H-IgG-FITC, and (b) anti-H-IgG-QSY/R-IgG-FITC. Figure 5-6A shows the changes of fluorescence signal in the receiver solution during the course of the transport experiments both for control 1 and control 2. It was observed that for both controls the fluorescence signal is reduced as the time of the transport experiment increases. Since any of these two controls comply with the conditions required for FRET, this inhibition in fluorescence signal was attributed mostly to two factors different of FRET process. One factor 161

PAGE 162

was endorsed to photobleaching of the fluorophore, and the second to a partial diffusion of the FITC-labeled proteins toward the direction of the feed solution. By assuming that during the time of the transport experiment the degree of labeling (DL) remains constant for the fluorescently protein, calibration data allowed us to convert these fluorescence intensities to molar concentration of unquenched protein in the receiver solution (Figure 5-6B). The observed variations in scale for the plots in Figure 6-6A and Figure 6-6B are due to the difference in the DL between H-IgG-FITC and R-IgG-FITC proteins (sensitivity). The DL for H-IgG-FITC was 4.2, whereas it was 4.7 for R-IgG-FITC. The plots in Figure 6-6B give the impression that the deactivation of the fluorescence intensity for these two controls was carried out via two different mechanisms. In the case of the anti-H-IgG-QSY/R-IgG-FITC pair, it was observed that during the initial hour of the experiment the rate of fluorescence quenching for this system was higher than for the anti-H-IgG/H-IgG-FITC pair. Although, there is not a specific affinity for these two proteins (control 1), this effect might be assigned to the possibility that the donor (FITC) and acceptor (QSY-7) fluorescence label moieties in the proteins collided with each other to deactivate the fluorescence intensity faster than the other control system, for which the presence of QSY-7 quencher is absence. The anti-H-IgG-QSY/R-IgG-FITC pair (control 1) was chosen as the control for the transport experiments performed through the specific molecular recognition reaction proposes in this approach. Effect of the Feed Solution Concentration on the Rate of Quenching The experimental system demonstrated here, relies on the strength of the binding affinity and interaction between the labeled antibodies to bring together the fluorophore and quencher to an appropriated distance in order that the fluorescence quenching occurs. To study the flux across silica nanotube membranes using this fluorescence quenchingmediated detection, we 162

PAGE 163

carried out experiments to determine the effect of the feed solution concentration on the rate of fluorescence quenching. For these experiments, the concentration of H-IgG-FITC in the receiver solution was 3.35 nM in all cases, whereas three different concentrations for anti-H-IgG-QSY were used in the feed solution. The three concentrations of anti-H-IgG-QSY were respectively, 16.6 nM, 33.3 nM, and 66.6 nM. These results are shown in Figure 5-7. Figure 5-7.Typical plots for (A) changes of fluorescence signal and (B) concentration of unquenched protein in function of time for three different feed solution concentrations: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control). Figure 5-7 shows the influence of feed solution concentration on the rate of fluorescence quenching in the receiver solution. As can be observed from Figure 6-7, transport experiments with higher concentration of anti-H-IgG-QSY yields to a faster rate of the fluorescence quenching. These results confirmed that the molecular recognition reaction between the labeled antibodies generated the appropriate distance between the donor (FITC) and acceptor (QSY-7) molecules necessary for this fluorescence quenching-mediated detection. This was also verified by comparing this effect with the control experiment (plot d in Figure 5-7). The significant differences in fluorescence intensity for the experiments shown in Figure 5-7were observed only during the initial four hours of the transport experiments. After this period of time, the changes in 163

PAGE 164

fluorescence signal were less evident. This behavior can be rationalized in that only a small fraction of the labeled proteins fulfilled an appropriate orientation of the dipole-dipole interaction between the donor-acceptor molecules for FRET. However, the possibility of contact quenching should be considered as well. Data Analysis Based on the Stern-Volmer Plots The role of fluorescence quenching can be experimentally studied by determining the quenching rate parameters using the Stern-Volmer plots that are drawn in accordance with the Stern-Volmer equation (Equation 5-5) 131 QKIISV10 (5-5) From the experimental data of this investigation, the terms I 0 and I in Equation 5-5 are known, and they are respectively the fluorescence intensity in the absence and presence of the quencher-conjugated protein. However, the concentration of the quencher-conjugated protein (anti-H-IgG-QSY) is not absolutely defined. Therefore, some assumptions were made in order to quantitatively describe the fluorescence quenching process in terms of the Stern-Volmer equation. By assuming that during the time of the transport experiment the degree of labeling (DL) remains constant for the fluorescently protein, the fluorescence intensities were converted to molar concentration of unquenched protein via calibration curve. It was also assumed that the reaction between anti-H-IgG-QSY and H-IgG-FITC during the quenching phenomenon was 1:1, since the molar concentration of unquenched protein was already established. Besides, it was assumed that the differences in molar concentration of unquenched protein with time represent the concentration of the quencher-conjugated protein (anti-H-IgG-QSY) that was able to have the right conditions to interact with the donor-labeled protein (H-IgG-FITC) to yield the quenching 164

PAGE 165

process. Thus, the I 0 /I ratio was associated with the assumed concentration of anti-H-IgG-QSY in the Stern-Volmer plot (Figure 5-8). Figure 5-8.Stern-Volmer plots of I 0 /I against concentration of anti-H-IgG-QSY in the receiver solution transported across silica nanotube membranes. The effect of the feed solution concentrations of anti-H-IgG-QSY is shown: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control). Figure 5-8 shows the linear segment of the Stern-Volmer plots from the data compiled in Figures 5-7A and 5-7B. The linear part of the data analysis was used with the purpose of comparing the slopes of these Stern-Volmer plots (K SV ) as a function of the feed solution concentration of anti-H-IgG-QSY (Table 5-1). The insert in Figure 5-8 displays the complete range of the data analysis. It should be mentioned that those plots were not linear, and a positive deviation in all of them was observed, which is clearly noticed in the insert in Figure 5-8. The no linearity of the plots suggested that the observed fluorescence quenching is not merely collisional; thereby the contribution of contact quenching should be also considered. It was found that the rate of fluorescence quenching depends on the flux of anti-H-IgG-QSY across the nanotube membrane, and that the rate of fluorescence quenching can be used to determine the concentration of the analyzed protein on the feed solution. Figure 5-9 and Table 5-1 summarize these results. 165

PAGE 166

The slopes (K SV ) calculated from the linear segments of the Stern-Volmer plots were compared on function of the feed solution concentration of anti-H-IgG-QSY (Table 5-1). The values obtained from the slopes (K SV ) confirmed that higher fluxes lead to faster rates of quenching. Figure 5-9.Effect of the feed concentration on the Stern-Volmer quenching constant K SV A linear relationship was found when the Stern-Volmer quenching constant (K SV ) was plotted versus the concentration of anti-H-IgG-QSY in the feed solution; this plot is shown in Figure 5-9. These data demonstrated that the linear relationship between K SV and feed solution concentration can be used to determine the concentration of the analyzed protein on the feed solution. Observing the values of K SV an additional conclusion that might be inferred, is that they can be associated with the binding affinity (nM -1 ) for the permeating protein to the fluorescent labeled antibody in the receiver solution. It should be pointed out that K SV is associated with the rate constant of the dynamic reaction between quencher and fluorophore. Therefore, this analytical approach proposed in these studies might also be used to estimate binding affinities of antibodies in solution. 166

PAGE 167

Table 5-1.Relationship between the Stern-Volmer quenching constant (K SV ) with the feed solution concentration of anti-H-IgG-QSY Feed Concentration of anti-H-IgG-QSY (nM) K SV (nM -1 ) 0 (control) 0.369 16.6 0.431 33.3 0.491 66.6 0.621 Data Analysis Based on log-log Plots of Fluorescence vs. Time A second and much simpler analysis for the obtained data was based on the logarithmic plots of the changes of fluorescence in function of time. The results for this analysis are shown in Figure 5-10. Figure 5-10.log-log plots of the of fluorescence signal vs. time showing the effect of feed solution concentration of anti-H-IgG-QSY: (a) 16.6 nM, (b) 33.3 nM, (c) 66.6 nM, and (d) 33.3 nM (control). The respective log-log plots of the Figure 5-7A is shown in Figure 5-10. After the experimental data was represented in a logarithmic scale the relation between fluorescence and time was almost linear, which confirmed that the relationship between these two physical variables is described by a power law formula. It was observed that as the concentration of anti-H-IgG-QSY in feed solution increases the slope of the log-log plot tends toward more negative values as a result of faster rate of quenching. The trend observed with this simple analysis (log167

PAGE 168

log plots) confirmed also that transport experiments with higher concentration of anti-H-IgG-QSY yields to a faster rate of the fluorescence quenching. These results are summarized in Table 5-2. Table 5-2.Relationship between slope of the log-log plots for the fluorescence signal vs. time with the feed solution concentration of anti-H-IgG-QSY. Feed Concentration of anti-H-IgG-QSY (nM) Slope of log-log plots Intercept of log-log plots 0 (control) -0.1198 2.4360 16.6 -0.1374 2.3406 33.3 -0.1693 2.3488 66.6 -0.2095 2.3344 The relation between the slopes obtained from the log-log plots with the feed concentration is shown in Figure 5-11. Figure 5-11.Linear relationship between slopes obtained from the log-log plots with the feed concentration of anti-H-IgG-QSY. A linear relationship was also obtained between the slopes of the log-log plots with the feed concentration of anti-H-IgG-QSY. It showed excellent linearity (correlation coefficient = 0.996) over the range from 0 nM to 66.6 nM. Therefore, by this simple data analysis using log-log plots, it is also possible to determine the concentration of the analyzed protein on the feed solution by simply knowing the slope of the log-log plot. 168

PAGE 169

Transport across Antibody-Modified Silica Nanotube Membranes Transport experiments were also carried out across antibody-functionalized nanotube membranes. For these experiments silica nanotube membranes were modified both with H-IgG or M-IgG. The results for these experiments are compared in Figure 5-12. Figure 5-12.Typical plots for (A) changes of fluorescence signal and (B) concentration of unquenched protein in function of time illustrating the transport of anti-H-IgG-QSY across (a) H-IgG-, and (b) M-IgG-modified membranes. The feed solution concentration was 33.3 nM for both experiments. The data analysis for the transport across antibody-modified membranes based on Stern-Volmer plots and log-log plots are summarized in Table 5-3. Table 5-3.Data for the Stern-Volmer quenching constant (K SV ) and slope of the log-log plots for antibody-modified membranes Antibody-Modified Membrane* K SV (nM -1 ) Slope of log-log plots H-IgG 0.382 -0.0995 M-IgG 0.415 -0.1218 *The feed solution concentration of anti-H-IgG-QSY was 33.3 for both experiments. The results shown that the fluorescence quenching in the receiver solution was higher for the transport experiments carried out across M-IgG modified than H-IgG modified membranes. This is due to the faster rate of transport of anti-H-IgG-QSY across M-IgG as a result of the lower binding constant between these two proteins. Therefore, it demonstrated that antibody-modified membranes can be use to modulated the rate of fluorescence quenching for any specific 169

PAGE 170

pair of quencher-fluorophore labeled antibodies. These results confirmed that the observed fluorescence quenching phenomenon is a result of the biomolecular recognition reaction between anti-H-IgG-QSY and H-IgG-FITC in the receiver solution. Conclusions It was described a new method to study transport of proteins across silica nanotube membranes by using a fluorescence quenching-based sensing approach. This analysis was accomplished by simply observing the fluorescence quenching of a donor-fluorescent labeled protein (H-IgG-FITC) in the receiver solution after immunoassay reaction with the transported acceptor-fluorescent conjugated protein (anti-H-IgG-QSY). The flux of anti-Human IgG conjugated with the quencher QSY-7 (anti-H-IgG-QSY) was monitored by observing the changes in fluorescence of a known concentration of Human IgG FITC-labeled (H-IgG-FITC) in the receiver solution. The rate of fluorescence quenching was studied both in the absence and the presence of quencher labeled protein transported across of bare silica nanotube membranes and antibody-modified membranes. It was demonstrated that the rate of fluorescence quenching depends on the flux of anti-H-IgG-QSY across the nanotube membrane, and that the rate of fluorescence quenching can be used to determine the concentration of the analyzed protein on the feed solution. 170

PAGE 171

CHAPTER 6 HYBRIDIZATION STUDIES OF DNA IN A PNA-MODIFIED NANOTUBE MEMBRANE PLATFORM Introduction The development of DNA-sensors devices have attracted substantial research efforts directed to gene analysis, detection of genetic disorders, analysis of single nucleotide polymorphism in genes, tissue matching and forensic applications among others. 304 307 Various techniques, including quartz crystal microbalance, 308 electrochemical, 309 fluorescence, 310 electrophoretic, 311 electrochemiluminescence, 312 enzymatic, 313 surface plasmon resonance, 314 and atomic force microscopy, 315 have been developed to detect DNA hybridization. However, the label-free detection methods have attracted more interest than other methods because they can be simple, rapid, highly sensitive, and inexpensive. 307 In 1953, Watson and Crick proposed a model for the structure of DNA. 316 This model predicted that DNA would exist as a helix of two complementary anti-parallel strands. Soon after the first description of the double helix by Watson and Crick, it was shown, that the two strands could be separated by heat or treatment with alkali. The reverse process, which underlies all the methods based on DNA renaturation or molecular hybridization, was first described by Marmur and Doty. 317 It was quickly established that the two sequences involved in duplex formation must have some degree of sequence complementarity and that the stability of the duplex formed depends on the extent of complementarity. These remarkable properties suggested ways to analyze relationships between nucleic acid sequences, and analytical methods based on molecular hybridization. Since then, these methods have been rapidly developed and applied to a variety of biological malfunctions. Base-pairing (A-T and G-C) or hybridization is the underlying principle of DNA-sensors, namely the strong interaction between two complementary strands. Accordingly, such devices 171

PAGE 172

rely on the immobilization of a single-stranded DNA molecule (the probe) for hybridizing with the complementary (target) strand in a given sample. Then, the specificity of DNA hybridization devices depends primarily on the selection of the probe (sequence and length) and secondary upon the hybridization conditions (mainly the temperature, and ionic strength as well). Peptide nucleotide acid (PNA) was introduced in 1991 as a structural mimic of DNA with the ability to invade double helical DNA and bind its complementary target sequence. 318 Due to its higher specificity in the recognition of complementary DNA; PNA has already established itself as an attractive recognition layer for DNA sensing. 319 Figure 6-1.Structures of (a) PNA, and (b) DNA. PNA is a DNA analogue with a neutral N-(2-aminoethyl) glycine backbone replacing the negatively charged phosphate backbone of DNA (Figure 6-1). 318 Since the spacing between the nucleotides is the same as in DNA, the conventional Watson-Crick base pairing rules apply between mixed PNA/DNA sequences. PNA highly discriminates mismatch DNA and has a stronger binding affinity for complementary DNA than its DNA counterpart. Then, the favorable hybridization properties of PNA have also inspired the incorporation of PNA into array and nucleic acid biosensor formats. 320 323 172

PAGE 173

This chapter describes the use of PNA-modified nanotube membranes to study a label-free detection for DNA hybridization. This was achieved by exploiting the unique structural and hybridization features of PNA. This approach was characterized by observing the changes in the membrane resistance using impedance spectroscopy. This method allowed carrying out an in situ characterization of the interfacial properties of the modified nanotube membranes during the hybridization process in the duplex formation PNA-DNA. The novel idea was based on the fact that PNA is a neutral molecule; thereby during hybridization events achieved in just highly purified water without any electrolyte in the solution, the only source of new charges in the system during the hybridization process emerge from the DNA molecules. This event exhibited a change of the membrane resistant that was detected by impedance spectroscopy which is an effective technique to probe the interfacial properties of modified electrodes. Experimental Materials Commercially available nanopore alumina membranes (60 m thick, nominal pore diameter 100 nm and 200 nm) were obtained from Whatman. Tetraethyl orthosilicate 99.999% (TEOS) (Sigma Aldrich) was used to deposit the silica nanotubes within the pores of these membranes. A trimethoxysilyl propyl diethylenetriamine (DETA) was used to introduce amino groups to the nanotubes was obtained from United Chemical Technologies. Sulfosuccinimidyl-6-(biotinamido) hexanoate (NHS-LC-Biotin) obtained from Pierce was used as biotin-streptavidin linker. A purified streptavidin was obtained from Sigma Aldrich. A biotinylated-PNA (PNA) with twelve nucleotides bases (BiotinGCG CTA GCC TAC) obtained from Applied Biosystems was used as the probe to detect DNA. To investigate the hybridization properties of the PNA-modified nanotube membranes a complementary twelve bases DNA (3CGC GAT CGG ATG -5) with a MW=3607 g/moles was used. This complementary target 173

PAGE 174

DNA was obtained from Alpha-DNA. All other chemicals were of reagent grade and used as received. Purified water, obtained by passing house-distilled water through a Barnstead, E-pure water-purification system, was used to prepare all solutions. Preparation of the Silica Nanotube Membranes Silica nanotubes were deposited in the pores of the alumina membrane using the sol-gel-based template-synthesis method. Briefly, a sol-gel precursor solution was prepared by mixing absolute ethanol, TEOS and water (40:5:10 by volume), and this solution was allowed to hydrolyze for 1 hour. The alumina membranes were then immersed into this solution for two minutes under sonication, and they were kept in this solution for ten additional minutes. The membranes were then removed and the surfaces swabbed with ethanol to remove precursor from the alumina surface. The membranes were then air dry for ten minutes at room temperature and cured in an oven for 12 hours at 150 C. Preparation of PNA-Modified Nanotube Membranes The biotinylated-PNA probe was immobilized on nanotube membrane surfaces using the streptavidin-biotin affinity system. 297, 324 Streptavidin is a 60-kDa protein purified from the bacterium Streptomyces avidini. The interest in streptavidin is the remarkable and unique interaction with its low-molecular weight ligand biotin (K d 10 -15 M). Because of its extremely strong affinity, the streptavidin-biotin affinity system has been developed as a general tool to use in various experimental systems. 325 The immobilization process used in this approach consisted of three basic steps. The first and second steps were respectively, a silanization which introduced amino groups to the surface, and conjugation between the amino groups and the NHS-LC-Biotin. The latter acylation reaction of an N-Hydroxysuccinimidyl (NHS) ester of biotin to the primary amino groups of the silane-modified surface yields to a stable amide bond. The next step was the coupling of streptavidin to 174

PAGE 175

the biotin-modified surface. It should be pointed out that the NHS-LC-Biotin had a spacer arm of 22.4 that can reduce steric hindrance associated with the binding many biotinylated molecules on one streptavidin, which in turns may enhance the sensitivity of the system. Figure 6-2.Immobilization of biotinylated-PNA on templated synthesized silica nanotube membranes. During silanization the silica nanotube membranes were immersed into freshly prepared 1% (v/v) solution of DETA in 1 mM acetic acid for 30 minutes at room temperature. The DETA-modified membranes were rinsed with water to remove the unreacted DETA. The amino-silane modified nanotube membranes were dried under nitrogen and fixed at 120 C in the oven for 15 minutes. For the biotinylation process the amino-silane modified membranes were exposed for 6 hours to a 0.5 mg/mL solution of NHS-LC-Biotin prepared in 0.1 M bicarbonate buffer pH 8.5. The biotinylated nanotube membranes were then incubated overnight at 4 C in a 1 mg/mL 175

PAGE 176

solution of streptavidin prepared in 10 mM phosphate buffer pH 7.4. The PNA immobilization was carried out by incubating the streptavidin-modified membranes for 8 hours in a solution that contained 1 M of biotinylated PNA dissolved in water at 4 C. A scheme showing the immobilization steps for the PNA-modified nanotube membranes and the hybridization reaction with the complementary DNA is shown in Figure 6-2. Nanotube Membrane Assembly After the silanization with DETA, the silica nanotube membranes were assembled in a suitable format for their characterization with impedance spectroscopy. In order to make the nanotube membrane conductive, they were sputter-coated for 90 sec on both sides with Gold-Palladium at 75 mTorr of system pressure, 45 mA of current, conditions that gave a coating of approximately 10 nm thicknesses. The metal sputtered on the membranes enhanced the conduction properties of the membrane, and provided the facilities for the characterization of the membranes. Figure 6-3.Nanotube membrane assembly. The sputtered membranes together with two thin platinum leads of 60 microns thickness (one in each opposite side of the membrane) were placed between two pieces of plastic tape punched with a hole of 0.31 cm 2 area. The two platinum leads were sandwiched between the two pieces of tape in such a way that the holes were aligned and the platinum leads were covered 176

PAGE 177

with the tape to avoid their contact with the solution during the impedance measurements. A schematic illustration for this assembly is shown in Figure 6-3. These holes in the assembled nanotube membranes defined the area of the nanotube membrane exposed for the subsequent modification steps and the area exposed during the impedance measurements. It is worth noticing that the external electrical connections involves dry metal-to-metal contacts, so that, the membrane itself performs the function of the electrodes (working and reference). Impedance Measurement for Membrane Characterization For the impedance measurements the assembled membrane was mounted between the two halves of a U-tube permeation cell. This sputter-coated nanotube membrane performed and replaced the function of the typically used remote electrodes located on either side of the membrane. For all the impedance experiments both half-cells were filled with 10 mL of highly purified water. During the hybridization experiments one of the half-cell was chosen to spike the concentration of DNA target, this side also belonged to the side where the reference electrode connections were placed. The impedance data was obtained using a Solartron 1287 electrochemical interface module connected to a Solartron 1255B frequency response analyzer and a personal computer. The impedance data was recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz. Z-plot and Z-view software packages (Scribner Asc. Inc., NC) were used to control the impedance experiments and analyzed the impedance data, respectively. Impedance Data Analysis The main interest of in these studies was to monitor the changes in the membrane resistance during the hybridization process in the duplex formation PNA-DNA. Even though, 177

PAGE 178

that the experimental impedance data revealed the contribution of Warburg impedance, in order to simplify the data analysis this contribution was not considered. In addition, since this represent a bulk property of the electrolyte solution (diffusion of ions from the bulk electrolyte); it is not affected for the chemistry modification occurring at the surface of the pores in the membranes. Therefore, the experimental data of impedance were analyzed by using the model equivalent circuit shown in Figure 6-4. Figure 6-4.General equivalent circuit model used for impedance data analysis. A scan be seen from Figure 6-4, the data were fitted using a constant phase element (CPE). The reason of this is because in impedance experiments the capacitors often do not behave ideally. 326 This CPE is defined by the Equation 6-1. njACPEZ)()( (6-1) Where j is an imaginary number, is the angular frequency, and A is a constant. When this equation describes a capacitor, the constant A is equal to the inverse of the capacitance, and the exponent n is equal to 1. For a CPE the exponent n is less than 1. Thus, by fitting the experimental data with the equivalent circuit shown in Figure 6-4, the changes in membrane resistance, R m were extracted. Results and Discussion In order to evaluate the effects of the various immobilization steps on the surface of the sputter-coated nanotube membranes, impedance measurements were carried out during the all steps of the immobilization process. All the assays were conducted in highly purified water. A biotinylated PNA probe with a sequence BiotinGCG CTA GCC TAC, which is fully 178

PAGE 179

complementary to the DNA target with a sequence 3CGC GAT CGG ATG -5, was used as receptor in the nanotube membrane sensing-platform. These studies begin, however, with the impedance measurements for the alumina template, and silica nanotube membranes and investigations of the equilibration times required for the impedance measurements. Impedance Characterization of the Unmodified Membranes The Nyquist plots for bare alumina template (200 nm nominal pore size) and silica nanotube membranes are shown in Figure 6-5. The x-axis and y-axis in this plot corresponds to real impedance, Z re and imaginary impedance, Z im respectively. Z re and Z im are mainly originated from the resistance and capacitance of the cell system, respectively. Figure 6-5.Nyquist plots (Z im vs. Z re ) for impedance measurement of (a) alumina and (b) silica-nanotube assembled membranes in highly purified water. As shown in Figure 6-5, changes after the deposition of the silica nanotubes in the alumina template were observed in semicircle part at high frequency region (usually associated with kinetic controlled process), whereas the linear tail at lower frequencies (usually associated with mass transfer process) maintained almost the same slope. The variations in the high frequency region were probably due to a change in the isoelectric point (pI) of the nanotube membrane. Since it is known that the pI of the alumina membranes range from 6.9 to 10.5, while the pI for 179

PAGE 180

silica has been ranged from 1.9 to 4.5. 327 Because of the different electrical characteristics of these materials, a change in the surface characteristics of the membranes should be expected when silica nanotubes are synthesized within the alumina membrane, as was observed in the Nyquist plots in Figure 6-5. In order to assess the stability of the impedance measurements a study of the wettability of the bare alumina membrane was carried out by monitoring the changes in resistance with the time. These results are summarized in Figure 6-7. Figure 6-7.Effect of the wetting time on the resistance impedance measurement for an assembled unmodified alumina membrane 200 nm pore diameter. As can be seen from Figure 6-7, the changes in resistance from impedance measurements of alumina membrane were pretty stable with the time. These results confirmed the stability of the sputter-coated membranes and the stability itself of the impedance measurements. Functionalization of the Silica nanotube Membranes The differences in membrane resistance before and after each modification step during the immobilization process were analyzed in order to characterize the accomplishment of the various modifications (Figure 6-8). The obtained values of R m from the Nyquist plots in Figure 6-8 are compiled in Table 6-1. 180

PAGE 181

The results revealed that the different immobilization steps introduced significant changes in the membrane resistance. The most significant change was found in the membrane resistance associated with the hybridization process of DNA with the PNA-modified nanotube membrane. This decrease in resistance was associated with the increase of negative surface charge density due to the duplex formation PNA-DNA on the surface. Figure 6-8. Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water: (a) alumina, (b) after synthesis of silica-nanotube, (c) after silanization, (d) after biotinylation, (e) after streptavidin immobilization, (f) after PNA immobilization, and (g) after DNA hybridization. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz. The increase of R m after the deposition of silica nanotubes was consistent with the fact that after deposition smaller effective pore size in the membrane might be reached. Conversely, the decrease in membrane resistance after silanization was explained through the formation of protonated amino groups on the surface at the pH of the experiment (highly purified water). However, part of this amino protonated groups were neutralized by the conjugation reaction with the biotin linker, which was confirmed for the increase in resistance after the biotinylation. The 181

PAGE 182

coupling of streptavidin also produced a significant change in resistance, since the size of this protein reduced the effective pore size of the membrane. For the immobilization of the biotinylated PNA a small decrease in resistance was observed compared with the streptavidin coupling. This effect was difficult of explain, but it might be due to the fact that PNA oligomers are purified by reverse phase HPLC in the presence of trifluoroacetic acid. The aqueous solutions of PNA posses a pH lower than 3. Therefore, at this low pH some residual positive charge in the structure of PNA might be introduced in the membrane that causes the reduction of the resistance. Table 6-1.Variations of R m at the different steps of modification on a 100 nm pore diameter alumina template Modification step R m () Alumina membrane 8100 Silica nanotube membrane 12500 Amino-modified membrane 1030 Biotinylated-modified membrane 4515 Streptavidin-modified membrane 18000 PNA-modified membrane 14717 DNA hybridization 327 Impedance Sensing of DNA-PNA Hybridization Figure 6-9 shows the Nyquist plots of the PNA-modified nanotube membrane before and after 10 minutes of the hybridization process. As can be seen after the addition of 3 nmoles of the complementary DNA to the half-cell containing highly purified water (side where the reference electrode connections were also sited) a significant decrease of the semicircle sector of the plot was observed. This result suggested that the resistance of the PNA-modified nanotube membrane became lower due to the increase in new negative charge in the system brought to the membrane surface for the negatively charged phosphate backbone of DNA molecules during the 182

PAGE 183

hybridization process. There was almost no change in the solution resistance (R s ) and Warburg impedance (Z W ), which demonstrated that they were not affected during the hybridization process on the membrane surface. Figure 6-9. Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water for a PNA-modified nanotube membrane: (a) before, and (b) after 10 minutes of DNA hybridization. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz. Effect of the Pore Size on R m The influence of the pore size on the membrane resistance was explored by using two different sizes of commercially available (Whatman) nanopore alumina membranes having 60 m thickness, and nominal pore diameters respectively of 200, and 100 nm. To study this effect, nanotube membranes of the specified pore size were modified with PNA. For the hybridization impedance experiments, 3 nmoles of the complementary DNA were added to the highly purified water in both cases (Figure 6-10). 183

PAGE 184

Figure 6-10.Effect of the membrane pore size on the membrane resistance: (A) 200 nm, and (B) 100 nm. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz. Figure 6-10 compares the streptavidin immobilization, PNA immobilization, and DNA hybridization processes, respectively, by using membranes with two different pore diameters (100 and 200 nm). The changes in membrane resistance during the immobilization steps for the 100 nm membrane showed the same trend observed for the 200 nm pore diameter membrane explained above. However, the resistance was for all modification steps higher in the 100 nm pore diameter membranes, as a result of the smaller pore size. In the case of the PNA immobilization, the resistance was increased in approximately 280% by using the 100 nm membrane. The effect of the pore size during the hybridization process revealed a reduction of 98% in the resistance of the membrane for the 200 nm-situation (Figure 6-10A), while for the 100 nm184

PAGE 185

situation the reduction was of 53% (Figure 6-10B). These differences in the resistance during the hybridization process might be attributed to the fact that membranes with higher resistance present longer dynamic range. However, the possibility of a higher immobilization of PNA might be also considered. Stability and Reproducibility of the PNA-modified Membranes The stability of the PNA-modified membrane was evaluated by comparison the effect of DNA hybridization in time with an unmodified control membrane. These results are compared in Figure 6-11. Figure 6-11.Study of the hybridization process stability with time: (A) unmodified control alumina membrane, and (B) PNA-modified membrane. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA. Figure 6-11 shows the stability of the impedance measurement for a control (Figure 6-11A), and PNA-modified membrane (Figure 6-11B) after adding 3 nmoles of complementary DNA to the experimental cell. It is clearly seen that the changes in membrane resistance are mainly due to the hybridization process (Figure 6-11B). However, some reduction on the membrane resistance was observed for the control membrane after the addition of 3 nmoles of DNA. This effect might be attributed to nonspecific interactions on the membrane surface. The 185

PAGE 186

changes in membranes resistance after 15 hours of adding DNA were 27098 ohms for the PNA-modified membrane, while this change was only 4271 ohms for the control unmodified membrane. However, after 20 minutes these changes were 25152 and 1806 ohms for the PNA-modified and unmodified membranes, respectively. These results clearly demonstrated that the assembled PNA-modified membrane configuration study herein might be used as a good platform for DNA sensors. An additional evidence of this conclusion can be observed in Figure 6-12 were the effect of DNA concentration on the impedance measurements is illustrated. Figure 6-12.Nyquist plots (Z im vs. Z re ) of impedance measurement in highly purified water for a PNA-modified nanotube membrane: (a) before, (b) after adding 3 nmoles of DNA, and (c) after 6 nmoles of DNA. The plots were recorded applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz. The reproducibility of the PNA-modified nanotube membranes were also evaluated (Figure 6-13). For these hybridization studies three different membranes modified with PNA were compared. Figure 6-13 shows the results of membrane resistance for these three PNA-modified membranes, before and after hybridization with DNA (20 minutes and 3 days). It was observed that after 20 minutes of adding 3 nmoles of the complementary DNA solution to each of the PNA-modified membranes, the reduction of the resistance membrane were almost in the same order of magnitude. The reduction in resistance for membrane 1, 2, and 3 in Figure 6-13 were 186

PAGE 187

54%, 42%, and 53%, respectively. Amazingly was to observe the stability of these membranes after three days of the duplex formation during the hybridization process. Figure 6-13.Reproducibility study of the PNA-modified nanotube membranes, before and after the hybridization process. The resistance values were obtained from the impedance data recorded by applying 5 mV alternative voltage in the frequency range from 1 MHz to 0.05 Hz after the addition of 3 nmoles of complementary DNA. The results presented in these studies demonstrated the good stability and reproducibility of the PNA-modified nanotube membrane describe here. Conclusions It was described the use of PNA-modified nanotube membranes to study a label-free detection for DNA hybridization. This was achieved by exploiting the unique structural and hybridization features of PNA. This approach was characterized by observing the changes in the membrane resistance using impedance spectroscopy. This method allowed carrying out an in situ characterization of the interfacial properties of the modified nanotube membranes during the hybridization process in the duplex formation PNA-DNA. The novel idea was based on the fact that PNA is a neutral molecule; thereby during hybridization events achieved in just highly purified water without any electrolyte in the solution, the only source of new charges in the system during the hybridization process emerge from the DNA molecules. This event exhibited a 187

PAGE 188

change of the membrane resistant that was detected by impedance spectroscopy which is an effective technique to probe the interfacial properties of modified electrodes. These results showed the feasibility of using PNA-modified nanotube membranes as a sensing-platform for DNA hybridization methods. This assembled nanotube membrane platform present the special characteristic that the membrane itself performed the function of the electrodes (working and reference respectively). It was demonstrated that PNA immobilized on the nanotube membranes function as a good probe to detect the DNA targets in condition where no electrolyte is present during the hybridization process. It was also demonstrated the good reproducibility obtained during the fabrication and detection with this novel sensing-platform. 188

PAGE 189

CHAPTER 7 CONCLUSIONS The main goal of the research presented in this dissertation has been to develop selective nanotube membranes based on affinity molecular recognition with interesting applications in biological analysis. The unique property of biomolecules and metal ions to recognized specific molecules was exploited to elaborate functional nanotube membranes with potential application in sensing and separations of proteins and DNA. The template synthesis method was use to obtain the silica nanotube within the pores of alumina membranes. The walls of these nanotubes were functionalized via silane chemistry to couple molecular recognition agents onto the modified surfaces. Chapter 1 introduce the background information for this dissertation including the template synthesis method, the description of the alumina template membranes, sol-gel and silane chemistry, the inspiration of living transport in the development of membrane-based separations, fundamental about proteins, and a synopsis of the instrumental techniques used during the development and characterization of the applications describe in this dissertation. In chapter 2, a detailed experimental study on antibody-functionalized nanotube membranes was carried out, including the effect of surface modification, pore size, feed concentration, temperature, and ionic strength, both on the transport as on the functional behavior of the membranes. It was demonstrated that the type of immobilized antibody changes the functional behavior of the antibody-modified membranes and the transport properties of the permeating proteins across them. These membranes showed a Langmuirian shape plot characteristic of a typical behavior of facilitated transport membranes, when the effect of feed solution concentration on the fluxes of the permeating proteins was investigated. Therefore, the antibody-functionalized membranes showed here exhibit a type of behavior for facilitated-transport membranes during protein separations. Important parameters such as the binding 189

PAGE 190

capacities of the antibody-modified membranes and the binding constants for the permeating proteins to the antibody carriers confirmed that the permeating protein anti-M-IgG binds strongly to the carrier M-IgG and that the permeating protein anti-H-IgG binds weakly to this carrier, likewise, the permeating protein anti-H-IgG binds strongly to the carrier H-IgG and that anti-M-IgG binds weakly to this carrier. It was also demonstrated that by controlling the orientation of the immobilized antibody using protein G the potency affinity of the antibodies is effectively preserved. Through these studies was also demonstrated that by selecting a template membrane with an appropriated pore size diameters these antibody-modified nanotube membranes have the ability to be used with a combination of both size and affinity selectivity for protein separation. Additional studies on the effect of temperature on the fluxes and the binding constant demonstrated that the functional behavior of these antibody-modified membranes is governed for the changes in free energy before and after interaction of the permeating proteins with the carrier antibody, which was attributed to the changes in enthalpy and entropy of the system. Other important finding during this investigation was that by controlling the ionic strength of the transport solution, these antibody-modified nanotube membranes can be used for situations were the transport of an unwanted protein needs to be shut down, in order to improve selectivity. This latter approach might be useful in the sample complexity reduction in proteomic analysis. Since the analysis and characterization of complex mixtures of proteins are the central aims of proteomics. It is also noteworthy to mention the low pM detection limits achieved during these studies. In general, the ability to alter the surface biochemistries and thus tailor the membrane for specific analytes, promises the use of these membranes for the analysis of a wide range of proteins. 190

PAGE 191

Chapter 3 describes the modeling and characterization of the transport properties of proteins across antibody-functionalized nanotube membranes by using the dual-mode model. This work demonstrated that the dual-mode model proposes here was a suitable model for the characterization of the transport properties of anti-H-IgG and anti-M-IgG in M-IgG modified membranes. The model demonstrated a good agreement with the experimental system studied here. Besides, the findings showed in these studies allow addressing the concept of modulated facilitated transport systems using antibody-modified nanotube membranes, and how the extent of the facilitation is determined and controlled by the binding affinity between carrier and permeating protein molecules. These studies can provide a new direction in the development of more effective membranes for membrane-based separation. In chapter 4, the development of metal ion affinity nanotube membranes was demonstrated by exploiting the concept of IMAC separations in a nanopore-membrane-platform. Silica nanotube membranes were functionalized with imidazole through a nucleophilic substitution reaction between 2-mercaptoimidazol and 3-chloropropyl modified silica nanotubes carried out in an aprotic medium. It was demonstrated that transport of histidine-tagged recombinant proteins across these imidazole-modified silica nanotube membranes is controlled by the nature of immobilized metal ion on the imidazole surface. Besides, the extensive characterization of the imidazole functionalized membrane allowed the understanding of how coordination chemistry with different divalent metal ions affects the functional behavior of the membrane and transport properties during the permeation of six-histidine recombinant proteins. Furthermore, additional results shown that by the appropriate choice of the transition metal ion for coordination with the imidazole ligand, the binding strength and optical properties within the functional membrane are also modified. Hence, these studies revealed how the presence of the immobilized metal ion 191

PAGE 192

affects the functionality of the imidazole membrane and its transport properties. The metal ion affinity nanotube membrane described here has the potential to be a versatile analytical platform, where different metal ions can be immobilized on the same prevailing conditions, additionally offering the possibilities of regeneration without apparent loss of the metal coordination properties of the functional membrane. These studies demonstrate an alternative way of immobilized metal ion affinity by using nanotube membranes. Chapter 5 described a new method to study transport of proteins across silica nanotube membranes by using a fluorescence quenching-based sensing approach. This analysis was accomplished by simply observing the fluorescence quenching of a donor-fluorescent labeled protein (H-IgG-FITC) in the receiver solution after immunoassay reaction with the transported acceptor-fluorescent conjugated protein (anti-H-IgG-QSY). The flux of anti-Human IgG conjugated with the quencher QSY-7 (anti-H-IgG-QSY) was monitored by observing the changes in fluorescence of a known concentration of Human IgG FITC-labeled (H-IgG-FITC) in the receiver solution. The rate of fluorescence quenching was studied both in the absence and the presence of quencher labeled protein transported across of bare silica nanotube membranes and antibody-modified membranes. It was demonstrated that the rate of fluorescence quenching depends on the flux of anti-H-IgG-QSY across the nanotube membrane, and that the rate of fluorescence quenching can be used to determine the concentration of the analyzed protein on the feed solution. In chapter 6, the use of PNA-modified nanotube membranes to study a label-free detection for DNA hybridization was demonstrated. This was achieved by exploiting the unique structural and hybridization features of PNA. This approach was characterized by observing the changes in the membrane resistance using impedance spectroscopy. This method allowed carrying out an 192

PAGE 193

in situ characterization of the interfacial properties of the modified nanotube membranes during the hybridization process in the duplex formation PNA-DNA. The novel idea was based on the fact that PNA is a neutral molecule; thereby during hybridization events achieved in just highly purified water without any electrolyte in the solution, the only source of new charges in the system during the hybridization process emerge from the DNA molecules. This event exhibited a change of the membrane resistant that was detected by impedance spectroscopy which is an effective technique to probe the interfacial properties of modified electrodes. These results showed the feasibility of using PNA-modified nanotube membranes as a sensing-platform for DNA hybridization methods. This assembled nanotube membrane platform present the special characteristic that the membrane itself performed the function of the electrodes (working and reference respectively). It was demonstrated that PNA immobilized on the nanotube membranes function as a good probe to detect the DNA targets in condition where no electrolyte is present during the hybridization process. It was also demonstrated the good reproducibility obtained during the fabrication and detection with this novel sensing-platform. 193

PAGE 194

LIST OF REFERENCES [1] Hagg, M.-B. Separation and Purification Methods 1998, 27, 51-168. [2] Baker, R. W. Membrane Technology and Applications, 2nd ed.; John Wiley & Sons, Ltd., 2004. [3] Striemer, C. C.; Gaborski, T. R.; McGrath, J. L.; Fauchet, P. M. Nature 2007, 445, 749-753. [4] Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655-658. [5] Yu, S.; Lee, S. B.; Martin, C. R. Analytical Chemistry 2003, 75, 1239-1244. [6] Yu, S.; Lee, S. B.; Kang, M.; Martin, C. R. Nano Letters 2001, 1, 495-498. [7] Chun, K.-Y.; Stroeve, P. Langmuir 2002, 18, 4653-4658. [8] Van Reis, R.; Zydney, A. Current Opinion in Biotechnology 2001, 12, 208-211. [9] Lee, S. B.; Mitchell, D. T.; Trofin, L.; Nevanen, T. K.; Soederlund, H.; Martin, C. R. Science 2002, 296, 2198-2200. [10] Kohli, P.; Harrell, C. C.; Cao, Z.; Gasparac, R.; Tan, W.; Martin, C. R. Science 2004, 305, 984-986. [11] Giorno, L.; De Bartolo, L.; Drioli, E. Membrane Science and Technology Series 2003, 8, 187-217. [12] Gavalas, V. G.; Wang, J.; Bachas, L. G. Membrane Science and Technology Series 2003, 8, 379-392. [13] Sexton, L. T.; Horne, L. P.; Sherrill, S. A.; Bishop, G. W.; Baker, L. A.; Martin, C. R. Journal of the American Chemical Society 2007, 129, 13144-13152. [14] Baker, L. A.; Choi, Y.; Martin, C. R. Current Nanoscience 2006, 2, 243-255. [15] Choi, Y.; Baker, L. A.; Hillebrenner, H.; Martin, C. R. Physical Chemistry Chemical Physics 2006, 8, 4976-4988. [16] Siwy, Z.; Trofin, L.; Kohli, P.; Baker, L. A.; Trautmann, C.; Martin, C. R. Journal of the American Chemical Society 2005, 127, 5000-5001. [17] Harrell, C. C.; Choi, Y.; Horne, L. P.; Baker, L. A.; Siwy, Z. S.; Martin, C. R. Langmuir 2006, 22, 10837-10843. [18] Kohli, P.; Wirtz, M.; Martin, C. R. Electroanalysis 2004, 16, 9-18. 194

PAGE 195

[19] Harrell, C. C.; Kohli, P.; Siwy, Z.; Martin, C. R. Journal of the American Chemical Society 2004, 126, 15646-15647. [20] Steinle, E. D.; Mitchell, D. T.; Wirtz, M.; Lee, S. B.; Young, V. Y.; Martin, C. R. Analytical Chemistry 2002, 74, 2416-2422. [21] Lee, S. B.; Martin, C. R. Journal of the American Chemical Society 2002, 124, 11850-11851. [22] Wirtz, M.; Yu, S.; Martin, C. R. Analyst 2002, 127, 871-879. [23] Martin, C. R.; Nishizawa, M.; Jirage, K.; Kang, M. Journal of Physical Chemistry B 2001, 105, 1925-1934. [24] Lee, S. B.; Martin, C. R. Chemistry of Materials 2001, 13, 3236-3244. [25] Martin, C. R. Science 1994, 266, 1961-1966. [26] Martin, C. R.; Mitchell, D. T. Analytical Chemistry 1998, 70, 322A-327A. [27] Hulteen, J. C.; Martin, C. R. Journal of Materials Chemistry 1997, 7, 1075-1087. [28] Cepak, V. M.; Hulteen, J. C.; Che, G.; Jirage, K. B.; Lakshmi, B. B.; Fisher, E. R.; Martin, C. R. Journal of Materials Research 1998, 13, 3070-3080. [29] Miller, S. A.; Young, V. Y.; Martin, C. R. Journal of the American Chemical Society 2001, 123, 12335-12342. [30] Kang, M.; Trofin, L.; Mota, M. O.; Martin, C. R. Analytical Chemistry 2005, 77, 6243-6249. [31] Xu, F.; Wharton, J. E.; Martin, C. R. Small 2007, 3, 1718-1722. [32] Liang, W.; Martin, C. R. Journal of the American Chemical Society 1990, 112, 9666-9668. [33] Van Dyke, L. S.; Martin, C. R. Langmuir 1990, 6, 1118-1123. [34] Yamada, K.; Gasparac, R.; Martin, C. R. Journal of the Electrochemical Society 2004, 151, E14-E19. [35] Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268, 700-702. [36] Brumlik, C. J.; Menon, V. P.; Martin, C. R. Journal of Materials Research 1994, 9, 1174-1183. 195

PAGE 196

[37] Harrell, C. C.; Lee, S. B.; Martin, C. R. Analytical Chemistry 2003, 75, 6861-6867. [38] Mitchell, D. T.; Lee, S. B.; Trofin, L.; Li, N.; Nevanen, T. K.; Soederlund, H.; Martin, C. R. Journal of the American Chemical Society 2002, 124, 11864-11865. [39] Klein, J. D.; Herrick, R. D., II; Palmer, D.; Sailor, M. J.; Brumlik, C. J.; Martin, C. R. Chemistry of Materials 1993, 5, 902-904. [40] Hou, S.; Harrell, C. C.; Trofin, L.; Kohli, P.; Martin, C. R. Journal of the American Chemical Society 2004, 126, 5674-5675. [41] Hou, S.; Wang, J.; Martin, C. R. Journal of the American Chemical Society 2005, 127, 8586-8587. [42] Cai, Z.; Lei, J.; Liang, W.; Menon, V.; Martin, C. R. Chemistry of Materials 1991, 3, 960-967. [43] Parthasarathy, R.; Martin, C. R. Nature 1994, 369, 298-301. [44] Hornyak, G. L.; Patrissi, C. J.; Martin, C. R. Journal of Physical Chemistry B 1997, 101, 1548-1555. [45] Foss, C. A., Jr.; Hornyak, G. L.; Stockert, J. A.; Martin, C. R. Journal of Physical Chemistry 1994, 98, 2963-2971. [46] Scopece, P.; Baker, L. A.; Ugo, P.; Martin, C. R. Nanotechnology 2006, 17, 3951-3956. [47] Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655-658. [48] Hulteen, J. C.; Jirage, K. B.; Martin, C. R. Journal of the American Chemical Society 1998, 120, 6603-6604. [49] Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Analytical Chemistry 1999, 71, 4913-4918. [50] Martin, C. R.; Kohli, P. Nature Reviews Drug Discovery 2003, 2, 29-37. [51] Gasparac, R.; Kohli, P.; Paulino, M. O. M.; Trofin, L.; Martin, C. R. Nano Letters 2004, 4, 513-516. [52] Hillebrenner, H.; Buyukserin, F.; Kang, M.; Mota, M. O.; Stewart, J. D.; Martin, C. R. Journal of the American Chemical Society 2006, 128, 4236-4237. [53] Hillebrenner, H.; Buyukserin, F.; Stewart, J. D.; Martin, C. R. Journal of Nanoscience and Nanotechnology 2007, 7, 2211-2221. [54] Buyukserin, F.; Kang, M.; Martin, C. R. Small 2007, 3, 106-110. 196

PAGE 197

[55] Ai, S.; Lu, G.; He, Q.; Li, J. Journal of the American Chemical Society 2003, 125, 11140-11141. [56] Kovtyukhova, N. I.; Mallouk, T. E.; Mayer, T. S. Advanced Materials 2003, 15, 780-785. [57] Li, F.; Zhang, L.; Metzger, R. M. Chemistry of Materials 1998, 10, 2470-2480. [58] Masuda, H.; Fukuda, K. Science 1995, 268, 1466-1468. [59] Wood, G. C.; O'Sullivan, J. P. Electrochimica Acta 1970, 15, 1865-1876. [60] Shingubara, S.; Morimoto, K.; Sakaue, H.; Takahagi, T. Electrochemical and Solid-State Letters 2003, 7, E15-E17. [61] Masuda, H.; Hasegwa, F.; Ono, S. Journal of the Electrochemical Society 1997, 144, L127-L130. [62] Masuda, H.; Yada, K.; Osaka, A. Japanese Journal of Applied Physics, Part 2: Letters 1998, 37, L 1340-L 1342. [63] Li, A. P.; Muller, F.; Birner, A.; Nielsch, K.; Gosele, U. Journal of Applied Physics 1998, 84, 6023-6026. [64] Che, G.; Lakshmi, B. B.; Fisher, E. R.; Martin, C. R. Nature 1998, 393, 346-349. [65] Hurst Sarah, J.; Payne Emma, K.; Qin, L.; Mirkin Chad, A. Angewandte Chemie, International Edition 2006, 45, 2672-2692. [66] Huber, C. A.; Huber, T. E.; Sadoqi, M.; Lubin, J. A.; Manalis, S.; Prater, C. B. Science 1994, 263, 800-802. [67] Kovtyukhova, N. I.; Mallouk, T. E. Advanced Materials 2005, 17, 187-192. [68] Park, S.; Lim, J.-H.; Chung, S.-W.; Mirkin, C. A. Science 2004, 303, 348-351. [69] Lee, W.; Scholz, R.; Nielsch, K.; Goesele, U. Angewandte Chemie, International Edition 2005, 44, 6050-6054. [70] Zhi, L.; Wu, J.; Li, J.; Kolb, U.; Muellen, K. Angewandte Chemie, International Edition 2005, 44, 2120-2123. [71] Wood, G. C.; O'Sullivan, J. P.; Vaszko, B. Journal of the Electrochemical Society 1968, 115, 618-620. [72] Masuda, H.; Satoh, M. Japanese Journal of Applied Physics, Part 2: Letters 1996, 35, L126-L129. 197

PAGE 198

[73] Masuda, H.; Yamada, H.; Satoh, M.; Asoh, H.; Nakao, M.; Tamamura, T. Applied Physics Letters 1997, 71, 2770-2772. [74] Asoh, H.; Nishio, K.; Nakao, M.; Tamamura, T.; Masuda, H. Journal of the Electrochemical Society 2001, 148, B152-B156. [75] Diggle, J. W.; Downie, T. C.; Goulding, C. W. Chemical Reviews 1969, 69, 365-405. [76] Jessensky, O.; Muller, F.; Gosele, U. Applied Physics Letters 1998, 72, 1173-1175. [77] O'Sullivan, J. P.; Wood, G. C. Proceedings of the Royal Society of London, Series A: Mathematical, Physical and Engineering Sciences 1970, 317, 511-543. [78] Parkhutik, V. P.; Shershul'skii, V. I. Journal of Physics D: Applied Physics 1992, 25, 1258-1263. [79] Siejka, J.; Ortega, C. Journal of the Electrochemical Society 1977, 124, 883-891. [80] Hench, L. L.; West, J. K. Chemical Reviews 1990, 90, 33-72. [81] Livage, J.; Henry, M.; Sanchez, C. Progress in Solid State Chemistry 1988, 18, 259-341. [82] Patrissi, C. J.; Martin, C. R. Journal of the Electrochemical Society 1999, 146, 3176-3180. [83] Sides, C. R.; Martin, C. R. Modern Aspects of Electrochemistry 2007, 40, 75-126. [84] Sides, C. R.; Li, N.; Patrissi, C. J.; Scrosati, B.; Martin, C. R. MRS Bulletin 2002, 27, 604-607. [85] Lakshmi, B. B.; Patrissi, C. J.; Martin, C. R. Chemistry of Materials 1997, 9, 2544-2550. [86] Lin, J.; Brown, C. W. TrAC, Trends in Analytical Chemistry 1997, 16, 200-211. [87] Collinson, M. M. Mikrochimica Acta 1998, 129, 149-165. [88] Avnir, D. Accounts of Chemical Research 1995, 28, 328-334. [89] Walcarius, A. Electroanalysis 1998, 10, 1217-1235. [90] Brinker, C. J. Current Opinion in Colloid & Interface Science 1998, 3, 166-173. [91] Walcarius, A. Electroanalysis 2001, 13, 701-718. [92] Stoeber, W.; Fink, A.; Bohn, E. Journal of Colloid and Interface Science 1968, 26, 62-69. 198

PAGE 199

[93] Hench, L. L. Sol-Gel Silica-Properties, Processing and Technology Transfer; Noyes Publications: New Jersey, USA, 1998. [94] Iler, R. K. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and Biochemistry, John Wiley & Sons, 1979. [95] Pyell, U.; Stork, G. Fresenius' Journal of Analytical Chemistry 1992, 343, 576-581. [96] Jal, P. K.; Patel, S.; Mishra, B. K. Talanta 2004, 62, 1005-1028. [97] Kiselev, A. V.; Lygin, V. I. Infrared Spectra of Surface Compounds; Wiley: New York, 1975. [98] Kiselev, A. V. Kolloidnyi Zhurnal 1936, 2, 17-26. [99] Chuang, I. S.; Maciel, G. E. Journal of Physical Chemistry B 1997, 101, 3052-3064. [100] Grushka, E.; Kikta, E. J., Jr. Analytical Chemistry 1977, 49, 1004A-1008A, 1010A, 1012A, 1014A. [101] Van der Voort, P.; Gillis-D'Hamers, I.; Vansant, E. F. Journal of the Chemical Society, Faraday Transactions 1990, 86, 3751-3755. [102] Tundo, P.; Venturello, P.; Angeletti, E. Journal of the American Chemical Society 1982, 104, 6551-6555. [103] Plueddemann, E. P. Silane Coupling Agents, 2nd ed.; Plenum Publishing Corporation: New York, 1991. [104] Loy, D. A.; Baugher, B. M.; Baugher, C. R.; Schneider, D. A.; Rahimian, K. Chemistry of Materials 2000, 12, 3624-3632. [105] Arkles, B. Chemtech 1977, 7, 766-778. [106] Dias Filho, N. L.; Gushikem, Y. Separation Science and Technology 1997, 32, 2535-2545. [107] Lehninger, A. L. Principles of Biochemistry, 2nd ed.; Worth Inc.: New York, 1993. [108] Kemp, N. J.; Noble, R. D. Separation Science and Technology 1983, 18, 1147-1165. [109] Lakshmi, B. B.; Martin, C. R. Nature 1997, 388, 758-760. [110] Noble, R. D.; Koval, C. A. Materials Science of Membranes for Gas and Vapor Separation 2006, 411-435. 199

PAGE 200

[111] Wirgau, J. I.; Crumbliss, A. L. Dalton Transactions 2003, 3680-3685. [112] Zaghbani, A.; Tayeb, R.; Bonnamour, I.; Felix, C.; Vocanson, F.; Lamartine, R.; Dhahbi, M.; Seta, P. Journal of Membrane Science 2005, 258, 5-7. [113] Villamo, O.; Barboiu, C.; Barboiu, M.; Yau-Chun-Wan, W.; Hovnanian, N. Journal of Membrane Science 2002, 204, 97-110. [114] Shentu, B.; Nishide, H. Industrial & Engineering Chemistry Research 2003, 42, 5954-5958. [115] Tsuchida, E.; Nishide, H.; Ohyanagi, M. Journal of Macromolecular Science, Chemistry 1988, A25, 1327-1338. [116] Barboiu, M.; Guizard, C.; Hovnanian, N.; Palmeri, J.; Reibel, C.; Cot, L.; Luca, C. Journal of Membrane Science 2000, 172, 91-103. [117] Gherrou, A.; Kerdjoudj, H.; Molinari, R.; Drioli, E. Separation Science and Technology 2002, 37, 2317-2336. [118] Choi, H. W.; Kim, D. B.; Choi, D. K.; Ahn, B. S.; Kim, H. G.; Kim, H. S.; Lee, C. H.; Sung, J. Y. Journal of Membrane Science 2006, 279, 403-409. [119] Kaucher, M. S.; Peterca, M.; Dulcey, A. E.; Kim, A. J.; Vinogradov, S. A.; Hammer, D. A.; Heiney, P. A.; Percec, V. Journal of the American Chemical Society 2007, 129, 11698-11699. [120] Janeway, C. A., Travers, P.,Walport, M., Capra, J. D. Immunobiology: The Immune System in Health and Disease, 5th ed.; Garland Publishing Inc.: New York, 2001. [121] Burton, D. R. Trends in Biochemical Sciences 1990, 15, 64-69. [122] Gellissen, G. Production of Recombinant Proteins : Novel Microbial and Eukaryotic Expression Systems; Wiley-VCH: Weinheim, Germany, 2004. [123] Merten, O.-W., Mattanovich, D., Lang, C., Larsson, G., Neubauer, P., Porro, D., Postma, P., Teixeira de Mattos, J., Cole, J.A. Recombinant Protein Production With Prokaryotic and Eukaryotic Cells : A Comparative View on Host Physiology; Kluwer Academic Pub.: Semmering, Austria, 2001. [124] Hochuli, E.; Doebeli, H.; Schacher, A. Journal of Chromatography 1987, 411, 177-184. [125] Smith, M. C.; Furman, T. C.; Ingolia, T. D.; Pidgeon, C. Journal of Biological Chemistry 1988, 263, 7211-7215. [126] Porath, J.; Carlsson, J.; Olsson, I.; Belfrage, G. Nature 1975, 258, 598-599. 200

PAGE 201

[127] Porath, J. Trends in Analytical Chemistry 1988, 7, 254-259. [128] Eftink, M. R. Biophysical Journal 1994, 66, 482-501. [129] Eftink, M. R. Methods in Enzymology 2000, 323, 459-473. [130] Methods in Enzymology: Fluorescence Spectroscopy, 1997, 278, 1-628. [131] Lakowicz, J. R. Principles of Fluorescence Spectroscopy, 2nd ed.; Plenum Publishers, Inc.: New York, 1999. [132] Guilbault, G. G. Practical Fluorescence. Modern Monographs in Analytical Chemistry, 2nd ed.; Marcel Dekker, Inc.: New York, 1990. [133] Mirabella, F. M. Principles, Theory, and Practice of Internal Reflection Spectroscopy, in "Internal Reflection Spectroscopy Theory and Applications"; Marcel Dekker, Inc.: New York, 1993. [134] Skourlis, T. P.; McCullough, R. L. Journal of Applied Polymer Science 1994, 52, 1241-1248. [135] Harrick, N. J.; duPre, F. K. Applied Optics 1966, 5, 1739-1742. [136] Tan, S., Belanger, D. Journal of Physical Chemistry B 2005, 109, 23480-23490. [137] Watts, J. F., Wolstenholme, J. An Introduction to Surface Analysis by XPS and AES; John Wiley & Sons Inc., 2003. [138] Goldstein, J., Newbury, D. E., Joy, D. C., Lyman, C. E., Echlin, P., Lifshin, E., Sawyer, L., Michael, J. Scanning Electron Microscopy and X-ray Microanalysis, 3rd ed.; Springer Science + Business Media, Inc.: New York, 2003. [139] Chilcott, T. C.; Chan, M.; Gaedt, L.; Nantawisarakul, T.; Fane, A. G.; Coster, H. G. L. Journal of Membrane Science 2002, 195, 153-167. [140] Gaedt, L.; Chilcott, T. C.; Chan, M.; Nantawisarakul, T.; Fane, A. G.; Coster, H. G. L. Journal of Membrane Science 2002, 195, 169-180. [141] Park, S.-M.; Yoo, J.-S. Analytical Chemistry 2003, 75, 455A-461A. [142] Kharitonov, A. B.; Alfonta, L.; Katz, E.; Willner, I. Journal of Electroanalytical Chemistry 2000, 487, 133-141. [143] Bilitewski, U. Analytica Chimica Acta 2006, 568, 232-247. 201

PAGE 202

[144] Ye, M.; Jiang, X.; Feng, S.; Tian, R.; Zou, H. TrAC, Trends in Analytical Chemistry 2007, 26, 80-84. [145] Domon, B.; Aebersold, R. Science 2006, 312, 212-217. [146] Domon, B.; Broder, S. Journal of Proteome Research 2004, 3, 253-260. [147] Marengo, E.; Leardi, R.; Robotti, E.; Righetti, P. G.; Antonucci, F.; Cecconi, D. Journal of Proteome Research 2003, 2, 351-360. [148] Barroso, B.; Lubda, D.; Bischoff, R. Journal of Proteome Research 2003, 2, 633-642. [149] Masselon, C.; Pasa-Tolic, L.; Tolic, N.; Anderson, G. A.; Bogdanov, B.; Vilkov, A. N.; Shen, Y.; Zhao, R.; Qian, W.-J.; Lipton, M. S.; Camp, D. G., II; Smith, R. D. Analytical Chemistry 2005, 77, 400-406. [150] Nedelkov, D.; Tubbs, K. A.; Niederkofler, E. E.; Kiernan, U. A.; Nelson, R. W. Analytical Chemistry 2004, 76, 1733-1737. [151] Zilberstein, G.; Korol, L.; Antonioli, P.; Righetti, P. G.; Bukshpan, S. Analytical Chemistry 2007, 79, 821-827. [152] Iritani, E.; Mukai, Y.; Murase, T. Filtration & Separation 1997, 34, 967-973. [153] Ghosh, R.; Cui, Z. F. Journal of Membrane Science 2000, 167, 47-53. [154] Harkensee, D.; Kokpinar, O.; Walter, J.; Kasper, C.; Beutel, S.; Reif, O. W.; Scheper, T.; Ulber, R. Engineering in Life Sciences 2007, 7, 388-394. [155] Guo, W.; Ruckenstein, E. Journal of Membrane Science 2003, 211, 101-111. [156] Ramamoorthy, M.; Raju, M. D. Industrial & Engineering Chemistry Research 2001, 40, 4815-4820. [157] Suen, S.-Y.; Liu, Y.-C.; Chang, C.-S. Journal of Chromatography, B 2003, 797, 305-319. [158] O'Connor, A. J.; Pratt, H. R. C.; Stevens, G. W. Chemical Engineering Science 1996, 51, 3459-3477. [159] Ho, A. K.; Perera, J. M.; Dunstan, D. E.; Stevens, G. W.; Nystrom, M. AIChE Journal 1999, 45, 1434-1450. [160] Li, Q. Y.; Pepper, D. S. Journal of Membrane Science 1997, 136, 181-190. [161] Nystrom, M.; Aimar, P.; Luque, S.; Kulovaara, M.; Metsamuuronen, S. Colloids and Surfaces, A: Physicochemical and Engineering Aspects 1998, 138, 185-205. 202

PAGE 203

[162] Nakatsuka, S.; Michaels, A. S. Journal of Membrane Science 1992, 69, 189-211. [163] Saxena, A.; Shahi, V. K. Journal of Membrane Science 2007, 299, 211-221. [164] Aravind, U. K.; Mathew, J.; Aravindakumar, C. T. Journal of Membrane Science 2007, 299, 146-155. [165] Poetz, O.; Schwenk, J. M.; Kramer, S.; Stoll, D.; Templin, M. F.; Joos, T. O. Mechanisms of Ageing and Development 2005, 126, 161-170. [166] Poetz, O.; Ostendorp, R.; Brocks, B.; Schwenk, J. M.; Stoll, D.; Joos, T. O.; Templin, M. F. Proteomics 2005, 5, 2402-2411. [167] Langmuir, I. Journal of the American Chemical Society 1918, 40, 1361-1402. [168] Yang, L.; Chen, P. Journal of Membrane Science 2002, 205, 141-153. [169] Weiss, M.; Welsch, W.; Schickfus, M. v.; Hunklinger, S. Analytical Chemistry 1998, 70, 2881-2887. [170] Saha, K.; Bender, F.; Gizeli, E. Analytical Chemistry 2003, 75, 835-842. [171] Igarashi, T.; Ishigatsubo, Y.; Ohno, S.; Shirai, A.; Aoki, I.; Okubo, T.; Klinman, D. M. Clinical Immunology and Immunopathology 1994, 72, 307-311. [172] Klinman, D. M. ; Banks, S.; Hartman, A.; Steinberg, A. D. Journal of Clinical Investigations 1988, 82, 652-657. [173] Gueell, C.; Davis, R. H. Journal of Membrane Science 1996, 119, 269-284. [174] Mueller, J.; Davis, R. H. Journal of Membrane Science 1996, 116, 47-60. [175] Ho, C.-C.; Zydney, A. L. Journal of Membrane Science 1999, 155, 261-275. [176] Jones, K. L.; O'Melia, C. R. Journal of Membrane Science 2000, 165, 31-46. [177] Dechadilok, P.; Deen, W. M. Industrial & Engineering Chemistry Research 2006, 45, 6953-6959. [178] Bhattacharjee, S.; Sharma, A.; Bhattacharya, P. K. Industrial & Engineering Chemistry Research 1996, 35, 3108-3121. [179] Johnson, C. P.; Jensen, I. E.; Prakasam, A.; Vijayendran, R.; Leckband, D. Bioconjugate Chemistry 2003, 14, 974-978. 203

PAGE 204

[180] Dancette, O. P.; Taboureau, J.-L.; Tournier, E.; Charcosset, C.; Blond, P. Journal of Chromatography, B: Biomedical Sciences and Applications 1999, 723, 61-68. [181] Rutherford, S. W.; Do, D. D. Adsorption 1997, 3, 283-312. [182] Paul, D. R. Journal of Polymer Science, Polymer Physics Edition 1969, 7, 1811-1818. [183] Tsuru, T.; Izumi, S.; Yoshioka, T.; Asaeda, M. AIChE Journal 2000, 46, 565-574. [184] Nomura, T.; Nakao, S.; Kimura, S. Kagaku Kogaku Ronbunshu 1987, 13, 811-817. [185] Chandler, D. Nature 2005, 437, 640-647. [186] Lin, F.-Y.; Chen, W.-Y.; Hearn, M. T. W. Journal of Molecular Recognition 2002, 15, 55-93. [187] Palma, E.; Klapper, D. G.; Cho, M. J. Pharmaceutical Research 2005, 22, 122-127. [188] Norde, W.; Lyklema, J. Journal of Colloid and Interface Science 1979, 71, 350-366. [189] Norde, W. Advances in colloid and interface science 1986, 25, 267-340. [190] Norde, W.; MacRitchie, F.; Nowicka, G.; Lyklema, J. Journal of Colloid and Interface Science 1986, 112, 447-456. [191] Hamilton, R. G. Production and epitope location of monoclonal antibodies to the human IgG subclasses. In Molecular Aspects of Immunoglobulin Subclasses; Pergamon Press: Oxford, 1990, pp 79-91. [192] Schmuhl, R.; Keizer, K.; van den Berg, A.; ten Elshof, J. E.; Blank, D. H. A. Journal of Colloid and Interface Science 2004, 273, 331-338. [193] Saksena, S.; Zydney, A. L. Biotechnology and Bioengineering 1994, 43, 960-968. [194] Zydney, A. L.; Pujar, N. S. Colloids and Surfaces, A: Physicochemical and Engineering Aspects 1998, 138, 133-143. [195] Lopez, M. F.; Melov, S. Circulation Research 2002, 90, 380-389. [196] Sun, L.; Crooks, R. M. Langmuir 1999, 15, 738-741. [197] Bath, B. D.; Lee, R. D.; White, H. S.; Scott, E. R. Analytical Chemistry 1998, 70, 1047-1058. [198] Baltus, R. E. Journal of Membrane Science 1997, 130, 157-172. 204

PAGE 205

[199] Galiatsatou, P.; Kanellopoulos, N. K.; Petropoulos, J. H. Journal of Membrane Science 2006, 280, 634-642. [200] Funke, H. H.; Noble, R. D.; Koval, C. A. Journal of Membrane Science 1993, 82, 229-236. [201] Smith, K. A.; Meldon, J. H.; Colton, C. K. AIChE Journal 1973, 19, 102-109. [202] Teramoto, M. Industrial & Engineering Chemistry Research 1994, 33, 2161-2171. [203] Vieth, W. R.; Sladek, K. J. Journal of Colloid Science 1965, 20, 1014-1033. [204] Paul, D. R.; Koros, W. J. Journal of Polymer Science, Polymer Physics Edition 1976, 14, 675-685. [205] Vieth, W. R.; Howell, J. M.; Hsieh, J. H. Journal of Membrane Science 1976, 1, 177-220. [206] Petropoulos, J. H. Journal of Polymer Science, Polymer Physics Edition 1970, 8, 1797-1801. [207] Mauze, G. R.; Stern, S. A. Journal of Membrane Science 1984, 18, 99-109. [208] Riedl, C. M.; Spencer, H. G. Journal of Applied Polymer Science 1990, 41, 1685-1693. [209] Noble, R. D. Journal of Membrane Science 1990, 50, 207-214. [210] Kanehashi, S.; Nagai, K. Journal of Membrane Science 2005, 253, 117-138. [211] Tsujita, Y. Progress in Polymer Science 2003, 28, 1377-1401. [212] Barrer, R. M.; Barrie, J. A.; Slater, J. Journal of Polymer Science 1958, 27, 177-197. [213] Stern, S. A.; Zhou, S.; Araux-Lara, J. L.; Ware, B. R. Journal of Polymer Science, Part C: Polymer Letters 1989, 27, 427-431. [214] Koros, W. J.; Paul, D. R. Journal of Polymer Science, Polymer Physics Edition 1978, 16, 1947-1963. [215] Huh, J.-K.; Song, D.-I.; Jeon, Y.-W. Separation Science and Technology 1999, 34, 571-586. [216] Huh, J.-K.; Song, D.-I.; Jeon, Y.-W. Separation Science and Technology 2000, 35, 243-259. [217] Wang, J.-S.; Kamiya, Y. Journal of Membrane Science 1999, 154, 25-32. 205

PAGE 206

[218] Dixon-Garrett, S. V.; Nagai, K.; Freeman, B. D. Journal of Polymer Science, Part B: Polymer Physics 2000, 38, 1078-1089. [219] Dixon-Garrett, S. V.; Nagai, K.; Freeman, B. D. Journal of Polymer Science, Part B: Polymer Physics 2000, 38, 1461-1473. [220] Meares, P. Journal of the American Chemical Society 1954, 76, 3415-3422. [221] Budd, P. M.; McKeown, N. B.; Fritsch, D. Journal of Materials Chemistry 2005, 15, 1977-1986. [222] Huang, S. H.; Margolies, M. N.; Rothstein, A. M.; Siekevitz, M.; Gefter, M. L. Pharmacology Reviews 1982, 34, 43. [223] Sem, D. S.; McNeeley, P. A.; Linnik, M. D. Archives of Biochemistry and Biophysics 1999, 372, 62-68. [224] Petropoulos, J. H. Advances in Polymer Science 1985, 64, 93-142. [225] Kempe, H.; Persson, P.; Axelsson, A.; Nilsson,B.; Zacchi, G. Biotechnology Bioengineer 2006, 93, 656-664. [226] Bosma, J. C.; Wesselingh, J. A. Journal of Chromatography B 2000, 743, 169. [227] Burczak, K.; Fujisato, T.; Hatada, M.; Ikada, Y. Biomaterials 1994, 15, 231-238. [228] La Flamme, K. E.; Mor, G.; Gong, D.; La Tempa, T.; Fusaro, V. A.; Grimes, C. A.; Desai, T. A. Diabetes Technology & Therapeutics 2005, 7, 684-694. [229] Leoni, L.; Boiarski, A.; Desai, T. A. Biomedice Microdevices 2002, 4, 131-139. [230] Bird, R. B.; Steward, W. E.; Lightfoot, E. N. Transport Phenomena, John Wiley, New York, NY, 1960. [231] Cussler, E. L. Diffusion. Mass Transfer in Fluid systems. Cambridge University Press, Cambridge, UK, 1997. [232] Edward, J. T. Journal of Chemical Education 1970, 47, 261-270. [233] Pluen, A.; Netti, P. A.; Jain, R. K.; Berk, D. A. Biophysical Journal 1999, 77, 542-552. [234] Terazima, M. Physical Chemistry Chemical Physics 2006, 8, 545-557. [235] Nishida, S.; Nada, T.; Terazima, M. Biophysical Journal 2004, 87, 2663-2675. [236] Moschou, E. A.; Bachas, L. G.; Deo, S. K. Analytical Chemistry 2006, 78, 6692-6700. 206

PAGE 207

[237] Hentz, N. G.; Daunert, S. Analytical chemistry 1996, 68, 3939-3944. [238] Lager, I.; Looger, L. L.; Hilpert, M.; Lalonde, S.; Frommer, W. B. Journal of Biological Chemistry 2006, 281, 30875-30883. [239] Dwyer, M. A.; Hellinga, H. W. Current Opinion in Structural Biology 2004, 14, 495-504. [240] Lager, I.; Fehr, M.; Frommer, W. B.; Lalonde, S. FEBS Letters 2003, 553, 85-89. [241] Nilsson, B.; Forsberg, G.; Moks, T.; Hartmanis, M.; Uhlen, M. Current Opinion in Structural Biology 1992, 2, 569-575. [242] Seidler, A. Protein Engineering 1994, 7, 1277-1280. [243] Zhang, Z.; Tong, K. T.; Belew, M.; Pettersson, T.; Janson, J. C. Journal of chromatography 1992, 604, 143-155. [244] Carlsson, J.; Porath, J.; Lonnerdal, B. FEBS letters 1977, 75, 89-92. [245] Chadha, K. C.; Grob, P. M.; Mikulski, A. J.; Davis, L. R.; Sulkowski, E. Journal of General Virology 1979, 43, 701-706. [246] Chaga, G.; Andersson, L.; Ersson, B.; Porath, J. Biotechnology and applied biochemistry 1989, 11, 424-431. [247] McCluskey, A. J.; Poon, G. M. K.; Gariepy, J. Protein Science 2007, 16, 2726-2732. [248] Lee, K.-M.; Ma, K.-W.; Shaw, P.-C.; Wong, K.-B. Analytical Biochemistry 2006, 358, 152-154. [249] Cantin, G. T.; Yi, W.; Lu, B.; Park, S. K.; Xu, T.; Lee, J.-D.; Yates, J. R. Journal of Proteome Research 2008, 7, 1346-1351. [250] Kermani, B. G.; Kozlov, I.; Melnyk, P.; Zhao, C.; Hachmann, J.; Barker, D.; Lebl, M. Sensors and Actuators, B: Chemical 2007, B125, 149-157. [251] Ueda, E. K. M.; Gout, P. W.; Morganti, L. Journal of Chromatography, A 2003, 988, 1-23. [252] Pearson, R. G. Journal of Chemical Education 1968, 45, 581-587. [253] Arnold, F. H. Biotechnology 1991, 9, 151-156. [254] Sulkowski, E. BioEssays 1989, 10, 170-175. [255] Min, C.; Verdine, G. L. Nucleic Acids Research 1996, 24, 3806-3810. 207

PAGE 208

[256] Hochuli, E.; Doebeli, H.; Schacher, A. Journal of Chromatography 1987, 411, 177-184. [257] Gaberc-Porekar, V.; Menart, V. Chemical Engineering & Technology 2005, 28, 1306-1314. [258] Sundberg, R. J.; Martin, R. B. Chemical Reviews 1974, 74, 471-517. [259] Gasparac, R.; Martin, C. R.; Stupnisek-Lisac, E.; Mandic, Z. Journal of the Electrochemical Society 2000, 147, 991-998. [260] Meier, M. A. R.; Schubert, U. S. Journal of Polymer Science, Part A: Polymer Chemistry 2003, 41, 2964-2973. [261] Herrero, M.; Tiemblo, P.; Reyes-Labarta, J.; Mijangos, C.; Reinecke, H. Polymer 2002, 43, 2631-2636. [262] Tiemblo, P.; Guzman, J.; Riande, E.; Mijangos, C.; Reinecke, H. Macromolecules 2002, 35, 420-424. [263] Dias Filho, N. L. Mikrochimica Acta 1999, 130, 233-240. [264] Kemlo, J. A.; Shepherd, T. M. Chemical Physics Letters 1977, 47, 158-162. [265] Varnes, A. W.; Dodson, R. B.; Wehry, E. L. Journal of the American Chemical Society 1972, 94, 946-950. [266] de Silva, A. P.; Gunaratne, H. Q. N.; Gunnlaugsson, T.; Huxley, A. J. M.; McCoy, C. P.; Rademacher, J. T.; Rice, T. E. Chemical Reviews 1997, 97, 1515-1566. [267] Fabbrizzi, L.; Licchelli, M.; Pallavicini, P.; Sacchi, D.; Taglietti, A. Analyst 1996, 121, 1763-1768. [268] Huang, F.; Tian, Y.; Chen, C.-Y.; Cheng, Y.-J.; Young, A. C.; Jen, A. K. Y. Journal of Physical Chemistry C 2007, 111, 10673-10681. [269] McFarland, S. A.; Finney, N. S. Journal of the American Chemical Society 2001, 123, 1260-1261. [270] You, C.-C.; Hippius, C.; Gruene, M.; Wuerthner, F. Chemistry--A European Journal 2006, 12, 7510-7519. [271] Shen, Y.; Mackey, G.; Rupcich, N.; Gloster, D.; Chiuman, W.; Li, Y.; Brennan, J. D. Analytical Chemistry 2007, 79, 3494-3503. [272] Hartwig, J. F.; Pil, P. M.; Lippard, S. J. Journal of the American Chemical Society 1992, 114, 8292-8293. 208

PAGE 209

[273] Thornton, N. B.; Schanze, K. S. New Journal of Chemistry 1996, 20, 791-800. [274] McFarland, S. A.; Finney, N. S. Journal of the American Chemical Society 2002, 124, 1178-1179. [275] Cotton, F. A., Wilkinson, G. Advanced Inorganic Chemistry, 5th ed.; John Wiley & Sons: NY, 1987. [276] Housecroft, C. E., Sharpe, A. G. Inorganic Chemistry; Prentice Hall: NY, 2001. [277] Silva, H. A. S.; Carlos, R. M.; Camargo, A. J.; Picchi, C. M. C.; Santos, R. H. d. A.; McGarvey, B. R.; Franco, D. W. Inorganica Chimica Acta 2004, 357, 3147-3154. [278] O'Sullivan, J. P.; Hockey, J. A.; Wood, G. C. Transactions of the Faraday Society 1969, 65, 535-541. [279] Kimura, Y.; Shiraki, H.; Ishibashi, K.-i.; Ishii, H.; Itaya, K.; Niwano, M. Journal of the Electrochemical Society 2006, 153, C296-C300. [280] Abbott, T. P.; Nabetani, H.; Sessa, D. J.; Wolf, W. J.; Liebman, M. N.; Dukor, R. K. Journal of Agricultural and Food Chemistry 1996, 44, 2220-2224. [281] Malhotra, V. M.; Jasty, S.; Mu, R. Applied Spectroscopy 1989, 43, 638-645. [282] Spinner, E. Journal of the Chemical Society 1960, 1237-1242. [283] Spinner, E. Spectrochimica Acta 1959, 95-109. [284] Spinner, E. Journal of the Chemical Society 1962, 3127-3128. [285] Katritzky, A. R.; Witek, R. M.; Rodriguez-Garcia, V.; Mohapatra, P. P.; Rogers, J. W.; Cusido, J.; Abdel-Fattah, A. A. A.; Steel, P. J. Journal of Organic Chemistry 2005, 70, 7866-7881. [286] Hayashi, T.; Mukamel, S. Journal of Physical Chemistry B 2007, 111, 11032-11046. [287] Waldron, R. D. Physical Review 1955, 99, 1727-1735. [288] Feenan, K.; Fowles, G. W. A. Journal of the Chemical Society 1965, 2449-2451. [289] Kyotani, T.; Sato, K.; Mizuno, T.; Kakui, S.; Aizawa, M.; Saito, J.; Ikeda, S.; Ichikawa, S.; Nakane, T. Analytical Sciences 2005, 21, 321-325. [290] Do, J.; Ahn, C. H. Lab on a Chip 2008, 8, 542-549. 209

PAGE 210

[291] VanWiggeren, G. D.; Bynum, M. A.; Ertel, J. P.; Jefferson, S.; Robotti, K. M.; Thrush, E. P.; Baney, D. M.; Killeen, K. P. Sensors and Actuators, B: Chemical 2007, B127, 341-349. [292] Medyantseva, E. P.; Varlamova, R. M.; Gimaletdinova, D. A.; Fattakhova, A. N.; Budnikov, G. K. Journal of Analytical Chemistry 2008, 63, 275-279. [293] Borisov, S. M.; Wolfbeis, O. S. Chemical Reviews 2008, 108, 423-461. [294] Goel, G.; Kumar, A.; Puniya, A. K.; Chen, W.; Singh, K. Journal of Applied Microbiology 2005, 99, 435-442. [295] Zeineldin, R.; Piyasena, M. E.; Sklar, L. A.; Whitten, D.; Lopez, G. P. Langmuir 2008, 24, 4125-4131. [296] Ko, S.; Grant, S. A. Biosensors & Bioelectronics 2006, 21, 1283-1290. [297] Liu, X.; Tan, W. Analytical Chemistry 1999, 71, 5054-5059. [298] Wu, P.; Brand, L. Analytical Biochemistry 1994, 218, 1-13. [299] Kagan, C. R.; Murray, C. B.; Nirmal, M.; Bawendi, M. G. Physical Review Letters 1996, 76, 3043. [300] Cheung, H. C.; Dong, W.-J. Reviews in Fluorescence 2006, 3, 445-462. [301] Schuler, B. Chemphyschem 2005, 6, 1206-1220. [302] Invitrogen. Amine-Reactive Probes, MP00143, Molecular Probes: Eugene, OR, 2007. http://probes.invitrogen.com/media/pis/mp00143.pdf [303] http://probes.invitrogen.com/handbook/boxes/0422.html [304] Piunno, P. A. E.; Krull, U. J.; Hudson, R. H. E.; Damha, M. J.; Cohen, H. Analytical Chemistry 1995, 67, 2635-2643. [305] Wang, J.; Xu, D.; Kawde, A. N.; Polsky, R. Analytical chemistry 2001, 73, 5576-5581. [306] Taton, T. A.; Mirkin, C. A.; Letsinger, R. L. Science 2000, 289, 1757-1760. [307] Drummond, T. G.; Hill, M. G.; Barton, J. K. Nature Biotechnology 2003, 21, 1192-1199. [308] Su, X.; Robelek, R.; Wu, Y.; Wang, G.; Knoll, W. Analytical Chemistry 2004, 76, 489-494. 210

PAGE 211

[309] Maruyama, K.; Mishima, Y.; Minagawa, K.; Motonaka, J. Analytical Chemistry 2002, 74, 3698-3703. [310] Zhao, X.; Tapec-Dytioco, R.; Tan, W. Journal of the American Chemical Society 2003, 125, 11474-11475. [311] Landers, J. P. Analytical Chemistry 2003, 75, 2919-2927. [312] Miao, W.; Bard, A. J. Analytical Chemistry 2004, 76, 5379-5386. [313] Castro, A.; Dalvit, D. A. R.; Paz-Matos, L. Analytical Chemistry 2004, 76, 4169-4174. [314] Wang, R.; Minunni, M.; Tombelli, S.; Mascini, M. Biosensors & Bioelectronics 2004, 20, 598-605. [315] Bui, M.-P. N.; Baek, T. J.; Seong, G. H. Analytical and Bioanalytical Chemistry 2007, 388, 1185-1190. [316] Watson, J. D.; Crick, F. H. C. Nature 1953, 171, 737-738. [317] Marmur, J.; Doty, P. Journal of molecular biology 1961, 3, 585-594. [318] Nielsen, P. E.; Egholm, M.; Berg, R. H.; Buchardt, O. Science 1991, 254, 1497-1500. [319] Nielsen, P. E. Current opinion in biotechnology 1999, 10, 71-75. [320] Wang, J. Biosensors & Bioelectronics 1998, 13, 757-762. [321] Arlinghaus, H. F.; Kwoka, M. N.; Jacobson, K. B. Analytical Chemistry 1997, 69, 3747-3753. [322] Wang, J.; Palecek, E.; Nielsen, P. E.; Rivas, G.; Cai, X.; Shiraishi, H.; Dontha, N.; Luo, D.; Farias, P. A. M. Journal of the American Chemical Society 1996, 118, 7667-7670. [323] Aoki, H.; Umezawa, Y. Analyst 2003, 128, 681-685. [324] Abel, A. P.; Weller, M. G.; Duveneck, G. L.; Ehrat, M.; Widmer, H. M. Analytical Chemistry 1996, 68, 2905-2912. [325] Methods in Enzymology: Aviding-Biotin Technology, 1990, 184, 1-617. [326] Macdonald, J. R. Impedance Spectroscopy; Emphasizing Solid Materials and Systems; Wiley Interscience publications: NY, 1987. [327] Winkler, B. H.; Baltus, R. E. Journal of Membrane Science 2003, 226, 75-84. 211

PAGE 212

BIOGRAPHICAL SKETCH Hctor Mario Caicedo was born in Cali, Colombia. He graduated with a B.S. in chemistry from the Universidad del Valle, in Cali. For his bachelors thesis he conducted research in analytical chemistry focusing on environmental pollutants under the direction of Dr. Nelly de Palacios (his mentor in chemistry), developing an analytical methodology for the evaluation of chromium in urine and blood by using electrothermal atomic absorption spectroscopy (ETAAS). After that, he worked as a chemist in the industrial field, for a combined time of almost four years, for both a biotechnology and pharmaceutical company. Then, he decided to go back to academia to conduct his Master of Science in analytical chemistry at the Universidad del Valle. In the summer of 2000, he visited the University of Florida as an international exchange student, conducting research in electrochemistry, as a part of his masters research work, using carbon fiber ultramicroelectrodes under the direction of Dr. Anna Brajter-Toth. After returning to the Universidad del Valle, he conducted research in electroanalysis of xanthine by using carbon fiber ultramicroelectrodes with surfaces electrochemically activated and polymer-modified, in the group of Dr. Alonso Jaramillo. Once he graduated with a M.S. in chemistry from the University del Valle, he continued his graduate studies in the Department of Chemistry at the University of Florida to pursue his Ph.D. He worked under the supervision of Dr. Charles R. Martin, where he conducted research in bioanalytical applications of affinity-based nanotube membranes for sensing and separations. He completed his final work in April 2008, when he received a Doctor of Philosophy degree. 212