|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 PROTEIN SEPARATION IN PL ASTIC MICROFLUIDIC DEVICES By CHAMPAK DAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007
2 2007 Champak Das
3 To my Parents
4 ACKNOWLEDGMENT I would like to take this oppor tunity to express my deepes t gratitude to my graduate advisor, Dr. Hugh Fan, for guiding me in my re search work and his patience in going through every draft I gave him. Special thanks go to Dr. Cattafesta, Dr. Ch auhan, Dr.Hahn, and Dr. Denslow for serving in my co mmittee and providing valuable suggestion to me for my dissertation. I would like to tha nk Mr. Zheng Xia for fabricating silicon and glass masters, Mr. Carl Fredrickson for plastic device optimizati on, undergraduate students Mr. Fernando Tavares and Mr. Andrew Simon for helping me with devi ce fabrication. I also would like to thank Dr. Alexander Stoyanov and Dr. Jiyou Zhang for helping me in understanding the fine points in electrophoresis and surface coating, as well as fru itful discussion on improving the resolution in two-dimensional separation. Special thanks go to my fellow gradua te students Qian Mei, Ruba Knouf, Karthik Pitchaimani, Jackie Viren and undergraduate students Corey Walker and Joey Wilson for the wonderful time I had with them in the lab. Thanks are also due to my friends Anales, Alpana, Jaya and Nadim. Finally, I would like to thank my family for their endless love, support and encouragement in the process of my learning. This work was supported in part by Dr. Fans startup fund from University of Florida and the grant from Army Research Office (ARO).
5 TABLE OF CONTENTS page ACKNOWLEDGMENT................................................................................................................. .4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES................................................................................................................ .........8 ABSTRACT....................................................................................................................... ............11 CHAPTER 1 INTRODUCTION................................................................................................................. .13 1.1 MEMS....................................................................................................................... ........13 1.2 Microfluidics.............................................................................................................. .......15 1.3 Electrophoresis............................................................................................................ .....19 1.3.1 Isoelectric Focusing................................................................................................19 1.3.2 SDS-PAGE Separation...........................................................................................21 1.4 Electrophoresis in a Chip.................................................................................................. 21 1.5 Current Challenges......................................................................................................... ..23 1.6 Objectives................................................................................................................. ........25 2 INSTRUMENTATION AND DEVICE FABRICATION.....................................................28 2.1 LIF Setup and Characterization........................................................................................28 2.1.1 Introduction............................................................................................................2 8 2.1.2 Instrumentation.......................................................................................................29 2.1.3 Photo Bleaching Effects.........................................................................................30 188.8.131.52 Detection limit..............................................................................................34 184.108.40.206 Optical correction.........................................................................................35 2.2 Device Fabrication......................................................................................................... ...37 2.2.1 Introduction............................................................................................................3 7 2.2.2 Design and Fabrication...........................................................................................38 2.3 Conclusion................................................................................................................. .......39 3 THEORITICAL AND EXPERIME NTAL RESULTS OF IEF.............................................46 3.1 Introduction............................................................................................................... ........46 3.2 Materials and Methods.....................................................................................................4 6 3.3 Theory of IEF.............................................................................................................. .....47 3.4 Experimental Results....................................................................................................... .54 3.4.1 Effects of Separation Medium................................................................................55 3.4.2 Effects of Separation Length..................................................................................57 3.4.3 Effects of Separation Voltage.................................................................................59 3.4.4 Focusing Time........................................................................................................59
6 3.4.5 pH Gradient Compression......................................................................................61 3.5 Conclusion................................................................................................................. .......62 4 TWO DIMENSIONAL SEPARATION OF PROTEINS......................................................71 4.1 Introduction............................................................................................................... ........71 4.2 Materials and Methods.....................................................................................................7 2 4.3 Polyacrylamide Gel Valve................................................................................................74 4.4 IEF in First Dimension..................................................................................................... 76 4.5 Numerical Simulation....................................................................................................... 78 4.7 SDS-PAGE Separation.....................................................................................................83 4.8 Two-dimensional Separation............................................................................................84 4.9 Conclusion................................................................................................................. .......89 5 CONCLUSION AND FU TURE DIRECTIONS..................................................................108 5.1 Conclusion................................................................................................................. .....108 5.2 Future Direction........................................................................................................... ...109 APPENDIX TRANSIENT SOLUTION OF IEF..................................................................112 LIST OF REFERENCES............................................................................................................. 123 BIOGRAPHICAL SKETCH.......................................................................................................130
7 LIST OF TABLES Table page 3-1 The effects of IEF dist ance on separation resolution.........................................................63 4-1 Effects of gel thickness and mask width on photo polymerization process.....................90 4-2 Comparison of resolution between di fferent channel lengths and 2-D device..................90 4-3 Five different labeled prot eins used for 2-D separation.....................................................91
8 LIST OF FIGURES Figure page 1-1 Representation of electrical double layer near a surface...................................................26 1-2 Illustration of basic process of IEF....................................................................................26 1-3 Typical microfluidic devices used for electrophoresis work.............................................27 1-4 Different microfluidic devices for 2-Dimensional separation...........................................27 2-1 LIF imaging system fo r protein separations......................................................................40 2-2 The fluorescence intensity values measur ed in microchannels with different laser power. The exposure of CCD camera is 100 ms...............................................................41 2-3 Temporal profile of photo bleaching effects......................................................................41 2-4 The cylindrical lens compresses the beam in one axis......................................................42 2-5 Spatial profile of photo bleaching effects..........................................................................42 2-7 Optical correction for rem oving Gaussian noise background............................................44 2-8 The layout of microfluidic devices for protein separation.................................................44 2-9 E-form used for this research.............................................................................................4 5 2-10 General methods for plastic device fabrication.................................................................45 3-1 Illustration of separation pr ocess for closely spaced peaks...............................................64 3-2 The temporal images of IEF separa tions of GFP in polyacrylamide gel...........................65 3-3 IEF of different proteins in gel and linear polymer...........................................................65 3-4 Comparison of front speed fo r different separation medium.............................................66 3-5 IEF of GFP, RPE in HPC/HEC linear polymer.................................................................66 3-6 The effects of the separation distance on IEF separation..................................................67 3-7 Separation resolution based on p eak width and separation distance.................................68 3-8 The effects of separation voltage on IEF. .......................................................................69 3-9 The relationship between the focusing tim e and the inverse of the electric field strength....................................................................................................................... ........69
9 3-10 Comparison of IEF electropherograms of GFP and RPE between pH 3-10 and pH 46 gradients. .................................................................................................................. ......70 4-1 Layout of a microfluidic device fo r two-dimensional protein separation..........................91 4-2 Photo polymerizati on in microchannels.............................................................................92 4-3 Polymerization time with acrylamide concentration.........................................................92 4-4 Polymerization time with HCPK concentration................................................................93 4-5 The schematic of photo polymerization of acrylamide inside the microchannel..............93 4-6 Micrograph of the valve arrays formed by in situ gel polymerization. Polymerized gels are dyed for easy visualization...................................................................................94 4-7 IEF in different microfluidic devices.................................................................................94 4-8 Isoelectric focusing of 4 proteins (Par albumin, Ovalbumin, BSA, and GFP) in cross channel........................................................................................................................ .......95 4-9 2-D device used for numerical simulation.........................................................................96 4-10 Transfer of negative species from cro ss channel to parallel channels due to an applied electric field of 100 V/cm.....................................................................................97 4-11 Transfer of protein plugs into parallel ch annel due to applicati on of electric field...........98 4-12 Protein migration pattern in all 29 parallel channels.........................................................98 4-13 Electric field distributions at differe nt location across the parallel channel......................99 4-14 Experimental results of location of protein front with respect to time............................100 4-15 SDS-protein complex migrat ion pattern in Gel due to application of electric field........101 4-16 IEF for RPE with and without SDS in 5 cm channel. The electr ic field is 500 V and pH gradient is 3-10...........................................................................................................1 01 4-17 The transfer of GFP and RPE in second dimension after IEF is performed in first dimension...................................................................................................................... ...102 4-18 2-Dimensional migration of a single prot ein. A) IEF of BSA in cross channel..............102 4-19 2-Dimensional separati on of three proteins.....................................................................103 4-20 2-Dimensional separation of three prot eins (BSA, Ovalbumin and Trypsin) in lower pH gradient (3-5 pH) during IEF operation.....................................................................104
10 4-21 2-Dimensional separations of three prot eins (BSA, Ovalbumin and Hemoglobin) in 3-10 pH during IEF operation..........................................................................................105 4-22 2-Dimensional separation proteins for proteins BSA (1), carbonic anhydrase (2), ovalbumin(3) and trypsin (4)...........................................................................................106 4-23 2-D map for BSA, ovalbumin, hemo globin and carbonic anhydrase in 10% acrylamide gel................................................................................................................. .107 5-1 The mask size is increased such that 500 m on both the sides of cross channel is left unpolymerized after UV exposure...................................................................................111 A-1 Assumed charge distributi on for different CA/ proteins in IEF focusing process for nonlinear case................................................................................................................. ..119 A-2 Assumed charge distributi on for different CA/ proteins in IEF focusing process...........119 A-3 Evolution of Gaussian peak with time for a particular ampholyte..................................120 A-4 Numerical results for front speed of CA / proteins with time at different. applied electric field................................................................................................................. ....121 A-5 Conductivity plotted with respect to time at different el ectric fields. .............................122
11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PROTEIN SEPARATION IN PL ASTIC MICROFLUIDIC DEVICES By Champak Das August 2007 Chair: Hugh Fan Major: Mechanical Engineering Conventional two-dimensional gel electropho resis has been extensively used for proteomics research, including disc overing biomarkers associated with a disease. However, the process is time consuming and labor intensive. To address the limitation, a plastic microfluidic device (1x 3) is developed that is capable of doing the same operation in a much shorter time with less labor. The devices are fabricated by compression molding, a similar technique for manufacturing compact discs. Fi rst two different separation mech anisms isoelectric focusing and polyacrylamide gel electrophoresis (PAGE) are se parately demonstrated in microdevices. Isoelectric focusing (IEF) is optimized in terms of the applied voltage an d separation medium. It is demonstrated that IEF is esse ntially independent of channel le ngth, allowing mini aturization of separation apparatus. The time required for IEF is also drastically reduced when the channel length is reduced. It takes only 3 minutes in a 2 cm channel compared to 10 hours in conventional apparatus. Further integration is achieved by selective photo polymerization inside microchannels, obtaining a reliable interface that prevents one separation medium from contaminating with the other medium. PAGE take s another 5 minutes to perform. When these two separation mechanisms are in tegrated, it takes about 15 minutes from loading the samples to finishing the experiment whereas the conventional system takes more than a day to finish. A
12 laser induced whole channel fluorescence imaging sy stem is assembled for detecting the proteins separated in the devices. The se quential images obtained from the imaging system helps in understanding the dynamic nature of protein separa tion such as IEF. The same imaging system can be used for single point detection for tw odimensional separation. The detection limit for this system was found to be around 1 nM of fluorescien solution and 0.03 ng/ l when Green fluorescent protein is used in isoelectric focusi ng. Fluorescently labeled proteins are used to demonstrate the viability of the miniaturized twodimensi onal protein separation system.
13 CHAPTER 1 INTRODUCTION Technology has come a long way from Ecke rt and Mauchys 1946s Eniac, a 50-ton computer that filled an entire room, to a palm top of todays world, many times powerful than its predecessor. System miniaturiza tion has benefits, including less power required, portability and, in most cases, fast and reliable operation. Miniatur ized systems can be optimally linked together in order to achieve specific mechan ical, electrical, optical or chem ical functions either one at a time or all combined together. In a broad sense, they take inputs from external stimulus, make decisions and give output. Miniaturized systems, though very exciting in terms of size and integration of several functionalities in one small packag e, also prove to be a challenge to fabricate and integrate with other functional parts. As the parts get sma ller, newer physical phenomena inapplicable or negligible to thei r larger counterparts start creating problems. Neverthe less, newer applications have been realized by finding an intelligent way to avoid these problems or making use of them. Although miniaturization has affected many facets of science, I will, in this thesis, focus on two areas, microelectromechanical systems (MEMS) and microfluidics. Subjects covered by these two terms overlap sometimes; but their definition and differences will be explained below. Since the application of this thesis is primarily rela ted to electrophoresis, I will also review the technique briefly. 1.1 MEMS A microelectromechanical system (MEMS) is an integration of sensors, actuators and other electrical or mechanical components, usually fabr icated in a silicon substr ate. This is achieved through microfabrication technology originally developed in the 60s for computer chips. With improvement of microelectronics, MEMS is poised to affect many parts of our life.
14 The fabrication of silicon devi ces is the most important part of MEMS and the process is well developed. A brief description of silicon fabrica tion relevant to this research is as follows. A silicon wafer is washed with tetrachloroethyle ne, acetone, methanol and deionized water before use. A photoresist is placed on the wafer and th en spun at high speed to a thickness of a few microns and then baked. A pattern is exposed us ing a mask aligner with UV light and is then developed. The wafer is baked again, followed by d eep reactive ion etching (DRIE), the left over photoresist after etching is removed by pl asma etch. This etching process produces microchannels, which are used for most of the work described herein. Though silicon devices are not used directly for any experiment in this resear ch, they are primarily used as a master to make plastic devices, which will be explained in later Chapters. One of the most successful applications of MEMS is the inertial sensing technology.1 Accelerometers in airbags are now used in millio ns of cars produced. Another important use of MEMS is micro mirrors, which can be used in a ho st of applications such as projection devices, scanners, laser printers and portable communication devices. More traditional MEMS-based devices are pressure sensors, wh ich can be used in tire pressu re sensing or in monitoring the pressure in the medical field. Radio frequency ap plications using MEMS are also proving to be popular nowadays. According to Nexus market anal ysis the market will grow at 16% per year from $12 billion in 2004 to $25 billion in 2009 for di fferent MEMS products, with new fields of application coming up every year. Another major application of ME MS is in the biomedical field. This has lead to emergence of a new field called biological microelectro mechanical systems (BioMEMS).The inherent characteristics of BioMEMS lead to design and fabrication of miniature, smart and low cost biomedical devices which have potential to revol utionize the biomedical research and clinical
15 practice. The success of BioMEMS in industry is dependent on a few issues. First unlike MEMS, BioMEMS devices are mostly non-silicon based. Wh ile the silicon-based fabrication process is well understood and theoretical models are avai lable for designing them, non silicon materials like PDMS, plastics and glasses are not. Hence device design and fabricat ion takes longer time. Second BioMEMS devices are cost-sensitive when co mpared to silicon devices, as for most of applications it is desirable to use disposable units. Other issues include biocompatibility, packaging, closed loop monitoring and controlli ng techniques, and power required for driving them. Important research thrusts related to BioMEMS include cell mani pulation, cell separation, drug delivery on cellular level, retinal implants,2 neural implants in ce ntral nervous systems, tissue engineering, and artificial organ creation. Cell separation am ong them is quite successful using BioMEMS. There are several methods to achieve th is, the most popular of which is dielectrophoresis.3-5 Depending on the frequency of an altern ating current and the size of the cell, the living and dead cells can be directed towards either the positive electrode or negative side of electrode. Other techniques include flow fractionations6 based on electrophoresis, electroosmosis, pressure driven, optical switching7 and using boundary layer manipulation.8 A discussion on the topic will be pr ovided in detail in Chapter 5. 1.2 Microfluidics Microfluidics involves similar techniques of microfabricat ion, which is used to make microchannels and microfeatures in silicon, glass, or other mate rials. However the emphasis is on the fluid movement inside th e channel as the name suggests. The research on microfluidics was first initiated about 30 year s ago at Stanford University,9 where miniaturization of a gas chromatograph was successfully implemented in a silicon wafer. Since then, more complex devices have been developed for many applications, for instance, chemical analysis to the benefit
16 of biochemists. Since movement of fluid is the u nderlying factor in microf luidics, great deal of research work has gone into modeling and understa nding the various physical attributes of fluid behavior in the microscale. Movement of fluid will invariably require some controls applied to it. Valves and pumps are most common ways to manipulate liquids insi de the channel. The valves, due to planar nature of microchannel, tend to have 2-D struct ures. The small size of valve gives a faster response and has lower power requirement, but it te nds to get clogged easil y, and is very difficult to remove the clogs from the microchannels. Stil l there are many of designs available for valve actuations in literature including piezoelectric,10 thermopneumatic,11 electrostatic,12, 13 electromagnetic14, 15 and bimetallic.16 Similarly there are different t ypes of propulsion system avai lable for moving liquid inside the microchannel like thermocapillary,17 electroosmosis,18 electrohydrodynamics, magnetohydrodynamics,19 pressure gradients etc. Micro pum ps are also very common in these systems. Mechanical pumps have the same problems as valves and actuators, namely moving parts, clogging in addition to the problems of low flow rates and pre ssure. Pumps without any moving parts also have been de signed, like diffuser/nozzle pumps,20-22 electrohydrodynamic pumps23 and electroosmotic pump.24 Since this thesis will focu s on electrophoresis (related to movement of charged particles under the influence of electric field), pumping using electroosmosis is discussed in detail below. When a liquid is placed inside a microfluidic channel, the inner surface of the channel acquires charge. This is due to ad sorption of ions from buffer onto the channel surface. In case of fused silica, the surface silanol (Si-OH) groups are ionized to ne gatively charged silanoate (Si-O-) group. These negatively charged groups then at tract the positively ch arged cations from the
17 buffer, which form an inner layer and are tightly held in position due to strong ionic interaction. But these cataions are not enough to achieve neut rality for the silanoate groups. Hence another outer layer of cation is formed, which is mobile in nature. When an electric field is applied, this mobile layer is attracted towards negatively ch arged chathode. Since th is outer layer is not strongly attracted by silanoate group, it moves and drags the bul k fluid along with it. Figure 1 shows the schematic of such a double layer, where the wall is negatively charged. The first layer which is closest to the wall is immobile and is called the Stern layer. The next layer is called Gouy-Chapman layer, where the ions are free to diffuse into the bulk fluid. The plane separating these two layers is called shear pl ane. Figure 1b shows the sketch of the potential associated with this double layer. The potential at the shear plane is calle d the zeta potential.25 Electroosmotic flow (EOF) results when an elec tric field is applied through a liquid-filled microchannel having electrical double layer (EDL). The flow results from the ion drag on the fluid in the EDL. The observed velocity of a particle in this system can be given by the following relationship EP EOF ObservedE v (1) whereObservedv is the velocity in cm/s, E is the electric field st rength in V/cm, and EOF and EP are the electroosmotic and electrophoretic mobility, respectively, in units of cm2/s V. The electroosmotic mobility is a measure of the speed of neutral material in the channel, and can be found by the relationship eEOF (2)
18 where is the thickness of the diffuse double layer in cm, e is the charge per unit surface area (coulomb/cm2), and is the viscosity in g/cm.s. It is desirable to re duce or eliminate EOF in some instances, and this can be accomplished by increasing the viscosity of the liquid in the channel. The electrophoretic mobility is a measur e of the speed of charged particles in the channel (which are more attracted to the anode or cathode than a neutral particle), and is related to a particles charge in coulombs, q and radius in cm, r by the equation r qEP 6 (3) EOF is important when consider ing flow in microscale as it effects the flow condition. Some of the electrophoretic techniqu e uses EOF to its advantage, while it is detrimental for some other techniques like isoelectric focusing (see Chap ter 3). EOF is also used for development of EO pump in microchannels. Overall, microfluidic systems are relatively new when compared to MEMS in terms of research work; however it has been successfully a pplied to quite a few commercial applications such as inkjet printing. The ot her application of mi crofluidics that has attracted immense importance is chemical analysis and separation te chnology The chemical analysis in microfluidic format can be traced back to the 1980s, much before the development of field of microfluidics itself. However the rapid deve lopment of microfluidics has immensely improved the chemical analysis system. As the microfluidics technology has matured in last decade, many chemical processes like separation, detection and purifica tion could be miniaturized in microfluidic format. In an analytical operation, there is sampling, prel iminary treatment, se paration, purification, measurement and interpretation. It is a dream of ch emist to have all of them integrated in one platform. The microfluidics brings about several advantages including a) sm aller size, b) faster
19 operation, c) lower reagent consumption, d) higher selectivity, e) lower detection limit, f) better precision and g) lower cost of fabr icating a device. This has lead to a related concept called lab on a chip (LOC) where one can have total chemi cal analysis system (TAS) in miniaturized format. Manz et al.26 first introduced the term -TAS based on this concept in 1990. One of the most important components of -TAS is separation of macromolecules in liquid medium. Electrophoresis is one such technique for separati on of electrically charge d macro molecules like proteins, peptides, deoxyribonucleic acids (DNA) or particles under th e influence of electric field in a conductive medium. The general form of electrophoresis is very well developed and has been in practice for the last 100 years, and is us ually done in the macro scale. Below is a brief history of electrophoresi s techniques available. 1.3 Electrophoresis Separation of different electrolyte phases under an electric field was first observed in the early nineteenth century and the first quantitat ive work was done by Friedrich Kohlrausch in 1897.27 Thereafter significant work was done on the de termination of ionic conductivities and the mobility of different electrophoretic systems. It was not until 1920 s that Tiselius first adopted this process for separa tion of proteins and colloids. The use of polyacrylamide gel leads to dramatic improvements in separation techniqu e because of its non-ioni c nature, low protein adsorption, minimal electroosmosis and anti-conv ective nature of the gel. Svensson further refined the technique by using amph olytes for stabilizing the pH gradient under the electric field. This process is known as isoelectric focusing (IEF) and is discussed next. 1.3.1 Isoelectric Focusing IEF is carried out in a pH gradient, which can be created by carrier ampholytes or an immobilized pH gradient. Under an electric field, a protein migr ates along the pH gradient (pH
20 increases from anode to cathode) an d is eventually focused at a location where the pH value is equal to its isoelectric point (pI), which is the pH at which the net charge of the protein is zero.28, 29 At the pI location, a dynamic equilibrium between IEF focusing and diffusion keeps the protein remaining in place.30 Since each protein has its unique pI, proteins can be separated along a pH gradient. Figure 1-2 shows the schematic of IEF. In his first seminal paper, Svensson pointed out that formation of uniform pH gradient formation, there needs a large number of lowmolecular-weight amopholytes with low buffering capacity and different isoelectric points. Thes e ampholytes were first synthesized through a reaction of acrylic acid with oligoamines by Vesterberg (1969) and was made commercially under the name Ampholine. This process wa s a huge success because of its very high resolution and was quickly adapted by labs around the world. IEF is usually carried out in polyacrylamid e gel (3% to 5% T, %T defines the total amount of polymer concentration in solution). This gel is a mesh-like struct ure and has very less electroosmotic flow, which is advantageous for IEF process. However there is one disadvantage of using polyacrylamide (PAA) gel, when it is us ed to separate high molecular weight proteins. When high molecular weight protein is bigger in size than the pore size of the mesh-like PAA gel, it will not be able to travel towards its pI point. Several researchers31-33 have proposed agarose gel as a suitable alternative to P AA. This is also popular but this has some electroosmotic flow. Linear polymers are also popular among researchers since they are viscous, preventing convection with no sieving effects. Alongside Righetti et al.34-36 invented a technique to covalently bond the pH gradient to the ge l matrix and marketed it under the name Immobilines. With an immobilized pH gradient there will be very less electroosmotic flow
21 and also the resolution can be as high as 0.001 pH unit. All these mediums can be used for IEF process. As mentioned above the sieving effect in P AA gel is used advantageously for protein separation on the basis of size. This topic is discussed next. 1.3.2 SDS-PAGE Separation The separation of proteins can be achieved based on their sizes. S odium dodecyl sulphate (SDS) is used to facilitate this purpose. SDS is highly anionic in nature and helps to denature the proteins. Merceptoethanol is used first to ope n up the di-sulphide bonds of the proteins. The negatively charged SDS then attaches to proteins to form complexes with similar charge-to-size ratio. The resultant electrostatic repulsion tends to make the prot eins rod-like structures. These proteins are then subjected to the electric field. While they move through the mesh-like polyacrylamide gel structure, th ey get separated based on the size of complexes. Usually the migration velocity shows an inverse relati onship to logarithm of molecular weight. OFarrell37 designed a methodology of combining IE F and SDS-PAGE gel electrophoresis and his protocol is still widely followed with minor modifications. This method uses the samples in fully denatured condition and the second dimens ion used different gradient of polyacrylamide gel for improved resolution. 1.4 Electrophoresis in a Chip Electrophoresis in microscale glass devices was first reported in 1993 by Harrison et al.38 Early work was based on glass or silicon. Each microfluidic devi ce was individually fabricated, requiring time consuming fabricating procedures.39 Nowadays materials like plastic and flexible polymer are used to fabricate devices, which can be batch produced, taking less time. The fabrication process for glass is di scussed here in brief. Thin la yers of metal and photoresist are applied to the surface of glass. The photoresist is masked and exposed to UV light to form a
22 pattern, which is then etched to define the channels. The channels are sealed by thermally bonding a glass cover sheet. Flexib le polymers such as polydimethylsiloxane (PDMS) have made the fabrication comparatively easy.40, 41 A pattern can be made in SU-8 on a glass or silicon substrate, and it is used as master, from which PDMS can be cast. Due to flexible nature of PDMS, it is easy to peel off from master, and the master can be used for a number of times. More recently plastic microfluidic devices42, 43 have become popular because a) they can be made in bulk, b) raw materials are comparat ively cheap, c) have very good optical properties for UV/Visible detectors, d) have structural rigidity and e) mo st importantly, the materials are almost inert to most of the macromolecules. Co mpared to silicon or PDMS, these plastic devices are more rigid thereby ensuring ease of handling a nd also cheap to fabricate. Usually the plastic devices are made by embossing or injection molding, though other fa brication techniques such as laser ablation and compression molding have been reported in laboratory settings. This research work uses plastic device that are fabricated using compression molding, wh ereby resin pellets are pressed against a nickel mold. Th e fabrication technique will be di scussed further in detail in Chapter 2. Usually in microscale electrophores is, the device consists of a main channel for separation with side channels for sample introduction. The entry to those ch annels is done by creating wells as can be seen in a typical device shown in Figure 1-3. The main channel (AB) is usually filled with buffer/separation solution; the sample to be separated is injected either by pr essure or EOF through a side cha nnel (CD). An electric field is then applied to main channel (AB) to pe rform electrophoresis. Once the constituents are separated, they can be detected by a detection system. Different t ypes of excitation sources like UV illumination or laser induced fluorescence (LIF ) are available. UV detection can be applied
23 to many compounds, but its sensitivity is low, whereas LIF is applied only to fluorescent molecules but has very high sensitivity. For thos e compounds that are not naturally fluorescent, they may be tagged by fluorescent molecules for LIF. Microscale electrophoresis has proved to be so successful that a few commercial products are available in the market, including instruments for electrophoresis from Agilient and Bio-Rad. Extensive research work is going to integrate mu ltidimensional separation in chip format. This research focuses mainly in this aspect, and is discussed briefly in objective section below. 1.5 Current Challenges The objective of this research is to miniatur ize 2-D slab gel electrophoresis system into microfluidic chip format, study the separation pro cess and develop an optical detection setup for the same. A short review on the work done so fa r in scientific community is discussed with current challenges that need to be solved before it can put into regular use instead of slab gel system. 1.5.1 Literature Review and Challenges: The grand challenge in microfluidics is to integrate the different components like valves, pumps, diffe rent separation mechanism, detection system etc. in one compact system. There have been se veral reports of standa lone separation systems like isoelectric focusing44, 45, free solution electrophoresis46, SDS-PAGE47, but very few attempts have been made to integrate th em. While it is comparatively easy to perform standalone systems, it is extremely difficult to integrate them because of chemical incompatibility, physical problems related to flow or difficulty in pumping the reagen ts from one location to others. The first effort in 2-dimensional protein separa tion was reported by Chen et. al.48 in 2002. They fabricated a complicated 3-dimensional structure of PDMS as can be seen from Figure 1-4. The device design is very complicated and involved six layers of PDMS. The device consists of a 25 mm long horizontal channel for IEF and several parallel channels (60 mm long).
24 The fabrication of this device requires ali gnment, bonding, removal, realignment and rebonding. The feasibility of such device is demonstrated with limited experiments. This system has a problem of proteins traveling in multip le channels in SDS-PAGE separation. More recently in 2004 Li et. al.49 tried to do the 2-D separation in plastic microfluidic device. They had one cross channel and severa l parallel channels embossed in polycarbonate substrate. The cross channel is used for IEF and parallel channels are used for SDS-PAGE. Figure 1-5 shows the device they ha ve used for their experiment. The main problem for this experiment is that they have used liquid medium for both the separations. Parallel channels are filled up with one liqui d medium (SDS-PAGE) and cross channel with another liquid medium (IEF). This gives a serious problem of cross contamination of the liquids at the junction of cross channels with parallel channels. Since there is no solid physical boundary at the junction of cross cha nnel with parallel channe ls, the proteins may diffuse into the parallel channels even before they are focused. These diffused proteins will then be present in all the parallel channels and will give false peaks in the second dimension separation. Also the parallel channe ls are spaced 1 mm apart. This reduces the resolution of the effective resolution of IEF quite appreciably as ev en if the proteins are separated in IEF channel; the separated proteins in between two parallel channels will go to one single channel. So there still remains a major challenge of integrat ing the two separation mechanism (IEF and SDSPAGE) with proper resolution in a single device. The simplicity in device fabrication will ensure lower cost and disposability. Plastic is both rigid and cheap and optic ally transparent for detection when compared with PDMS. Hence plas tic is chosen for my research work. Simple design of one cross channel and multiple parallel channels are chosen for easy sample loading and ease of experiment. However instead of using liquid medium for both the separation
25 mechanism (as in Li et.al.), second dimension sepa rations in parallel channels are performed in gel. All the channels except the cross channe l is polymerized. This will prevent the cross contamination of two separation medium. Presence of gel on the junction of cross channel with parallel channel will prevent eas y diffusion of proteins onto the parallel channels. The parallel channels are also designed to be closer than previous designs as it will make the separation more precise. Both IEF and SDS-PAGE are separately evaluated in plastic microfluidic device and their parameters are optimized before the integra tion of both of them is tried. IEF is optimized with respect to channel length and electric field to find out the be st parameters and SDS-PAGE is optimized with respect to gel strength, buffer c oncentration and electric field. These conditions are then used to integrate the two separation mechanism in one device. The outlines of the subsequent Chapters are give n in the following paragraphs. 1.6 Objectives Chapter 2 discusses a whole channel image detect ion system for IEF in a microfluidic chip. Laser induced fluorescence detection is use d. The photo bleaching eff ect of the fluorophore molecules due to laser illumination is invest igated and an optimum condition with minimum photo bleaching was found for IEF. In addition devi ce fabrication for plasti c microfluidic chips is also discussed. Chapter 3 describes the dynamics of isoelect ric focusing of proteins in microfluidic devices. Theory for isoelectric focusing process is discussed and the experimental results are compared with theoretical results. Once the IEF process is optimized in microf luidic devices, a device consisting of one cross-channel for IEF process and several pa rallel channels for SDS-PAGE separation is designed and fabricated as discussed in Chapter 4. The device is then tested for two separation mechanism (IEF and SDS-PAGE). IEF is perfor med first on the cross channels followed by
26 SDS-PAGE in parallel channels. Critical parameters are discussed; optimiza tion is done to attain the high separation resolution in both the separa tion mechanism. Chapter 5 discusses the conclusi on and future of the work. Figure 1-1. Representation of el ectrical double layer near a surface. A) The distribution of ions near the surface can be divided into two layers, closely packed ions in stern layer, and more freely moving ions in Gouy-Chapman layer. B) Plot of potential near the wall of a negatively charged wall showing zeta pot ential, wall potential and shear plane. Figure 1-2. Illustration of basic process of IEF, ampholytes gets focused with application of electric field, creating uniform pH gradient in which protein migrate to their pI point and remain there in dynamic equilibrium. The focused proteins can be seen as dark bands. Lower pH Higher pH pH Distance Hi g h Volta g e Focused proteins Shear Plane GouyChapman layer Stern Layer Y Y Y=0, wall Potential wall A B
27 Figure 1-3.Typical microfluidic devices used for electrophoresis work. AB is the main channel where the electrophoresis is performed. The sa mple is introduced from side channel CD. The detection shown here is one poi nt detection where an excitation source illuminates a single point and the detector dete cts the change in intensity due to either absorption or emission by samples. Figure 1-4. Different microfluid ic devices for 2-Dimensional separation. A) 3-Dimensional structure in PDMS. B) Separation of BS A and ovalbumin in the PDMS device. [Reprinted with permission from Chen, X.; Wu, H.; Mao, C.; Whitesides, G. M. Anal Chem 2002 74 1772-1778] A B A D C B Detection Excitation
28 CHAPTER 2 INSTRUMENTATION AND DEVICE FABRICATION* 2.1 LIF Setup and Characterization 2.1.1 Introduction Fluorescence is an optical phenomenon, wher e a fluorophore molecule absorbs a high energy photon and re-emits it as a low energy photon. The absorption of energy leads to the electronic transition of an electron to an excited state. The excited electron then drops back to its ground state emitting energy in form of fluores cence. The energy difference between the absorbed and emitted light is expressed in the form of molecular vibration (heat energy). The wavelength emitted is dependent on the absorbance curve and Stokes shift of a particular fluorophore molecule.50 Use of a laser as an exciting source is popular because of the selectivity of the wavelength and high en ergy density availability. Laser-Induced Fluorescence (LIF) detection is a widely used method in capillary electrophoresis (CE) and microfluid ic devices because of its high sensitivity. Most LIF systems use one point detection, in which a laser beam continuously illuminates a fixed point along the separation capillary (or channe l) to detect the signal when fluorescent molecules pass by.51 For IEF applications, however a consid erable length of channel needs to be illuminated and imaged because proteins do not move after focusing a nd any effort to move them for single point detection results in distortion of peak resolution, destruction of pH gradient and longer analysis time. This whole-channel imaging is important be cause it can also give dynamic information of separation process like proteinprotein interaction and protei n front evolution with time. A part of this Chapter has been published: Das, C.; Xia, Z.; Stoyanov, A.; Fan, Z. H. Instrumentation Science and Technology 33, 379-389,2005
29 Sensitivity and detection limit of the LIF system depends on the efficiency of fluorescence emission and selective collection of that em ission from background noise. The fluorescence emission depends on laser intensity, uniformity of the line beam, and photo bleaching of fluorophore molecules. The imp act of photo bleaching has been studied using single-point detector and the study suggested that photo ble aching process takes place in accordance with a first-order rate Equation in its simplest form.51 This Chapter is to investigate photobleaching in microchannels using the whole channel imaging sc heme. Also studied are the effects of laser intensity, uniformity of the line beam, and phot o bleaching on the detection sensitivity and characterization of the imaging sy stem for protein separations. 2.1.2 Instrumentation A LIF imaging system is assembled in house a nd it is schematically illustrated in Figure 21. The light source is an Ar+-ion laser (488 nm, 30 mW, JDS Uniphase, Model no: 2214-30sl, 0.69 mm beam diameter @ 1/e2), which is directed by mirrors to a beam expander (HB-20X, Newport). For isoelectric focusing, an average po wer of 3 amps (~ 6 mW) is used whereas for second dimension separation, power up to 20 mW is used. The laser has to be run for at least 5 minutes prior to any experiments to stabili ze. The beam expander (Newport: HB 20 XAR.14) converts the laser into a column of light, whic h is focused into a line beam (15 mm) using a cylindrical lens. A capillary or a microfluid ic device is mounted vi a a holder onto an XYZ translation stage, which is used for aligning the device to the position of the beam. It is tricky to adjust the beam expander with the laser line. The laser is adjusted with the help of mirrors to shine a spot on the microfluidic device. The device is kept on fixture so the final laser position is fixed. The beam expander is adjusted in such a way that laser beam glances the external surface of the beam expander. Since the beam expander is circular cylinder, when the beam glances the outside surface it should be parallel to the cylind er surface giving the indica tion that the angle of
30 beam expander is right. The expander then move d and then placed exactly in the right position so that the laser beam comes out as expanded beam with Gaussian intensity distribution. The slight adjustment needed can be done by changing the mi rror angle and the adjusting the knobs in the expander itself. A scientific-grade, cooled char ge-coupled device (CCD) cam era (14 bit digital, 2184 x 1472 pixels, Apogee) is mounted vertical ly above the microfluidic device. The fluorescence emission is collected by the cam era after passing through a band pass filter (HQ535/50 nm, Chroma Technology, Rockingha m, VT). The CCD array size is 6.8 m x 6.8 m. The dynamic range of the camera is 77 dB. For IEF, a high voltage power supply (0-3 kV, Glassman High Voltage Inc., High Bridge, NJ) is used. The power supply is controlled by a computer using softwa re written in Labview (National Instrument, Austin, Texas). A 10-K resistor is connected in series between the channel and ground electrode so that the current in the device can be monitored by measuring the voltage drop across the resistor. The set-up is first studied for photo-bleaching effects, detection limit and optical characterization. 2.1.3 Photo Bleaching Effects Continuous excitation of fluor ophore molecules causes a reduction in their ability to absorb and re-emit light, resulting in permanent or temporary bleaching of them. As a result, the effect of photo bleaching must be taken into account when a so lution is subjected to a prolonged exposure of laser illumination because it will aff ect the sensitivity of the imaging system. The theoretical formulation for photo bleaching can be expressed as a first-order rate process. Therefore, the concentr ation of unbleached dye molecules, c(t) remaining after continuous illumination from time t0 to time t, can be expressed as 51, 52 0 0exp ) ( t t Q c t cb 
31 where c0 is the concentration of the dye at time t0, Qb is the quantum bleaching efficiency, is the absorption cross section, and is the photon flux of the excitation source. Qb is defined as the number of dye molecules that are bleached pe r absorbed photon, and can vary with the type of dye solvent, the dye concentration, th e excitation wavelengt h, and the photon flux.51 The photon flux is given by I h where I and are the intensity and frequency of the monochromatic excitation source, respectively, and h is Plancks constant. Since Qb and are a function of dye concentratio n, excitation wavelength and photon flux. In reality however absorption of photons from th e incident laser flux pr omotes some flurophores to the first excitation state, which may al so absorb and go to second excited state53, 54. These excitation states will have diffe rent absorption cross section a nd total absorption coefficient ( ) will depend on population density in each excitation state. The absorption coefficient in that case is: 1 1 0 0  where 1 0 and are the population densities of fluor ophores in first and second excited states. 1 0 and are the absorption cross section for firs t and second excited states. The temporal evolution of these population densities: tdt I0 1 0exp  It is difficult to get any estimate of populat ion density from the above Equation without knowing the value of absorption cross section. Th is variation in the absorptive properties of fluorophores are however not signif icant when power is very low. A simple experiment (Figure 2-2) is done to show the there is a linear re lationship between laser power and fluorescence intensity, so that it can be assumed that the fluo rphores are not saturated and a single excitation
32 state will be sufficient. The laser power used for most of the experiments in subsequent Chapters is around 6 mW. The actual energy de livered is still less as the la ser beam is expanded 20 times. Therefore, the above condition can be simplified by assuming51 that the coefficient Qb can be mathematically described by 0,,bQfcI For our experiment and I are constant, this equation reduces to0c f Qb At any given time, t the fluorescence, F is proportional to intensity, I and the concentration of the remaining, unbleached dye molecules, c(t) Therefore, F can be described as 000exp FIcfctt  where and are constants. Using Equation  as the theoretical model, the fluorescence intensity is studied as a function of time. The temporal intensity profil es observed in experiment s generally agree with mathematical values calculated using the model, as shown in Figure 2-3. The data indicates that photo bleaching is very signifi cant for higher concen trations of fluores cein, reducing the intensity by 12% in 100 s and by 48% in 500 s for 1 M fluorescein solution. The results also suggest that the effect of photo bleaching drastically decreases with concentration. The photo bleaching is negligible for a fluorescein solution at a concentration of 0.01 M. Since the profile of the laser light is gaussi an, the beam shape after the beam expander should have a similar envelope, i.e ., stronger intensity at the center than on the edge as in Figure 2-4. Therefore the effect of sp atial variation must also be taken into account. Assuming the power is conserved during expansion, the intensity profile can be described by 2 2 22 / ) (y xe I 
33 where x and y are orthogonal axes and is the standard deviation of gaussian profile of the beam. The cylindrical lens compresses one axis to make a line beam, so that the intensity in xaxis can be found by integrating intensity in y-direction. 2 22 2 2 / 2 / ) ( 02 2 2 2 2x r erf e dy e Ix y x y  where r is radius of the expanded beam. Substituting I into Equation  gives 0 0 2 2 2 / 0 2 2 2 /2 2 exp 2 22 2 2 2c f t t x r erf e c x r erf e Fx x  The Equation  has both spatial and temporal dependency. As a result, both spatial and temporal effects of photo bl eaching can be investigated. Figure 2-5.A shows spatial intensity profiles of fluorescence, imaged at different times for 1 M fluorescein solution in a 200 m diameter capillary. The intensity values are normalized against the maximum intensity value after one minute exposure. The results indicate that fluorescence decreases both spatially and tem porally. For longer exposure times, the dye molecules at the center of gaussian irradiati on undergo significant photo bleaching, apparently due to stronger light intensity. As a result, a di p forms at the center of the intensity profile while two side lobes have relatively higher intensity. A similar result is obtained using numerical simulation according to Equati on 5, as shown in Figure 2-5.B. The agreement between the numerical values with experiment al data suggests that the simple theoretical model used, can be employed to predict the degree of photo bleaching and help to find optimum experimental conditions. It should be pointed out that photo bleach ing is significantly reduced for lower concentrations of fluorescein as s uggested in Figure 2-2. Equations  and  also predict that
34 the effect of photo bleaching is reduced for lowe r concentrations of fluorophores and lower laser power. It was found that the effect of photo bleach ing is negligible when the laser power is less than 3 mW and the exposure time is less th an 60 seconds for a fluorescein solution at a concentration of 1 M or less. These results help in determining detection sensitivity of the imaging system and applying the apparatus for protein separations in capillaries or micro fabricated devices. 220.127.116.11 Detection limit After studying photo bleaching and finding out the conditions in which there is negligible photo bleaching, the sensitivity and detection limit of the imaging system was investigated. The limit of detection is found out by estimating the S/N ra tio. This is defined as the ratio of the mean value of the signal ( E ) to rms noise (Es). Mean value of signal is defined as arithmetic mean of n observations (E). n E En i i 1  The rms noise is the standard deviation of the signal de rived from n measurements. 2 1 11 n E E sn i i E  The S/N ratio of 2 is taken as the limit of de tection. Proper care is taken while analyzing the data as the signal is gaussian in shape and the noise is over this gaussian signal. So one particular spatial point (usually the maximum intensity) is chos en along the gaussian beam and data is obtained for n repeated experiments. The detection limit of the imaging system is determined to be 1 nM (nanomolar) of fluor escein (S/N ratio is 2.45) as suggested by the
35 calibration between fluorescence intensity and fluo rescein concentration in Figure 2-6 A. The calibration curve is linear over five orders of magn itude at an incident la ser power of 3 mW with an exposure time of 50 seconds. The data collected is averaged over entire width of the channel to eliminate possible su rface irregularities. Using fluorescein for calibration helps in understanding the limit of detection of the system. However separation processes like IEF leads to focusing of proteins and as the proteins get concentrated, very low initial concentrati on can be used. Figure 26 B shows the calibration curve using GFP in IEF. The minimum detection limit is 0.03 ng/ l and is much lower than the case when fluorescein was used. For higher concentrations (more than 1 M) of fluorescein, the nonlinear effect of photo bleaching is prominent even at this power and exposure time. This can be avoided by having less exposure time. Very low intensity fluorescence measurement has a different problem. The gaussian nature of the lase r line induces similar gaussian characteristics in fluoresecence emission. Hence the intensity is ma ximum at center and gradually reduces on two sides. This requires mathematical correc tion of the profile an d is discussed next. 18.104.22.168 Optical correction The sensitivity of the detecti on system is inversely proporti onal to the square of the distance between the object and the camera. The s horter the distance is, the better the sensitivity is. The shorter distance also increases the spat ial resolution; however, it reduces the field of view. A trade-off between the maximum field of view and the best sensitivity is chosen depending on the application. Due to spatial di fference in the laser intensity, there is a dependence of the fluorescence intensity on the locati on of a protein. If a pr otein is focused near the center of a gaussian beam, the protein exhi bits maximum fluorescence. If a protein is
36 focused near the edge of the gaussian beam, the pr otein exhibits less fluor escence. As a result, a visual distortion exists in terms of rela tive intensities of th e focused proteins. This distortion can be corrected by refractive beam shaper or mathematically. A refractive beam shaper though commercially available is ve ry expensive and still tend to have some distortion at the edges. Mathematically it can be corrected by taking in to account the original intensity profile of the laser beam. At any given time, fluorescence intensity can be approximated by F IC where is a proportionality constant, I is the laser intensity, and C is the concentration of protein. A ssuming the initial concentration of the sample is uniform along the channel, the initial light intensity can be described by00() FIIxC where Io is the maximum intensity at the center and I(x) is a normalization function along the channel. In other words, a normalized intensity profile between 0 to 1 will give a profile of I(x) i.e., ()start normalizedFIx After IEF and proteins are focused, the fluorescence can be described as ) ( ) (0 0x C C x I I F where C(x) is also a normalization function along th e channel, but for the variation in concentration. The normalized intensity pr ofile between 0 to 1 will give a profile of()() I xCx, i.e.()()final normalizedFIxCx Therefore C(x) in the image can be correctly represented by ()final normilzed start normalizedF Cx F Figure 2-7 shows an electropherogram of different proteins before and after mathematical correction. At the beginning of experiment, the protein is uniformly distributed along the channel. Initial intensity profile shows a gaussian distribution, which is consistent with that of laser beam profile as discussed ear lier. After IEF, the protein is focused at its isoelectric point, represented by a peak in the electropherogram. The raw data for Figure 2-7 A shows the effect
37 of gaussian beam profile. The prof ile is flat after removing the ga ussian effect. Similar correction shows a clear shoulder peak in Figure 2-7 B. This shoulder peak is expected due to microheterogeneity of GFP, which refers to sl ight structure differences among isoforms of a protein synthesized biologically. GFP is often observed having a large portion of the primary component with a small portion of other is oforms as discusse d in the literature.55 Slight increase of intensity in the corrected prof ile at both ends is probably due to amplification of noise near the edge of the gaussian beam. 2.2 Device Fabrication 2.2.1 Introduction A growing amount of research is being devoted to the study of microfluidic devices for a number of applications, including DNA analysis, protein separations, and cell manipulation. While many researchers use PDMS glass, or silicon to fabricat e their microfluidic devices, others choose thermoplastics due to the followi ng advantages. First, plastic devices are potentially inexpensive, resulting from lowcost raw materials and vast experience in manufacturing high-volume plastic pa rts. For example, the manufact uring cost of a compact disc (CD), a two-layer structure made from acrylic or polycarbonate and containing micro scale features, is less than 40. 42 As a result, it will probably be affordable for plastic devices to be disposed off after a single use. Being disposable could have tremendous impact in applications where cross-contamination of sequential samples is of concern, including point-of-care clinical diagnostics, drug screening, and fo rensic analysis. In addition, plastic devices are compatible with biological and chemical reagents, evident fr om the extensive use of plastic labware such as micro centrifuge tubes and micro plates. The plastic materials investigated for micr o-fabrication and micr o fluidics include polymethyl-methacrylate (PMMA),56-62 polycarbonate (PC),62-65 cyclic olefin copolymers
38 (COC),66-71 polyester, fluorinated ethylen e propylene, and poly(ethylene terephthalate). COC is used since it offers a number of advantages, including increased so lvent resistance, higher optical clarity, and reduced absorption of moisture.66 The major techniques for fabri cating plastic microfluidic de vices include hot embossing, injection molding, and laser ablati on. Casting is another approach that is almost exclusively used for PDMS.72 The general methods for fabricating plasti c devices have appeared in literature and have been reviewed. 42, 73, 74 In the following sections th e device fabrication and design parameters adopted for producing microfluidic devices for this work is described. 2.2.2 Design and Fabrication The device was designed using AutoCAD and the layout is s hown in Figure 2-8. The 6channel device (Figure 2-8A) is fabricated for isoelectric focusing. The channel length ranges from 1 cm to 6.5 cm, with a side channel a pproximately 5 mm in lengt h and 5 mm away from one end of channel. The side channel is desi gned for sample injection. The channels are approximately 120 m wide and 30 m deep. The second device (Figure 2-8 B) consists of one vertical channel (AB) for the fi rst dimensional separation (IEF) and 29 horizontal channels (CD) for the second dimensional separation (PAGE). The feature size of etched channels is approximately 120 m wide while the CD cha nnels are separated by 240 m. Plastic microfluidic devices were produced us ing the procedure sim ilar to what has been reported in the literature.42, 66 A glass master with the de sired pattern was defined by photolithography. Chemical isot ropic etching was exploited to create glass masters with channels approximately 30 m deep and 120 m wide. Electroplating ag ainst the master created a metal electroform42 and it was carried out by Optical Elec troForming (Clearwater, FL). The
39 electroplating material is nickel alloy and the thickness of th e electroform is about 1 mm (Figure 2-9). The topology of the electroform is exactly the opposite of the glass master; a channel in the master becomes a ridge in the electroform. This electroform served as a molding die that would produce plastic parts from Topas 8007 resin (Ticona, Florence, KY) using a hydraulic press (Carver, Wabash, WI). The molding temperature is around 257 oF and the compression force is 5000 lbs. About 15 gm of Topas 8007 resin is sandwiched between E-form and a 6x 6 inch glass plate and placed inside the pres s as illustrated in Figure 2-10A. After preheating for 5 minutes at 257 oF, the lower platen is raised to compress the resin at 5000 lbs and kept at that pressure for 5 minutes for pressure moldi ng. Then the molded plastic is cooled at controlled temperature of 70 oC for 5 minutes and is removed from the E-form. Holes were drilled at the end of channels using a computer numerically controlled (CNC) mill (Flashcut 2100; Flashcut, Menlo Park, CA) to func tion as reservoirs. Th e device is completed by sealing the plastic substrat e with a cover film (~70 m thick Topas 8007 film) using thermal lamination.42 Figure 2-9 B shows the schematic of la mination process. The plastic device and film sandwiched between two Mylar films (2 mil thick) are initially preheated for 1 minute at 85 oC, and then is run through a laminator (Catena 35, GBC). The roller temperature for the laminator is set at 110 oC. The devices thus fabricated are checked fo r any channel blockage, which can happen due to variation in lamination temp erature and applied pressure.75 These devices are then surface treated for carrying out electrophoresis, the deta il of which is discusse d in next Chapter. 2.3 Conclusion A laser-induced fluorescence imaging system was assembled for viewing the full length of microchannels. The advantage of the system is that its ability to st udy the protein separation
40 dynamics as a function of time since the sequent ial images can be obtained. This method is useful for studying IEF process, as proteins need not be moved after focu sing, thereby preventing the loss of resolution of protein peaks. The dete ction limit for this system was found to be around 1 nM of fluorescien solution. Photobleaching wa s found to be an issue, when the fluorophore concentration is higher above 1 M of concentration. As a result ca re is taken to carry out all the subsequent experiments within th at range, though the degree of photobleaching effect may differ in different types of fluorophore molecules. Fo r very low intensity fluorescence measurement, gaussian beam profile correction needs to be done. In addition, I di scussed the fabrication process of microfluidic devices. The comp ression molding has good reproducibility, and hundreds of devices can be fabr icated from one master. Figure 2-1. LIF imaging system fo r protein separations. A) The lase r beam is transformed into a line beam using a beam expander and cylindrical lens. A line beam is needed to illuminate and image an entire channel during isoelectric focusing of proteins. B) Photograph of the set-up in an optic table. A B Mirror Mirror Beam expander Microchannel device Band pass Filter (535+/25nm) Lens ND Filter
41 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 050010001500Time (s)Intensity (Normalized) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 050010001500Time (s)Intensity (Normalized) 0.01 M 0.1 M 1 M 10 M Figure 2-2. The fluorescence inte nsity values measured in micr ochannels with different laser power. The exposure of CCD camera is 100 ms. Figure 2-3. Temporal profile of photo bleaching effects. The data poi nts are experimental results, whereas the solid lines are numerical calculation using Equation . The concentration of fluorescein solution is indicated. The fl uorescence intensity values are normalized with their respective maximu m values, so that the gradients can be compared in the same plot. The power of laser line is 3 mW. R2 = 0.9912 400 500 600 700 800 900 1000 4567891011 Power (mW)Fluoresence (a.u)
42 Figure 2-4. The cylindrical lens compresses the beam in one axis Figure 2-5. Spatial profile of phot o bleaching effects. A) Experimental results for sample of 1 M concentration of fluorescein solution. The length of laser line is 20 mm. The intensity values are normalized with the maximum value of intensity of the first minute profile. B) Numerical simulation fo r photo bleaching eff ects using the same parameters as in A. Cylindrical lens -10 -5 0 5 10 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Spatial location (mm)Intensity (a.u) 1 minute 5 minutes 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes -10 -5 0 5 10 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Spatial direction (mm)Intensity (a.u) 1 minute 5 minutes 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes A B
43 Figure 2-6. Calibration curve for the detecti on system. A) Calibration curve of fluorescein solution in a glass capillary using the imaging syst em in Figure 2-1. Both axes are in log scale. B) Similar calibration of the system using GFP in Isoelectric focusing. The channel length used is 4 cm at 100 V/cm. R2 = 0.9905 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 -3.5-3-2.5-2-1.5-1-0.500.511.5 Log (Conc. mM)Log (Intensity) R2 = 0.9983 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 00.20.40.60.822.214.171.124 Concentration (nano gm/ micro liter)Intensity (a.u) A B
44 Figure 2-7. Optical corr ection for removing Gaussian noise background. A) Isoelectric focusing of four proteins green fl uorescent protein, R-phycoery thrin, B-phycoerythrin and phycocyanin in pH gradient of 4-6. B) IEF for GFP in pH gradient of 3-10 formed in channel of length 2 cm. The electropherogr am is mathematically corrected in both cases by taking into account of spatial laser intensity variation. The raw data in left image (blue) after correction (red) shows a fl at intensity profile. Similarly for right one, a small hump for isoform of GF P is visible after correction. Figure 2-8. The layout of microflu idic devices for protein separa tion. The size of the device in general is 1 x 3. A) Device used for isoe lectric focusing of prot eins. B) Device used for 2-dimensional electrophoresis. Cross cha nnel AB is to be used for IEF whereas parallel channels CD are used for separation based on sizes. 4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6 -200 0 200 400 600 800 1000 1200 Gaussian background correction pHIntensity (a.u) 0 5 10 15 20 25 0 1000 2000 3000 4000 5000 6000 Distance (mm)Intensity (a.u) Initial gaussian profile Raw data after IEF Corrected profile A B Raw data after IEF Initial gaussian profile Corrected profile A B CDA B
45 Figure 2-9. E-form used for this research. Figure 2-10. General methods for plastic device fabrication. A) The schematic of molding operation for plastic device usi ng E-Form. B) The plastic device and film is initially heated and then run through laminator at 230 oC Hea t Mylar sheet Plastic device Roller A B
46 CHAPTER 3 THEORITICAL AND EXPERIME NTAL RESULTS OF IEF 3.1 Introduction With the advent of microfluidics, effort s were made to miniaturize electrophoretic processes to the microfluidic chip format. IEF has been adapted to the microfluidic format due to potential advantages of miniaturization.59, 64, 76-85 Possible advantages include less sample amount, faster analysis, higher se paration efficiency, potentially lo wer cost, and the ability to integrate with other components including a detector. In this Chapter, I will discuss IEF in the 6 channel device described in Chapter 2, a nd a comparative study between theory and experimental results. The details of experimentation will be first described, followed by theoretical and experimental results. 3.2 Materials and Methods Carrier Ampholytes (CA) (p H 3-10, 4-6) were purchase d from Bio-Ra d Laboratories (Hercules, CA) while ethanolamine, 2-hydroxyet hyl cellulose (HEC, MW 90,000, 150 cps @ 5% w/w in water), hydroxypropyl ce llulose (HPC, MW 80,000, 150700 cps @ 10% w/w in water), Myoglobin and Bovine Serum Al bumin (BSA) were from Aldr ich-Sigma (St. Louis, MO). Myoglobin and BSA were labeled with Alex a488 Protein Labeling Kit (A-10235) from Molecular Probes (Eugene, OR). Green Fluorescent Protei n (GFP, 1 mg/mL stock) was obtained from BD Biosciences Clontech (Palo Alto, CA ) and R-phycoerythrin (RPE, 10 mg/mL stock) was obtained from Cyan otech (Kailua-kona, HI). Acrylamide:bis-acrylamide (electrophoretic grade, 5%C), TEMED (N, N, N', N'-tetramethyl ethylenediamine), glycerin, acetic acid, and microscope slides were purchased from Fisher Scientific (Atlanta, GA). All solutions were prepared using water purified from Barn stead Nanopure Water System (Model: D11911, A part of this Chapter has been published: Das, C.; Fan, Z. H. Electrophoresis, 27, 3619-3626, 2006
47 Dubuque, Iowa). Solutions of 10 mM acetic ac id and 10 mM ethanolamine served as the anolytes and the catholytes, respectively. All experiments were performed either in glass or plastic microfluidic devices. The channels in the device were fi rst cleaned with 1% (0.18 M) potassium hydroxide (KOH). The solution is allowed to stay for 5 minutes inside th e channel and then flushe d out with water. The channel is then filled with IEF separation medi um, which consists of 5% ampholytes (pH 3-10) stock solution (at 40% su pplied by the vendor), 10% glycerol st ock solution (80% v/v in water), and 85% HPC/HEC stock solution (1.83% HPC and 0.73% HEC in water, except where specified otherwise). A mock IEF is performed w ith all the constituents (including proteins) in the presence of an electric field, followed by flushi ng out with water. This step helps to block all the active sites in the surface of channel, ther eby preventing protein adsorptions in future experiments and helps conditioning the wall surfaces unde r an electric field. It has been reported that most plastics, including poly(cyclic olefin), possess much smaller EOF, typically a factor of five or so smaller than silica/glass.42 Electroosmotic flow (EOF ) are further suppressed by dynamic coating of HPC/HEC contained in the IEF separation medium, as reported in the literature.86-88 Nevertheless, a minute amount of EO fl ow was observed, especially when the electric field strength was high. A protein sample can be introduced into a mi crofluidic device by using two approaches. One is to fill proteins in an entire channel 59, 76, 78, 79 whereas the other is to inject a plug of protein samples into the IEF channel 89 as demonstrated in many microfluidic CE devices.90, 91 It is chosen in this work to fill an entire channel with analytes to ensure uniformity in pH gradient. 3.3 Theory of IEF Electrophoresis in a microchannel is a comp licated three dimensional problem. The analytical electrophoretic model used here is assumed to be one-dimensional in nature. The
48 motive of this analysis is to find out the physical factors behind the focusing characteristics and relation of time of completion wi th that of conductivity drop in th e system due to IEF process. Several simple assumptions are made for the following semi-analytical solution. The process of IEF in a typical microfluidic devi ce or capillary can be viewed as follows. The channels are initially filled with carrier ampholytes and proteins (both are charged macromolecules). As the electric field is applie d, a pH gradient is es tablished. Once the pH gradient is established the prot eins (macromolecules) starts mi grating towards their pI point where their net charge is zero. At a steady state the proteins stay stagna nt at their pI points. Standard transient electrophoret ic process can be defined as2,3,4,92 2 2) ( ) ( ) ( ) ( x t x C D x x u t x C t t x C  where u and D are the velocity and diffusion coefficient of the component, t and x are time and distance variables, and C(x,t) is the concentration of the co mponent, which is a function of location and time. The equation is the Ficks diffusion law93 modified with an additional term for electrophoresis. Both the transient process and final steady state is of importa nce in separation science. The transient solution shows the dynami cs of protein front evolution, how the finer structures of proteins evolve with time and that the proce ss proceeds towards completion asymptotically. The transient equation is difficult to solve, but if certain assu mptions are made, the underlying physics of the problem can be understood. The eq uation is non-dimensi onalised for the purpose with following variables. **** 0 00;;; tu x uC xtuC lluC
49 where l is the length of channel, 0u is the initial velocity and 0C is the initial concentration of the species/proteins. The corres ponding non-dimensional equation is of the following form ** *2* ***21 Cu CC txPex  where 0lu Pe D is defined as the Peclet number. This is defined as the ratio of product of length and fluid velocity to that of di ffusion. When the Peclet number is small, the diffusion is dominant in the system, whereas when it is high, diffusion can be neglected when cons idered with the size. Let us consider the case when diffusion dominates, the equation simplifies to: If1 Pe 2* *20C x  **CAxB  Hence the concentration is a linear function of distance. The channel length is very short when compared to diffusion of species. The species will diffuse into the channel very fast and IEF will fail in this condition. Th e limiting conditions for this case can be taken as the theoretical limit for electrophoresis. 01 lu Pe D 0DD l uE  Where is the electrophoretic mobility for IEF and E is the applied electric field. Using diffusion constant for a model protein94 (BSA in this case) 727.7510/ Dcms, electrophoretic mobility 4112410 sVcm, E= 100 V/cm, the length of channel is calculated to be:
50 5~10 lcm (0.1 m)  The above phenomenon occurs when the channe l length is less than 0.1 m. It can be assumed that IEF can be safely performed fo r longer channel length. However as it will be demonstrated, there is practical limitation to the minimum size beyond which it becomes practically impossible to perform IEF. The other extreme of th e above equation is to have very high peclet number. 1Pe Equation 2 then changes to ** **0 Cu C tx  This equation can be solved analytically with additional information for velocity from IEF. For general electrophore tic movement the velocity is dependent on electric field as: E x u ) (  Svensson 95 made an assumption that the pH gradient, d(pH)/dx, and the mobility difference at pI value, d/d(pH), are constant Then the mobility can be expressed by x dx pH d pH d dx) ( ) (  Hence u can be written as a linear function of x *ukxd  Where k, d are the constants. The final so lution for this case can be written as **.k t kCAekxd  Where is also a constant. This however gives very sharp concentration gradient which decays exponentially with time. The sharp concentr ation gradient is becaus e of very high peclet
51 number. The full equation is solved though w ith some assumptions and can be found in literature96, 97. The final analytical soluti on is given here and the fu ll solution can be seen in Appendix A: ) ( 2 exp 1 exp 2 ) ( 1 ,2 2 0t s t t x x t s t x CpI  This solution gives a moving front of peak with Gaussian concentration profile which moves towards its isoelectric point and ultimately stops there. Equation 12 can be analyzed for different conditions At time t = 0, ) ( t g 0 ) ( t x, the center of mass for the CA/protein is at x=0 2 22 exp 2 1 0 x x C  At time t 0) ( Dt t g pIx t x ) ( So at very long time the CA/protein peaks center of mass will be exactly at their pI point. 0 2 02 exp 2 1 Dt x x Dt x CpI  Equation 12 has the spatial location of center of mass of protein peak. Differentiating with respect to time will give the front velocity The front speed thus can be defined as: 0 0exp ) ( t t t x t vpI  This front speed can be monitored in real time by the imaging system developed for whole channel viewing and the numerical results can be experimentally verified. However the steady state analysis gives us the vari ous experimental conditions requi red for successful completion of IEF.
52 For IEF, a steady state is presumably reached at the completion of focusing, i.e, 0 ) ( t t x C Then the concentration is a function of location, x only. The velocity of a component can be calculated from its electrophoretic mobility at its location, (x) and the electrical field, E i.e., E x u ) ( (assuming zero electroosmotic flow). As a result, Equation 1 is simplified to Equation 16, dx x dC D E x x C ) ( ) ( ) (  where x is now defined as being x = 0 at the concentration maximum of a peak. To solve this Equation , Svensson 95 made an assumption that the pH gradient, d(pH)/dx and the mobility difference at pI value, d/d(pH) are constant. Then the mobility can be expressed as: x dx pH d pH d dx) ( ) (  After substituting (x) into Equation 16, the equation can be in tegrated on both sides as follows, x C Cxx C x dC xdx D E dx pH d pH d d00) ( ) ( ) ( ) (  where C0 is the concentration of the component at th e peak maximum. Solving this equation will give Equation 19 D E dx pH d pH d d x C Cx) ( ) ( 2 exp2 0  To fit with the traditional gaussian distributio n of a peak, Equation 19 can be rewritten as Equation 6,95, 98 2 2 02 expx C Cx 
53 where is defined by ) ( / ) ( / pH d d pH d dx E D and its physical meaning is the standard deviation of the gaussian peak. The IEF resolution can be described by the mini mum pI difference required to separate two proteins. In order to disti nguish two peaks, the distance between them must be at least 3 Figure 3-1b shows graphically the requirement for 3 between the peaks for proper resolution. The superimposition of the two gaussian profil e if separated by 2, gives only a single peak, whereas 3 separation resolves them. Therefore, th e minimum pH difference required is ) ( / / ) ( 3 ) ( 3 ) (minpH d d dx pH d E D dx pH d pI  When a constant voltage, V is applied and a uniform pH gradient is used, then E = V/L and L pH dx pH d ) ( where L is the separation distance. Equatio n 7 is simplified to Equation 8, ) ( / 3 ) (minpH d d pH V D pI  in which there is no term related to the distance. Equation 8 clearly indicates that the minimum pI difference required for two proteins to be se parated by IEF is independent of the separation distance. As a result, a short focusing length is advantageous, es pecially for a microfabricated device, since it should provide more rapid analysis without sacrificing the reso lving power. It should be noted, however, that a few assumptions are made during derivations, and they (e.g., negligible EO flow) may not be true in some e xperimental conditions. In addition, the separation voltage must remain the same while reducing the separation distance. As a result, the electric
54 field strength and current will in crease; possibly producing Joule heating that is not negligible anymore. From the theoretical results above, the factors that need to be optimized for miniaturization are a) electric field strength, b) length of channe l, c) time of separation for proper optimization. The theory predicts that length of channel should not be a factor in separation, as the minimum pI difference required for proper re solution of protein peaks is inde pendent of channel length. So a very long capillary needed for electrophoresis can be shrunk into a small chip like a microfluidic device. Since the time of completion of IEF is inversely proportional to electric field, if proper heat dissipati on from the chip can be designed, high electric field can ensure the IEF can be completed with in minutes. Since the ch annel length is reduced the high field strength can be maintained very easily without need for very high voltage pow er source. There are several other parameters that co me directly from practical sta ndpoint and experimental issues. All these will be discussed in next sections with emphasis on channel le ngth independency, the effects of field strength, focusing ti me and pH gradient compression. 3.4 Experimental Results The theory described above is idealized and has many assumptions, which will lead to mismatch between experimental results and theo retical calculation. Correctly analyzing those anomalies will lead to better understanding of process and thus improving the theoretical analysis. The first set of experiments were carri ed out by focusing one protein at a time and then gradually making the system mo re complicated by adding more proteins. The whole-channel imaging system designed and characterized before is used for detecti ng proteins after their separation by IEF. Since GFP is naturally fluor escent, it has been used for characterization of the system. The dynamics of IEF was inve stigated and the temporal images of GFP separation are shown in Figure 3-2. This result indicates that a front starts focusing from each end after
55 applying a voltage across a separation channel. Focu sing fronts start to be observed at the fourth minute. Both fronts progressively move towards th e middle, and finally combine together at the focusing point after about forty minutes. The cat hode front is rather wide and faint during focusing whereas the anode front is sharper. Protein bands at th e anode front are clearly visible soon after the front appears (aft er ~5 min at 200 V). The conductiv ity of system drops while the pH gradient is being established. As a result, a higher electric field may be applied after initial focusing. For example, the applied voltage was increased up to 500 V for final zone compression and better resolution. The separation medium for this case was polyacrylamide gel. 3.4.1 Effects of Separation Medium The separation medium used is either polyacryl amide gel or linear polymer; both of them have been successfully implemente d in either microfluidic channe ls or glass capillaries. The polyacrylamide significantly reduces electroosmotic flow (EOF), especially when the gel form is used. However polymerization inside a microcha nnel is a significant problem. Introducing the solution inside the channel becomes difficult if the chemical polymerization is started beforehand. UV polymerization can be done in situ but it destroys the fluorescence of proteins and thus can not be detected. Elimination of EOF flow keeps the focused protein bands in place, as well as enhances separation resolution. The cross linked polymer (for polyacrylamide) is a mesh like structure and it has sieving properties for proteins moving across it. This leads to increased drag and hence longer separation time. Mesh like structure also hampers separation of large molecular weight proteins. The linear pol ymer on the other hand is a viscous liquid and does not have mesh like structure. So it is easy to introduce inside the channels and does not restrict the flow of bigger proteins. Next three different proteins are focused in the microchannel. They are GFP, BSA, and Myoglobin. Both BSA and Myoglobin are labeled with Alexa 488 dye, so that they can be
56 visible when illuminated by argon laser. Figure 3-3 shows the separation of these proteins in both linear polymer in Figure 3-3a and pol yacrylamide gel in Figure 3-3 b. The results in Figure 3-3 suggest that the reso lution in the linear polym er is slightly better than in cross linked polymer. The images reco rded at regular interval by CCD camera are analyzed to track the protein front with tim e. Figure 3-4 shows th e protein front speed comparison with time for the case above (Figure 3-3). Time required for separation is less for linear polymer (40 minutes for gel and 20 minutes for linear polymer). It is also observed linear polymer can withstand higher level of electrical field without appreciabl e heat generation when compared with polyacrylamide gel. The reas on behind maybe of conv ective cooling possible with linear polymer but not with cross linked gel. The sample prepared can be used multiple times in case of linear polymer, whereas the pol yacrylamide gel can be used only once. Hence it is chosen for all subsequent work. Since BSA, and Myoglobin labeled with Alexa4 88 gave broad peaks due to heterogeneity in labeling,99 we did not use them for studying the IEF pr ocess. It is necessary to know the exact pI point of the proteins for evaluating the comp letion of IEF. Also the dyed peaks have broad irregular shapes which make it difficult to study the resolution issues in IEF, as it is easier to assume the clean peaks as gaussian in nature a nd then calculate the resolution thereof. GFP and RPE are two naturally fluorescent pr oteins with well defined pI poi nt. They can be excited with same wavelength (488 nm) and have similar emission spectrum. So these two naturally fluorescent proteins are chosen, whose pI points are well characterized. The next sets of experiments were started similarly by uniformly filling a channel with a mixture containing ampholytes, lin ear polymers (HPC/HEC) and prot eins. The ampholytes with a pH gradient of 3-10 were used and the separa tion voltage was 500 V (other voltages were also
57 studied as discussed below). As mentioned previously,100 the protein peaks were observed almost immediately after an electric field was app lied. These peaks traveled from the anode side toward the middle of the channel. The intensity of the protein peaks was initially very low, and then increased as they traveled along the ch annel while more proteins were focused and accumulated in the peaks. Figure 3-5 shows IE F of these proteins at 100 V/cm. The GFP is having one dominant peak flanke d by two small peaks. These small peaks are due to micro heterogeneity of GFP. The pI points for these peaks are 4.88, 5.0 and 5.19. Similarly for RPE three peaks are visible; the pI for th e dominant peak is 4.2. The length of the channel is 5 cm and protein concentration is 5 g/ml. The proteins thus optimally focused inside the microchannel and be ing detected by the CCD camera indicated that this setup can be used for further studying the e ffect of channel length, electric field and other parameters. 3.4.2 Effects of Separation Length According to Equation 8, the minimum pI di fference between two proteins separated in IEF should be independent of the separation di stance if the voltage is maintained and all assumptions are met. Figure 3-6 shows the IEF electropherograms of GFP and RPE in channels II-V with different separation distances (2.1, 3.2, 4.3 and 5.4 cm respectively). The channels of length 1 cm and 6 cm were not used due to their too short or too long a ch annel. All three peaks of GFP in all four separation distances were obser ved, thus it can be claimed that all of them have capability to separate protei ns with pI difference of 0.1 pH units. However, it is clear that the degree of separation between these peaks is le ss when the separation distance shortens. The concept of separation resolution is used to an alyze the results further as discussed below.
58 Separation resolution (R) can be calculated by dividing the separation distance of two adjacent peaks by their average width, i.e., 21 12() 1 () 2 x x R ww  where x1 and x2 are the location of two peaks and w1 and w2 are the width of the peak at the base, measured between the tangent lines of the peak sides.101, 102 Larger values of R mean better separation, and smaller values of R poorer sepa ration. In order to maintain the separation resolution when the separation distance shortens (i.e., x2 x1 becomes smaller), peaks must be sharper (i.e., w1 and w2 also become smaller). This is theoretically possible because a higher electric field results when a shorter channel is used while the separation voltage remains the same; a higher electric field like ly leads to sharper peaks. Figure 3-7 shows the comparison between different resolution values Also shown is the effect of superposition of two peaks of different peak heights. Comparing the case for R= 1, the superimposed image of different peak heights (1:0.75) appears to be ba rely resolved, whereas same peak height looks better visually. As discussed above, separation resolution (R) can be calculated by using Equation 1. The R values between peaks 1 and 2 of GFP (can be seen in Figure 3-8) as a function of the separation length at different separation voltage s are tabulated in Table 3-1. Statistical analysis using ANOVA (analysis of variance) indicates that the resolution values at 300 V and 500 V in Table 1 are the sa me at the 95% confidence level for different separation lengths. The resolution values at 200 V and 400 V are statistically different based on ANOVA, but the trend of the change is not clear, especially for those at 200 V. Relatively large standard deviation in these values is primarily due to the fact that each peak corresponds to 15 to 25 pixels of CCD. One pixel variation will result in 4-6% deviation. The results indicate that the
59 separation resolution is independe nt, within the experimental e rrors, of the separation length, which is in agreement with the th eory predicated. It is should be pointed out that the conclusion is true for the most experiments when the separation voltage changed from 200 V to 500 V, which correspond to the electric field strength of 38 to 263 V/cm. 3.4.3 Effects of Separation Voltage Effects of the separation voltage on IEF by applying different separation voltages across 5.2 cm channel are investigated. Figure 3-8 shows a few representative IEF electropherograms of GFP and RPE when different se paration voltages were used. There is little difference in the electropherogram, though it showed sharper peaks when voltage increased initially. Th e initial increase in the peak sh arpness is also evident from the increase of separation resolution listed in the Table 3-1. This is also the basis of the theoretical prediction that the separation resolution is indepe ndent of the separation distance, as discussed above. On the other hand, higher electric fields could also result in Joule heating, which increases peak dispersion. This may explain the general trend that the separation resolution increase initially, level off, and then decrease later on as indicated in Table 1 and Figure 3-8.The key advantage of using a higher separation voltage is a shorter analysis time, as discussed below. 3.4.4 Focusing Time A short analysis time is one of major motivations to go for miniaturization. Obviously, the IEF focusing time will be affected by both the separation length and voltage. When the separation resolution is minimally affected, a shorter separation length and higher separation voltage, thus higher electrical fiel d strength, will be pr eferred as they will lead to a shorter analysis. However, IEF analysis time is difficult to meas ure. Theoretically in IEF, the proteins will reach its pI point asymptotically and hence wi ll take almost infinite time towards total
60 completion. Experimentally, completion of a c onventional IEF in a slab gel or a tube is facilitated by a dye contained in a sample when ampholytes are used. When the dye reaches to the end of the capillary tube or slab gel, it is regarded as the point to stop. It is likely that the property of the dye is different fr om proteins in the sample. IE F experiments are often helped by empirical experience and the protocols supplie d by manufacturers. For immobilized pH gradients (IPG), protocols often recommend an overnight IEF since the pH gradient will not move away. The whole-channel imaging detection (WCID) sy stem described in Chapter 2 allows us to monitor IEF dynamically. Temporal profiles of protein separations provide the information about the time when the peaks do not move any furthe r, and IEF is considered to be completed. However, EOF was not completely eliminated as mentioned above, especially when a higher voltage was used. Therefore the completion of IEF is determined by the significant slowingdown of peak moving, and it gave a very good id ea after a certain amount of experience. The completion of IEF is confirmed by the expected pI location assuming the pH gradient is uniform and linear in the channel. Figure 3-9 shows the relationship between th e focusing time and the electrical field strength when a variety of separati on voltages were used in differen t separation lengths. The standard deviations repres ented by the error bars were ob tained from three sets of repetitive experiments. A voltage range of 200 V to 1000 V was used for channel V as in Figure 3-8 whereas a smaller range of 200 V to 500 V was used for channels II and IV. The focusing time ranged from 1.5 to 46 minutes depending on th e separation length and electrical field used. The data suggest that the focusing time decreases w ith the electrical field strength and there is a
61 linear relationship between the time and the invers e of the electric field st rength. In addition, the focusing time decreased with the se paration length as expected. 3.4.5 pH Gradient Compression Recently Cui et al. reported that the pH gradie nt in a microfluidic device is compressed to the middle of a channel, rather than in the w hole channel starting from the anode and ending at the cathode.79 Similar compression phenomenon was found to be existed in these experiments. For the electropherograms in Figure 3-2, the compression ratio is 83, 84, 88 and 93% for channels II-V, respectively. The compression ra tio was the actual distance (in pixel) between two major peaks of GFP and RPE divided by the cal culated distance from their pI difference and the pH gradient. The results suggest that cathol ytes and anolytes entered into IEF channel for a short distance (7-17% of the cha nnel) to facilitate the formation of pH gradient. The degree of compression decreased with the leng th of separation channel. The effects of different pH gradients, pH 4-6 and pH 3-10 on IEF focusing and gradient compression was also investigated. As expected, the peaks were better resolved in the pH 4-6 gradient, having larger peak separation than that of the pH 3-10 gradient as shown in Figure 310. Both experiments were run in 5.2 cm chan nel. The separation voltage was 1000 V. Their compression ratios were calculated 56% and 61% for the pH 3-10 gradient and the pH 4-6 gradient, respectively. The resu lts suggest that catholytes and anolytes entered about the same level for both pH gradients. Cui et al. suggested that an increase in the viscosity of catholytes and anolytes may reduce the compression of pH gradient.79 This was studied by adding diffe rent concentration of glycerol in both catholytes and anolytes (glycerol concentration at 25 %, 50% and 65%). There was no evidence of any decrease in pH gradient compre ssion. In fact, it was observed that there was decrease in the peak distance between GFP a nd RPE (137, 122, and 98 pixels for 25, 50, and
62 65% glycerol, respectively), indicating an incr ease in the compression. Also observed was an increase in the focusing time at a higher viscosit y, suggesting a longer time is required to form a pH gradient. 3.5 Conclusion IEF theory predicts that the IEF resolution is independent of se paration length under the conditions that meet the assumptions in the deri vation. The assumptions include (1) negligible EOF; (2) negligible Joule heating; (3) constant pH gradient, d(pH)/dx,; and (4) constant mobility difference at pI value, d /d(pH). The conclusion is significant since it supports th e use of IEF in a microfabricated device without sacrificing the resolving powe r, and at the same time the analysis time is significantly reduced due to shor ter channels and higher el ectric field strengths. Although IEF was performed in channels III-V for separation of GFP peaks with a pI difference of 0.1 pH units, there was a slig ht increase of separation resoluti on with the separation length. The deviation of the experimental results from th e theory is possibly due to the presence of EOF and/or Joule heating as discussed above. The LIF whole-channel imaging detection (WCI D) system is a very useful tool for studying IEF dynamics. Simultaneous illumination of an entire channel with an expanded laser line and collection of fluorescence emission by a CCD camera enable elimination of the mobilization step that is often practiced in capill ary or channel IEF. In addition, WCID provides useful information about the dynamic behavior of protein migration during the process of IEF, and the temporal information could be used for determining the shortest focusing time as demonstrated in this work. pH gradient compression repor ted recently by Cui et al.,79 that is, the pH gradient is compressed to the middle of a channel exits, ra ther than crossing the whole channel from the
63 anode to cathode. This phenomenon is important because it conveys the exact distance of a pH gradient actually formed, rather than the assumed capillary or slab length. In addition, presence of EOF also affects the actual distance of a pH gradient sin ce some separation medium might flow out of the separation channel or capillary These factors suggest an immobilized pH gradient is probably a direction for future work,103-105 especially considering the claim that a true steady state in absence of immobilized pH grad ients cannot be achieved experimentally at all according to theoretical simulation.30 The IEF process performed in 6 channel device gave an idea of process parameters. These are now used to perform the first dimension separation in 2-D microfluidic device and are discussed in later Chapters. Table 3-1. The effects of IEF distance on se paration resolution. Sepa ration resolution was calculated as discussed in the text. The standard deviation wa s obtained from three repeating experiments. The separation vo ltage for each row was listed in the first column while other experimental conditi ons were the same as in Figure 2. Channel II Channel III Channel IV Channel V Separation Length (cm) 1.9 3.0 4.1 5.2 Resolution (200V) 0.95 + 0.07 1.1 + 0.2 0.7 + 0.1 1.5 + 0.4 Resolution (300V) 1.1 + 0.4 1.2 + 0.1 1.4 + 0.2 1.6 + 0.3 Resolution (400V) 1.0 + 0.1 1.5 + 0.3 1.7 + 0.2 2.1 + 0.2 Resolution (500V) 1.2 + 0.3 1.4 + 0.3 1.6 + 0.9 1.7 + 0.4
64 Figure 3-1.Illustration of separation process for cl osely spaced peaks. A) Representative diagram of channel with two closely spaced proteins B) The real picture of two proteins GFP and RPE focused in microchannel. RPE on the right has two closely spaced peaks which are barely resolved. C and D) I llustration of minimum spatial difference required between two peaks to be resolved properly. Two closely spaced Gaussian peaks (green and blue, both have sa me characteristics) are separated by 2 in picture C and 3 in picture D. It should be at least 3 of the gaussian curvature of the peaks for them to be resolved. The red line is s uperposition of two gaussian lines green and blue. -5 0 5 0 0.5 1 1.5 2 2.5 2 -5 0 5 0 0.5 1 1.5 2 2.5 3 2 to 5 cm 100 m A C B D
65 0 5 10 15 20 0 10 20 30 40 0 100 200 300 400 Distance (mm) Time (min) Intensity (a.u) Figure 3-2. The temporal images of IEF separati ons of GFP in polyacrylamide gel. The length of channel is 3 cm. The electric field applied is 200 volts and total time for focusing is 40 minutes. The device used was made in glass. Figure 3-3. IEF of different pr oteins in gel and linear polym er. A) IEF of GFP, BSA and Myoglobin in linear polymer. B) in polyacrylam ide gel. The electric field strength in both the cases is 66 V/cm. Time of comp letion for linear polymer is 20 minutes whereas for gel is 40 minutes. 0 500 1000 1500 2000 2500 280480680880108012801480 Pixels Intensity (a.u) 100 200 300 400 500 600 700 800 0200400600800100012001400 pixelsIntensity (a.u ) A B Myoglobin Myoglobin GFP GFP BSA BSA
66 Figure 3-4. Comparison of front speed for different separation medium Figure 3-5. IEF of GFP, RPE in HPC/HEC linear polymer. A) line intensity profile at the center of the channel. B) Full channel intensity prof ile in the software where the intensity is plotted over the region of interest. C) The image acquired by the CCD camera. The minor peaks of GFP are lost when the image is converted to JPEG format. Comparison of front speed 0 0.2 0.4 0.6 0.8 1 1.2 1.4 3.513.523.533.5 Time (min)Speed (mm/min) Linear polymer Cross-Linked polymer IEF of GFP-rpe @ 500 V, ch-4(50mm) 1.83%HPC 200 400 600 800 1000 1200 500100015002000 PixelIntensity (a.u ) 500 v A B C GFP RPE Pixels
67 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0500100015002000250030003500 PixelIntensity (a.u) Channel-II Channel-III Channel-IV 0.05 0.1 0.15 0.2 0.25 280320360400 0.05 0.1 0.15 0.2 0.25 22002700 Channel-V GFP RPE Figure 3-6. The effects of the separati on distance on IEF separation. GFP (5 ng/ l) and RPE (3.85 ng/ l) were separated in IEF with channel length of 2.1, 3.2, 4.3 and 5.4 cm. Minor peaks of GFP are shown in the expa nded views of a part of electropherograms for 2.1 cm and 5.4 cm channel. A voltage of 500 V was applied to all channels; and the cathode was on the left and the anode was on the right. The pH gradient was 310, and the separation medium was prep ared from 1.83% HPC/0.93% HEC stock solution. The IEF times were 3, 6, 11, and 18 minutes for channels of length 2.1, 3.2, 4.3 and 5.4 cm, respectively. Each pixel in x axis approximat ely corresponds to 20 m. The IEF electropherograms were shifte d in both axis for clarity, the degree of up-shift is indicated by the background intensity and that of right-shift is by the starting point.
68 Figure 3-7. Separation resolution based on peak width and separation distance. A) Three different resolution R values (1, 1.25, and 1.5) and its corresponding grey scale image is shown. The relative peak hei ghts are same. Dotted lines represent the original pulses whereas the firm lines are afte r superposition. B) Same as in A, but the relative peak heights are at a ratio of 1:0.75, 1:0.5, and 1:0.25. 0 2 4 6 8 10 12 0 0.5 1 1.5 DistanceConcentration 0 2 4 6 8 10 12 0 0.5 1 1.5 DistanceConcentration R= 1 R= 1.25 R= 1.5 R= 1.25 R= 1 R= 1.5 0.25 0.50 0.75 A B
69 0 0.5 1 1.5 2 2.5 3 3.5 9001100130015001700 PixelIntensity (a.u)1000 V 700 V 500 V 200 VGFP-2 GFP-1 GFP-3 RPE Figure 3-8. The effects of separation voltage on IEF. IEF electropherograms at separation voltages indicated. 5.2 cm channel was used for all experiments. Other conditions were the same as in Figure 3-6. 1/E (cm/V)Time (min.) 0 10 20 30 40 50 60 0.0000.0100.0200.030 Channel-II Channel-IV Channel-V Figure 3-9. The relationship between the focusi ng time and the inverse of the electric field strength. The pH gradient used was 310. The separation length is represented by the channel number while the separation vol tages were indicated by the electrical field strength. Other conditions were the same as in Figure 3-7
70 0 0.5 1 1.5 2 2.5 0500100015002000IEF Location (pixel)Intensity (a. u.)pH 3-10 gradient pH 4-6 gradient GFP RPE Figure 3-10. Comparison of IEF electropherogr ams of GFP and RPE between pH 3-10 and pH 4-6 gradients. Channel V was used with a separation voltage of 1000 V. Other conditions were the same as in Figure 3-2.
71 CHAPTER 4 TWO DIMENSIONAL SEPARATION OF PROTEINS 4.1 Introduction Two dimensional protein separation in the slab gel format is well developed and has been around for almost half a century. Although this method provides very high resolution, it has a few drawbacks. The process involves complex handling procedures. First IEF is performed separately and then transferre d in gel for second dimension se paration. Thereafter a tedious staining process is required to vi sualize the separated proteins. The time involved is also very long, more than ten hours. Several groups have explored the applicati on of microfluidics for multidimensional separation of proteins/peptide s because of faster operation and other advantages offered by microfluidics as described previously. Chen et al.48 had used 6 layers of PDMS for performing IEF and SDS-PAGE, but this process involves alignment, bonding, debonding and realignment of PDMS layers which is also very tedious process. Li et al.49 had used plastic microfluidic devi ce to couple IEF and SDS-PAGE using one cross channel and 10 parallel channels that are ort hogonal to the cross channel. The cross channel is used for IEF whereas the orthogonal channels are used for SD S-PAGE. The medium of separation is viscous liquid. This is a pretty novel te chnique, but there is a problem of contaminating first dimensional constituents with the second dimension. The cross-talking between different media is of important concern when performing this type of multidimensional separation. In this research, an array of microfluidic valves is created to obtain a reliable interface which prevents one separation medium from contaminating with the other separation medium. A part of this Chapter has been published : Das, C.; Fredrickson, C.K.; Xia, Z., Fan, Z.H. Sensors and Actuators A 134, 271-277, 2007
72 4.2 Materials and Methods The reagents used for IEF is already descri bed in Chapter 3. The additional chemicals required are as follows. Sodium dodecyl sulpha te (SDS), Tris-HCl (1M) and Tricine is purchased from Fisher scientific (Atlanta, GA). Hydroxycyclohexylphenyl-ketone (HCPK) and acrylamide monomers (acrylamide / bisacrylam ide, 19:1 ratio in 40% stock solution) are purchased from Sigma (St. Louis, MO). The SDSPAGE medium is made of polyacrylamide (812% T), whereas IEF medium is made of linear polymer, constituents of which are same as discussed in Chapter 3. %T defines the total amount of polymer concentration in solution (the amount in gram of acrylamide + gram of Bis per 100 ml of solution). An acrylamide concentration of 8-12%T is generally used depe nding on the type of proteins separated. For exclusive SDS-PAGE, the proteins are precomplexed with SDS (1%) at 95 0C for 5 minutes before separation (section 4.7). Similarly 0.1% SD S is electro kinetically injected inside the acrylamide gel before IEF and then 10% SDS solu tion in slot for SDS-PA GE only in Figure 4-2. For all other cases, 10% SDS is injected inside the gel from slot only after IEF is performed. Following are the detailed steps required to be performed for a standard 2-D separation. 1. Photopolymerization: The monomer solution fo r photopolymerization is prepared by adding 125 l of acrylamide/bis-acrylamid e monomer solution (19:1), 20 l of Tris-HCl (1 M concentration), 20 l of Tricine (1 M concentration), 65 l of HCPK (100 mM in propanol) to 260 l of DI water in a vial. The contents are mixed well and kept under cover to prevent it from room light. The 2-D device is then placed under microscope and the monomer solution is added to the top slot and two reservoirs (for IEF). Vacuum is used to suck the monomer solution into the parallel channels and cross channel. Care should be taken such that there are no bubbles anywhere inside the channel. The rema ining liquid is then drained. DI water (2 l) is added in each circular reservoir. Since it is impossible to drain the monomer solution totally from IEF reservoirs, so adding DI water dilutes the remaining monomer solution and will not polymerize in th e reservoirs. The monomer solution is also drained from the two slots but it is ensured a th in layer of solution is present near the junction of parallel channels with slot such that no air bub ble is trapped inside the channel during polymerization. The device is then flipped over a nd the mask is placed carefully on the top of the device. The mask is carefully placed such that the cross channel is completely covered by the mask. The device thus made ready with mask covering the cross cha nnel, is exposed to
73 UV light. The high intensity of UV light (75 Wa tt lamp) polymerizes the parallel channels within 50 seconds. The diameter of UV light is around 2.5 cm. So during photopolymerization only 19.6 square cm area of the device is polymerized. The device is then removed from UV source. The mask is removed. The cross channel did not polymerize due to presence of mask. One of the IEF reservoirs is filled up with DI water and contents of the cross channel is sucked out of the other reservoir. The water is added to the reservoir is to ensure air does not go inside the cross channe l when monomer solution is removed from the cross channel. Now both the slots are filled up with monomer solution and the device is again exposed to UV light for furthe r polymerization, but this tim e without the mask. The whole device is polymerized for about 5 minutes. Th e device is again removed from the UV source and gel from the slot is removed. All the slot s and reservoirs are filled up with DI water and the device is kept under DI water for future use. 2. IEF in cross channel: The next step is to pe rform IEF in cross channel. The IEF medium is prepared as described previously. 5 l of this medium is pipetted in to one of the reservoirs of cross channel. The liquid is sucked into th e cross channel and the remaining liquid is removed. Anolytes and catholytes are loaded onto two reservoi rs. The negative electrode slot is filled with 10% SDS solution whereas the posit ive electrode slot is filled up with TrisHCl buffer (20 mM). The device is then placed in the platform where the experiment is performed. 4 electrodes required for the 2-D separation is carefully lowered into the reservoirs and slots of the device. The electrode s used for IEF is 0.5 mm thick platinum wire whereas flat 2 cm platinum foil (0.05 mm thick) is used for SDS-PAGE electrodes in slots. The laser line is focused onto th e cross channel and 100 V/cm is applied to the cross channel whereas the slot electrodes are kept floating. Proteins will focus inside the cross channel and will be visible as distinct fluorescent peaks. It takes around 5 minutes to perform IEF. 3. SDS-PAGE: The laser line (20 mW) is now sh ifted from the cross channel towards the parallel channel. The laser line is usually pl aced either at 6.5 mm or 2 cm away from the cross channel. After IEF operation is over, 1200 V is applied across th e parallel channels through the slot electrodes. 10% SDS inside the negative slot is electro kinetic ally injected inside the gel. It meets with the proteins al ready focused on the basis of charge inside the cross channel. The SDS then binds with respec tive proteins to form the negatively charged complexes inside the parallel channels. These pr oteins as they travel inside the parallel channels get separated on the basis of molecu lar weight. These separa ted proteins can be seen arriving at different times to the zone wh ere they are detected by a laser line. The whole process is continuously recorded by the CCD camera (exposure time 1 second with full analog gain). The images are later processe d in MATLAB to obtain a 2-D map. It takes around 5-8 minutes to perform SDS-PAGE. Figure 4-1 A) shows the schematic of two type s of devices used for the 2-D separation. The first design consists of one cross channel a nd several parallel channels. The cross channel AoBo is for IEF process. Once the IEF is comple ted, the second dimension separation is done in orthogonal parallel channels CD. Two different separation medi um is used for IEF and SDS-
74 PAGE. Figure 4-1 B) shows the similar layouts of device but a pa rt of the CD channels are staggered with each other. These staggered intersect ions ensure all samples in the first dimension flow into the second dimension. These devices are made through a glass master, and the channel is 100 m wide while the CD cha nnels are separated by 360 m (center to center). The number of orthogonal channels is 29 for both design, but th ere is an additional channel on the left for the staggered design to realize the staggering arrang ement. The most important aspect of 2-D separation is to achieve reliable ge l valve, which is discussed next. 4.3 Polyacrylamide Gel Valve An array of pseudo valves is designed by having in situ gel polymerization for all parallel channels. The solution for making gel valves is prepared by mixing HCPK, a photo-initiator ( 100 mM in propanol) with acrylamide monomers (8 %T), and DI water in a volumetric ratio of 1:3:6. A mask is used to define the physical lo cation of gel such that the IEF channel is left unpolymerized. Figure 4-2A shows the mask pattern us ed to check the fidelity of pattern transfer to gel. The gel is formed by introducing mono mer solution of acrylamide between two glass slides separated by a spacer of measured th ickness (Figure 4-2B). The spacer is made by spinning photoresist SU-8 at a certain speed to gi ve the desired thickness. Then it is exposed to UV light with a mask to form the desired gel pa ttern. The lines on the mask will block UV light exposure and lead to no polymerization in the region. Table 4.1 shows the effects of mask dimension as well as gel thickness on the width of gel formed. The fidelity of pattern transfer is better for thinner gels when a ll experimental conditions (exposu re time, energy density of UV light and gel composition) are kept constant. Mask widths are the actual line thickness printed out.
75 Another aspect of photopolymerization is to reduce the time required for exposure to UV light. Photopolymerization is not strictly de fined by direction of light. With long time of exposure to UV, solution will polymerize nearby ev en in the presence of mask as the reaction initiated by HCPK can propagate in all the direction. Monomer a nd photoinitiator concentrations are varied to investigate their effect on time of exposure. Figure 4-3 shows the plot of polymerization time with respect to acrylamide concentration, keeping photo initiator ratio concentration constant. The plot is linear in nature, suggesting the exposure time should be reduced for higher concentration. Below 4% T, monomer is still a viscous liquid after polymerization. Above 8% T the polymeri zation process takes less than 5 seconds. The percentage of photoinitiator also affects the polymerizat ion time. Figure 4-4 shows the rate of polymerization as a function of the c oncentration of HCPK at a constant acrylamide concentration (6% T). The polymerization process is almost independent of time below a certain concentration of HCPK (< 4%), as with ve ry long time of exposure even very small concentration of photo initiator (HCPK) w ill ultimately polymerize the whole liquid. These results discussed above are obtained by performing experiments on a glass substrate. The optical depth of glass slides and plasti c devices are different. Hence the time of polymerization will be different in plastic device but the overall trend will remain the same. The UV exposure time and power required is arrived at by performing a series of experiments at different power and exposure time. For higher powe rs the polymerization is fast but bubbles tend to form inside the channel due to heat from UV light, whereas at low power the required time of exposure is long, and it tends to dry up the wells and air goes inside th e channel. The optimum condition of polymerization in pl astic device is as follows: Acrylamide / bisacrylamide: HCPK = 2:1 a. UV-energy density @ 365 nm wavelength = 40 mW/cm2
76 b. Time for polymerization = 50 seconds Next step is to perform selective polymerizati on in parallel channels in 2-D device. All the channels in the device in Figure 41 are filled with the solution. The device is then turned upside down and a chrome mask with a 100 m line is placed on the top (the film side) of the device. To make gel valves at precise locations, an alig nment is made between the line in the mask and AB channel in the device. Then the device is ex posed to UV light for about 50 seconds. Figure 45 shows the schematic of th e UV-polymerization process. Photo-initiated polymerization took place in all parallel cha nnels (CD from Figure 4-1 A) because they are exposed to light, whereas the solution in the cross channel (AoBo from Figure 4-1 A) channel did not polymerize since UV light is blocked by the mask. Non-polymerized solution in the AoBo channel is removed by vacuum and replaced with the first dimensional separation medium. Figure 4-6 shows the sele ctive polymerization of acrylamide in microchannel. The gel is dyed with congo red. M onomer solution in the proximity of the region exposed to UV light tends to polymerize as well. This obstacle is overcome by adjusting the size of the mask line and UV exposure time. Once the selective photopolymer ization is completed, the nonpolymerized cross channel (AoBo) monomer solution is sucked out. IEF solu tion is then introduced in that channel and IEF is performed in first dimension, details of which are discussed in next section. 4.4 IEF in First Dimension IEF in first dimension is performed by appl ying 200 V across the channel AoBo filled with IEF medium as shown in Figure 4-7 B. The parallel channels are polymerized with polyacrylamide gel as discussed above. The IEF medium is linear polymer of 1.83% HPC / HEC and all other constituents are same as the expe riments performed in the 6 channel device. Figure
77 4-7 B shows the separation of GFP and RPE in cross channel whereas Figure 4-7 A shows similar separation in 6 channel device They are very similar in nature. As seen from Figure 4-7 the PAA gel in the parallel channel did not contaminate the IEF medium. Also the IEF medium did not penetrate in side the gel, resulting in clean separation of two proteins. However these two proteins can not be used for SDS-PAGE separation as they lose their fluorescence after complexation with SD S. This will be discussed in detail later. So new fluorescently labeled proteins are also used to demonstrate th e robustness of the gel valve in the cross channel. Figure 4-8 shows IEF of four proteins (t hree labeled and one naturally fluorescent protein) in the cross channel. The proteins used are paralbumin (pI 4.10), ovalbumin (pI 4.6), BSA (pI 4.6) and GFP (pI 5.0). The Fi gure 4-8 shows three distinct peaks because BSA and ovalbumin share the same isoelectric point and they could not be separated in IEF. The linear polymer for IEF medium is increas ed to 2.75% HPC to reduce the effect of hydrostatic imbalance. The separation is achieved by applying 300 V across the cross channel.Table 3 shows the comparison of resolu tion between the cross channel of 2-D device and that of 2, 3 and 4 cm channels. The resolu tion of 2-D device (calcula ted for major peaks of GFP and RPE) is 1.95 whereas for 2 cm channe l it is 2.12. The peak widths and distance between the peaks for two proteins are also si milar. The experimental conditions like field strength (100 V/cm) and IEF medium (similar concentration of linear polymer and CA and protein concentration) are same for all the cases. The resolution for 3 cm (2.38) and 4 cm (2.64) is however better than 2D device (1.95) mainly because of channel length. The increase in channel length leads to better control of hydrodynamic flow and less effect of pH gradient compression.79, 106 However these two proteins can not be used for SDS-PAGE separation as they lose their fluorescence after complexation with SDS. So fluorescently labeled proteins are
78 also used to demonstrate the r obustness of the gel valv e in the cross channe l. Figure 4-8 A shows IEF of four proteins (three labe led and one naturally fluorescent protein) in the cross channel. The proteins used are paralbumin (pI 4.10), ovalbumin (pI 4.6), BSA (pI 4.6) and GFP (pI 6.0). Three distinct peaks (electrophe rogram in Figure 4-8 B) are only visible because BSA and ovalbumin share the same isoelect ric point and they could not be separated in IEF. Figure 4D shows the plot of isoelectric point of the above proteins with respect to their physical location inside the cross channel. The linear trend (R2=0.99) of the plot is very important as it shows the pH gradient generated in IEF is uniform and lin ear in nature. However if the data is further analyzed, it reveals a high compression of pH gradient.79, 106 The cross channel is 15 mm long and if it is assumed that the pH gradient is es tablished uniformly along the entire channel length, 1 pH unit will span a length of 2.14 mm of the cr oss channel. But the experiment shows a length of only 0.8 mm of cross channel is used for 1 unit pH gradient. This translates to a pH gradient compression of around 60%. This suggest some amount of catholytes and anolytes have entered inside the cross channel to facilitate the gr adient formation. This fact can however be advantageously used in design of the device. The cross channel can be made longer without increasing the number of parallel channels or spacing between them for the similar resolution. The increase in length will also ensure less hyd rodynamic instability and less effect on IEF due to electroosmotic flow. 4.5 Numerical Simulation Integrated microfluidic systems with complex network like the 2-D device needs optimizations before being put to real experi ments. CFD simulation can accelerate the design, development and optimization of these integrat ed microfluidic devi ces. Though it is still impossible to simulate the exact conditions prev ailing in experimental co nditions for 2-D device,
79 appropriate assumptions can make the problem defi nition simpler and results thereof can be used to understand the underlying physics of the problem. One of the most important issues of the 2-D se paration is the transfer of protein from cross channel to parallel channels. The relative positi on and blob size of proteins with respect to parallel channel will decide the way the protein will move while getting transferred. In order to understand this transfer phenomenon CFD-ACE mult iphysics software is used for simulation. The problem description is defined below, fo llowed by the methodology used for simulation. Geometry: Two dimensional planar. Rectangular grid is generated by CFD-ACE software. Figure 4-9 shows the geometry. The full scale of the device is used for modeling. The parallel channel has a total length of 6.5 cm while th e cross channel length is 1.7 cm. An individual channel is 100 m wide and the distance between the parallel channels is 360 m.This is a problem of transient electrophoresis of charged molecules/species. To do that all the parallel channels are divided into small segments, so th at a particular species can be initially placed anywhere in the channel and its migration under an elec tric field can be studied. In order to simulate this type of problem, there is a f eature in CFD-ACE which allows the electrical conductivity to be calculated as a function of ion concentration. Two different types of species are selected with different mass diffusivities (2.5 x 10-9 5.5 x 10-9 m2/s) and one or both the species are selected as negatively charged. Th e reason behind selecting the negative charge for species is that the protein in SDS-PAGE separation becomes negatively charged after complexation with SDS. The common buffer strength is 4x 10-6 1/Ohm-m. Usually an electric field of 1000 V is applied across the electrodes. A time step of 0.001 s and Euler first order scheme is chosen for the transient solu tion with 20 iterations in each step.
80 4.5.1 Results and Discussion: The cross channel is filled up with one species whereas the parallel channel are filled up with a different species. An electr ic field of 100 V/cm is applied to the parallel channels via the slot. The negative cross channel species are gradually seen to move into the parallel channels. Figure 4-10 shows such a transfer of species from cross channel. The numerical results in Figure 4-10 A matches pretty well with expe rimental results shown in 4.10 B. The fluorescent proteins are seen to split less uniformly in one direction. Two possible reasons can be given, A) The para llel channels are staggered, so the electric field can not travel straight from parallel channels above the cro ss channel to the staggered ones below the cross channel. Two possible cases of elec tric field orientations are shown in 4.10 C and D. In first case the electric field splits equally into channels whereas the second case shows that the field does not split but follows a tortuous line. Either of the cases can happen. So for this experiment, proteins split preferentiall y in the direction of least resistance of flow path (electrical field) as in Figure D. It is difficult to justify similar pref erential movement in numerical simulation. B) Presence of continuous hydrostatic flow in th e cross channel, which may have moved the proteins in one particular dire ction in cross channel. This co ndition however is not present in numerical simulation. If IEF is performed in cross channel first, the proteins will focus as bands in the cross channel and when these bands of proteins are tr ansferred in the parallel channels, they will transfer either into a single channel or into multiple channels depending on the position and size of the bands. A few cases are simulated as s hown in Figure 4-11. Figure 4-11 A shows the case where the protein after IEF is focused in between two parallel channels (topside) but since the parallel channels are staggered, the proteins are trifurcated into three channels while it is transferred. Figure 4-11 B shows the similar case, but here the protein plug is shifted in such a
81 way that it is in between two lower parallel channels. So during transfer the protein gets bifurcated neatly into two channels. Figure 411 C and D shows a smaller plug size. The plug is just above one lower parallel channel and gets en tirely transferred in that channel (Figure C) whereas for Figure D the plug position is shifte d and it gets bifurcated again. Figure 4-11 E shows the experiment result for protein plug transfer from cro ss channel to parallel channel. The position of electrode in the slot of parallel channels can distort the electric field distribution inside the different pa rallel channels, which will in turn effect the electrophoretic movement of charged species. If a conventional pointed electrode is used, the channels near the electrode will have higher electric field whereas the channels away from the electrode will have less electric field. The differen ce in electric field di stribution will lead to uneven electrophoretic speed of the charged species. The second dime nsion separation is based on the electrophoretic mobility of SDS complexed proteins. The detection of these proteins is done by illuminating the channels with a laser line at one particular loca tion near the end of the parallel channels. As the proteins move past the detection zone, the CCD cam era collects the signal with time. Proteins of different molecular weight will travel with differ ent electrophoretic mobility (due to same charge to mass ratio attained during complexation of SDS with proteins, higher molecular weight proteins will have higher resistan ce to flow in gel due to bigger physical size and will have lower mobility) in the gel filled channels. Any artific ial change of that mobility due to different electric field will cause the proteins to travel fast er or slower than their counterparts in different channels. This will cause an error in the migratio n. An experiment is performed to demonstrate the problem. The parallel channels are polymerized with gel and the cross channel is then filled with a protein (GFP 1 g/ml) solution. Two pointed electrodes are placed on the two slots (at the center of slot) at the end of the parallel channels and an electric field is applied. Figure 4-12
82 shows the irregular positions of protein plug (GFP in this case) in different channels due to distortion of electric field. Non uniformity of gel can also be a reason for this behavior. It is difficult to ensure the uniformity of gel, so grea ter control on electric fiel d uniformity is applied. It is very important to ensure that the electric field is constant across all the channels. This can be ensured by using an electrode which covers the entire slot. Full scale simulation is carried out in CFD-ACE to evaluate the effects of elec trode size and position on the transfer of plugs in parallel channels. Three cases are evaluated, a) pointed electrode at center of slot b) pointed electrode at one corner of slot c) a flat electrode which covers the entir e slot. The purpose is to ev aluate the distortion of electric field at different condi tion. Electric field of 1000 V is a pplied across the electrodes. The buffer conductivity is 4 x 10-6 1/Ohm-m and the mass diffusivities are 2.5 x 10-9 m2/s for parallel channels and 5.5 x 10-9 m2/s for cross channel. The mobility of the proteins or any other charged species will depend directly on the magnitude of the electric field. Figure 4-13A shows the field distribution for all the three case and as expected the distortion of electric field is more for the pointed electrode when compared with flat el ectrode. Figure 4-13 B show s the approximate field strength at a particular loca tion across all the parallel channe ls. The configuration where the pointed electrode is at one corn er shows maximum distortion in field of around 8% between two extreme channels. This will lead to an error of around 9 % in time taken by proteins to travel the same distance in channels. The center electrode c onfiguration will lead to an error of around 2% on time. The flat electrode shows almost no variat ion in electric field. The CCD camera used for LIF detection can also used to monitor the protei n front movement inside the parallel channels. Since SDS complexation renders uniform charges to mass ratio to the proteins, the protein of one particular mass will travel with cons tant velocity inside the parallel channels at constant electric
83 field.107 Experimental results for pointed el ectrodes show a bow shaped (deviationm 133 ) migration pattern. This results in error of kDa 2 1 in the estimate of molecular weight of ovalbumin. The same experiment when performed w ith flat electrodes, gives more uniform field strength across the channels and results in sim ilar migration velocity and hence flat shaped (deviation m 23 )migration pattern in all the channels. The resultant error will be much less at kDa 2 0 Figure 4-14 shows the plot for location of th e protein peak with respect to time and the trend is very linear (R2 =0.999). The inset plot shows almost constant velocity profile of the same data (4 pixels/s or 80 m/s) with second order accuracy. Next different electrode configurations to control the two di mensional separations are discussed. 4.7 SDS-PAGE Separation In conventional SDS-PAGE technique, prot eins are allowed to bind with SDS through hydrophobic interaction. A constant amount of approximately 1.4 gram of SDS is complexed by each gram of protein to form the negatively charge d complexes, with similar charge to size ratio. The gel used for this separation is a random mesh work of cross linked individual polyacrylamide mesh and has sieving effect on proteins during the separation procedure. When SDS-PAGE is performed by applying electric field the complexe s migrate with a velocity which is dependent only on size of complex.108 Hence this process can separate th e proteins based on their sizes. The linear relationship between the logarithm of molecular weight and mobility allows one to determine unknown sample molecular weight by comparing with standards. SDS-PAGE separation is first tried on two di mensional devices. The proteins are first complexed off chip with 1% SDS at 95 0C for 5 minutes. Four prot eins are used a) trypsin Inhibitor (20 kDa), carbonic Anhydrase (31 kD a), ovalbumin (45 kDa) and bovine serum albumin (66 kDa). The device is filled with monomer solution and UV photopolymerization is
84 performed to form gel. The cross channel is cove red with a mask as discussed before and is not polymerized. Two different gel st rengths are used (10 & 12%). The gel also contains Tris HCl (40 mM) and tricine buffer (40 mM) for supporting the electric field required for PAGE operation. The precomplexed proteins are loaded into the cross channel and an electric field of 1200 V is appl ied across the parallel channels. The proteins traveled from cross channel into the parallel channel and got separated due to difference in molecular weight. They are observed as separate peaks by the CCD camera when the channels are illuminated by a laser line at a distance of 6 mm from the cross channel. As seen from the Figure 4-15 the proteins separated shows consistent linear trend when their molecular weight is compared with their electrophoretic mobility. This is a very important characteristic for SDSPAGE separation as proteins of unknown molecula r weight can be separated and its molecular weight can be found out from the linear rela tionship with its elec trophoretic mobility. 4.8 Two-dimensional Separation The next step is to integrate both IEF and SD S-PAGE separation in 2D device. In flat bed SDS-PAGE system, the gel medium contains abou t 1% SDS, so that when the proteins are transferred to the second dimension, the SDS in side the gel can form complexes. However in 2D device, since the polymerization of parallel cha nnel are done before IEF, any SDS in the cross channel can severely effect the IEF process. So the cross channel needs to be thoroughly cleaned before the IEF separation is tried. Also th e gel inside the paralle l channel should be well polymerized so that SDS does not leak out from the parallel channel into the cross channel. To study the effect of SDS on IEF, a separate experiment is performed in 6 channel device. A small amount of SDS (0.1%) is added to IEF so lution. As seen from the Figure 4-16, the SDS, even in small quantity, can affect the focusing process. RPE
85 forms a sharp peak without SDS. With SDS, how ever, a broad peak is visible just at the beginning of the experiment but became more and more diffused as the experiment progressed. Same effect is also observed when the experime nt is performed in 2-D device with SDS in PAA gel in cross channel. Next a different approach is tried, where th e parallel channels are polymerized without SDS, so that there is no chance of IEF medium coming in contact with SDS. After the IEF is performed in cross channel, SDS is introduced in negative electrode slot of parallel channels. Thereafter an electric field (800 V) is applied and SDS is electr okinetically injected into the parallel channel. As the SDS we nt and interacted with proteins, they formed complexes and started traveling along the para llel channels. Figure 4-17 sh ows the results of IEF and corresponding transfer of proteins in the second dimension. The laser line for illumination in the second dimension is kept very close to the cross channel (5 mm away) to observe the transfer of proteins. As seen from the Figure the proteins are seen are transf erred smoothly into the parallel channels (GFP in 7th and RPE in 10th Channel). Since the detection is made very close to cross channel, it is difficult to ascertain ab out the SDS complexation with proteins. Though GFP and RPE are naturall y fluorescent proteins and gi ves good peaks in IEF, they are not very useful for second dimension separation as the prot eins loose their fluorescence as soon as they form complex. Although the proteins are seen traveling in the parallel channels, they gradually disappear and if the laser line is placed further down the channel, no fluorescence is detected. However there will be no such pr oblem with labeled proteins as the SDS will not react with fluorescence of the la beling dye. Hence it is decide to label non fluorescent protein with a fluorescent dye.
86 Molecular probes fluorescent labeling kit is ch osen for labeling the proteins. The dye is insensitive to pH gradient of 4-10 and is idea lly suited for our application. The absorption and fluorescence emission maxima are 494 nm and 519 nm respectively. The labeling procedure is well documented in literature and is discussed he re in brief. The proteins are all separately labeled. The protein of concentr ation of 2 mg/ml is mixed with sodium bicarbonate solution (~ pH 8.3) to raise the pH of reaction mixture as the dye reacts efficiently at pH 7.5-8.5. The solution is then mixed with the dye and stirre d with a magnetic stir rer for 1 hr at room temperature. After the reaction is over the labeled protein is puri fied from excess dye by passing it through a resin column and PBS (phosphate buff er) elution buffer. As the protein solution passes through the column, it gets separated into two fluorescent bands. The first band is that of protein and the following band is th at of unlabeled dye. The protein is collected and stored at -20 0C. Five proteins are labe led in this way. The proteins are ta bulated below in the Table 4.3. The labeled proteins are used to for all subsequent experiments. First a sing le protein is used to demonstrate the IEF and SDS-PAGE together. The protein is first focused in IEF and then transferred in second dimension for PAGE. Figure 4-18 A and B shows the focused peak of BSA in both IEF and SDS-PAGE. The protein is focused in IEF and then is nicely transferred to one of the parallel channels. Next three proteins are tried together. The prot eins are so chosen that all proteins can not be separated alone in any one dimension. Separa tion of BSA, trypsin and hemoglobin is tried as in Figure 4-19. The experimental conditions are same as described before. The three proteins are focused in IEF first. Since BSA and Trypsin ar e having same isoelectric point, they are not
87 separated in IEF. So there are tw o major bands in IEF. The band to wards the acidic side contains both BSA and trypsin whereas the other band contains hemoglobin. The second dimension (SDS-PAGE) however sepa rates the BSA and trypsin as they differ in their molecular mass. Trypsin being the smallest in terms of molecular weight travels the fastest in gel followed by hemoglobin and BSA. He moglobin is already separated in IEF itself and it travels in separate parallel channel. IEF for most of the cas es are done at 3-10 pH gradient, but there may be specific need to do the IEF at less pH gradient. One of them is improving the resolution in IEF. The next expe riment is performed with 3-5 pH gradient in IEF but with different set of proteins (B SA, ovalbumin and trypsin). The proteins got separated as expected based on their molecular weight as in Figure 4-20. But there is some discrepancies regarding the isoe lectric point of the prot eins. All the proteins have same pI point and hence can not be separated in the cros s channel in IEF. However SDSPAGE should separate all three because of difference in molecular weight. Trypsin and ovalbumin got separated in the same channels but BSA is seen traveling in the channel far away from the other two proteins. One possible explana tion is the change of pI point of BSA due to labeling. The correlation between molecular weig ht and electrophoretic m obility is quite high (20.9928 R ). Further experiments are performed with th ree or four proteins to characterize the system at different molecular weight and isoelectric point. The above Figure 4-21 shows the 2-D map using three different proteins (BSA, ovalbumin and hemoglobin) in 3-10 pH gradient. Hemoglobin is more basic protein and hence is focused towards the right side and transferred to the ex treme right channels. Trypsin is more acidic (pI 4.6) and has lower molecular weight (20 kDa) a nd got focused to wards left side and traveled
88 fastest down the parallel channels. Carbonic anhydrase has pI of around 5.9 and molecular weight of 31 kDa and hence is in the mi ddle with respect to other two proteins. There after four proteins are tried. They are BSA, ovalbumin, trypsin inhibitor and carbonic anhydrase. First three protei ns have same isoelectric point (4.6) whereas the last one has isoelectric point of 5.9. So ideally first th ree proteins should get separated only in second dimension and in same perpendicular channel. One of the important parameter for 2-D separation is to en sure proper on-chip complexation of SDS with protein. Presence of SDS already in gel enhances the complexation process. However it is difficult to photopolymeri ze gel inside the channels with SDS in monomer solution. So SDS (0.1%) is elect rokinetically injected after polymerization into the gel before IEF. IEF in cross channel is not heavily affected because of very low co ncentration of SDS in gel. The injection time is about 10 minute to en sure uniformity of concentration of SDS though out the entire length of parallel channels. Figure 4-22 shows the se paration of all four proteins. The proteins are well separated and the peak shap e is very good and there are very less spurious peaks due to leaking of proteins in all different proteins during IEF. However there is very high amount of pH gradient compressi on observed in the cross channe l. The total length of cross channel is around 17 mm and since a pH gradient of 7 units is used (Carrier ampholytes of pH 310), the entire channel should have linear pH gradient starting with pH of 3 at the anode end and pH of 10 at the cathode end. In that case the minimum difference between the proteins BSA (pI 4.6) and carbonic anhydrase (pI 5. 9) should be 3.15 mm; otherwise th e proteins will be atleast 8 channels apart. The difference observed in actu al experiment is only around one channel gap or around 0.3 mm. This gives an approximate 10 times pH gradient compression.
89 The proteins thus compressed in IEF, got transferred in parall el channels. Another important aspect of IEF in linear polymer is th e presence of electro-osmotic flow. The presence of electro-osmotic flow from anode to cathode tends to shift the focusing action more towards the right side of the cross channel. The electr oosmotic flow is irregular due to highly random structure of gel on the interface of parallel channels. Some times presence of gel structure in cross channel and geometry enhances or reduces the electro osmotic flow. Figure 4-23 shows the similar focusing of four proteins (BSA, ovalbum in, hemoglobin and carbonic anhydrase) where the proteins are well distributed along the cross channel and got transferred in the parallel channels. Though hemoglobin has lower molecular weight that BSA, yet it traveled slower in the paralle l channels. The probable reason may the gel inconsistency between different channels. BSA and ovalbumin have similar pI (4.6) but different molecular weight. The map is not very clean as proteins seem to have leaked into different channels. 4.9 Conclusion Gel valve array for introduction of two type s of separation media in orthogonal channels for implementing two-dimensional protein separati on is studied in a microfluidic device. The valve arrays are fabricated by using photo-definable, in situ gel polymerization. The precise location is provided by the photomask and optical alignment. The locally-polymerized gels function as pseudo-valves, leading to a fluidic network that allows (1) the first dimension based on IEF and the second dimension based on PAGE; (2 ) full sample transfer from the first to second dimension; and (3) negligible di sturbance between two dimensions. Isoelectric focusing of two proteins, green fl uorescent protein and R-phycoerythrin, in the first dimension is demonstrated after gel valves are formed in the second-dimension channels. This result further showed the f unction of the gel valve array, ve rifying the overall concept of the
90 fluidic interface designed for two-dimensional pr otein separation in a microfluidic device. Different fluorescently labe led proteins are used apart from na turally fluorescent protein to study the viability of 2-D separation. The proteins are separated both on basis of charge (IEF) and molecular weight (SDS-PAGE). The separation base d on mass showed a linear correlation when molecular weight is compared with electrophor etic mobility. The data matches well with literature. The separation in first dimension (IE F) takes around 5 minutes at 100 V/cm whereas the second dimension takes around 2-8 minutes de pending electric field (120 V/cm-185 V/cm) and detection zone (distance aw ay from cross channel). Presence of SDS in the gel helps onchip protein complexation with SDS better. Table 4-1. Effects of gel thickness and mask width on phot o polymerization process. Mask width ( m) Gel thickness ( m) 56 66 68 25 54 68 75 50 52 61 66 80 Nonpolymerized gel width ( m) 47 47 52 Table 4-2. Comparison of resolution between different channel lengths and 2-D device Peak width GFP RPE Distance between peaks Resolution 2-D device (1.7 cm channel) 39 46 83 1.95 2 cm channel 28 53 86 2.12 3 cm channel 37 83 143 2.38 4 cm channel 41 105 193 2.64
91 Table 4-3. Five different labeled proteins used for 2-D separation No. Protein Isoelectric point Molecular weight (kDa) 1 Trypsin Inhibitor 4.6 20 2 Carbonic Anhydrase 5.9 31 3 Ovalbumin 4.6 45 4 Hemoglobin 7.1 65.5 5 Bovine serum albumin 4.6 66 Figure 4-1. Layout of a microfluidic device for tw o-dimensional protein sepa ration. It consists of one vertical channel (AoBo) and 29 horiz ontal channels (CD). An exploded view of the intersections of each device is illustra ted on the right. The size of the device is 1 x 3. A) The device with intersections of aligned CD channels. B) The device with intersections of staggered CD channels. A B C D A B Ao Bo
92 Figure 4-2. Photo polymeriza tion in microchannels. A) The photomask printed out in transparency. The minimum line thickness is 10 m, but while it is printed using commercial printer, minimum resolved line is 50 m. B) The schematic of experiment for gel polymerization between two glass slides. Figure 4-3. Polymerization time w ith acrylamide concentration. R2 = 0.973 0 5 10 15 20 25 30 35 40 23456789 % T AcrylamideTime (s) Gel UV Radiation Mask Glass slide Gel A B
93 Figure 4-4. Polymerization time with HCPK concentration. Figure 4-5. The schematic of photo polymerizatio n of acrylamide inside the microchannel. The channels are first filled with the monomer solution. The two slots at the extreme end of parallel channels are chemically polym erized, so that during UV polymerization, air does not enter inside the channels from open slots. Chemical polymerization is not required for the end wells of the cross cha nnel, since the liquid present inside the cross channel is sucked out after polymeri zation. The device is then flipped over and mask is carefully placed on top of the cro ss channel. Thereafter it is exposed to UV light for about 50 seconds. The mask is then removed, and the unpolymerized monomer solution is removed a nd replaced by the IEF solution. 0 5 10 15 20 25 30 35 2468101214 % HCPKTime (s) Chemical polymerization UV-light Mask Mask
94 Figure 4-6. Micrograph of the valve arrays formed by in situ gel polymerization. Polymerized gels are dyed for easy visualization. A) Fr om a device shown in Figure 4-1 A. B) An exploded view of the marked area in A. C) From a device shown in Figure 4-1 B. Figure 4-7. IEF in different microfluidic devi ces. A ) IEF in 5 cm channel at 500 V. B) IEF performed in cross channel AoBo of 2-D device, while the parallel channels are polymerized with polyacrylamide gel using sa me experimental condition as in A. in situ polymerized gels IEF medium Ao Bo A B in situ polymerized gels IEF medium C IEF of GFP-rpe @ 500 V, ch-4(50mm) 1.83%HPC 200 400 600 800 1000 1200 500100015002000 PixelIntensity (a.u ) 500 v 5 cm 100 m IEF of GFP+RPE in cross channel-200v 0 200 400 600 800 1000 1200 1400 1600 1800 134234334434534634734 PixelsIntensit y IEF in LPA B PAA Gel 2 cm Ao Bo GFP RPE GFP RPE
95 Figure 4-8. Isoelectric focusing of 4 proteins (Paralbumin, Ovalbumin, BSA, and GFP) in cross channel. A) Raw image. Experimental conditions: 300 V, 2.75% HPC in IEF medium, 10% T acrylamide polymer in parall el channels. Ovalbumin and BSA could not be separated because they have same is oelectric point. B) El ectropherogram of A. C) Plot of isoelectric point of proteins in B with their physical location in cross channel. The linear trend of the plot indica tes the uniform and linear pH gradient of IEF. R2 = 0.9983 4 4.5 5 5.5 6 6.5 4090140190 location along the channel (pixel)Isoelectric point 0 50 100 150 200 250 300 0100200 PixelsIntensity (a.u) Paralbumin Ovalbumin + BSA GFP A B C
96 Figure 4-9. 2-D device used for num erical simulation. A) The lengt h of parallel channels is 6.5 cm and that of cross channel is 1.7 cm. B) Close up view of cross channel with staggered parallel channel with quadril ateral mesh. The channel width is 100 m and distance between parallel channels is 360 m. 6 cm 0.5 cm 1.7 cm long A B
97 Figure 4-10. Transfer of negative species from cross channel to parallel channels due to an applied electric field of 100 V/cm. A) Numerical simula tion showing gradual change of concentration gradient with time. B) Same phenomenon observed in experiments, here the cross channel is filled up with fluorescent proteins whereas parallel channels are polymerized with Acrylamide. C) Possi ble electric field flow path between parallel channels and cross channel. This c onfiguration allows equa l splitting of cross channel analytes. D) Another possible conf iguration of electric field orientation which allows preferential splitt ing of cross channel analytes. A B C D
98 Figure 4-11. Transfer of protein plug s into parallel channel due to application of electric field. A) The plug is at center of tw o parallel channels and th e size of plug is around 400 m. The plug gets trifurcated into three parall el channels. B) The plug is before two parallel channels and gets bifurcated into two channels C) Plug is just above one channel and gets entirely transferred to that single channel. D) Similar to case B but the plug size is small and it still gets bi furcated in to channels. E) Experimental results for plug transfer from cross channel to parallel channel. Figure 4-12. Protein migration pattern in all 29 para llel channels. As the pr otein migrates through the gel the protein peak location varies in different channel due to gel inconsistency and variation of electric field. A B C D E Experiment
99 Figure 4-13. Electric field distri butions at different location ac ross the parallel channel. A) Different electrode configurat ion and position, B) plot of electric field across the entire 29 channels for all 3 cases meas ured halfway betwee n the electrodes. Electric field distribution870 880 890 900 910 920 930 940 950 960 051015202530 ChannelElectric potential (V) Flat electrode Electrode at center Electrode at corner A B Electrode at one corne r Flat electrode Electrode at cente r
100 Figure 4-14. Experimental results of location of protein front with respect to time. The trend is linear indicating constant speed for the pr otein fronts inside the parallel channels. R2 = 0.9992 50 100 150 200 250 300 350 400 450 0102030405060 Time (s)Location (Pixel) 0 2 4 6 8 10 020406 0 Time (s)Velocity (pixels/s)
101 Figure 4-15. SDS-protein complex mi gration pattern in Gel due to a pplication of electric field. A) Plot of Molecular weight of protei ns vs. electrophoretic mobility. They show linear trend in both 10% and 12% gel. The data is obtained from 10-15th channels from left end B) Schematic of experiment for SDS-PAGE separation of 4 proteins in two dimensional devices. Figure 4-16. IEF for RPE with and without SDS in 5 cm channel. The electric field is 500 V and pH gradient is 3-10. Gel+ Tris + Tricine Protein + 1%SD S Laser line R2 = 0.9376 R2 = 0.9922 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 00.0020.0040.0060.0080.010.012 Electrophoretic Mobility (mm2 V-1s-1) Log Molecular weigh t 12 % Gel 10 % Gel Linear (12 % Gel) Linear (10 % Gel) A B 0 500 1000 1500 2000 2500 3000 3500 4000 4500 02004006008001000120014001600 pixelsIntensity (a.u) 0.1 % SDS No SDS RPE without SDS RPE with 0.1 % SDS
102 Figure 4-17. The transfer of G FP and RPE in second dimension af ter IEF is performed in first dimension. Electric field of 800 V is appl ied in parallel cha nnels for separation second dimension. First dimension separati on (IEF) is done in 3-10 pH gradient at 200 V. Figure 4-18. 2-Dimensional migration of a single pr otein. A) IEF of BSA in cross channel. The pH gradient is 3-10. The voltage applied is 200 V. B) SDS-PAGE of BSA in parallel channels after IEF is performed. The detecti on zone is 6 mm from the cross channel. The voltage applied is 1200 V acr oss 6.5 cm parallel channels. GFP RPE SDS-PAGE SDS-PAGE-BSA0 200 400 600 800 1000 1200 02004006008001000120014001600 PixelsIntensity (a.u) BSA-IEF-cross channel200 250 300 350 400 450 500 020040060080010001200 PixelsIntensity (a.u) A B
103 Figure 4-19. 2-Dimensional separation of thr ee proteins. A) IEF of three proteins (BSA, hemoglobin and trypsin) in the cross channel. The pH gradient used is 3-10. BSA and trypsin has same isoelectric point of 4.6 wh ereas the Hemoglobins isoelectric point is 7.1. Hence two broad peaks are visible in IE F. B) The proteins are further separated according to their molecular weight in para llel channels. Trypsin being the smallest (20 kDa) travels the fastest followed by hemoglobin (65.5 kDa) and BSA (66 kDa). LANETIME (x 5 seconds) 5 10 15 20 25 5 10 15 20 25 30 35 40 45 50 55 Trypsin Inhibitor BSA Hemoglobin 100 200 300 400 500 600 700 800 10 20 30 40 50 60 70 80 90 100 110BSA, Trypsin Hemoglobin A B
104 Figure 4-20. 2-Dimensional separation of three proteins (BSA, Ovalbumin and Trypsin) in lower pH gradient (3-5 pH) during IEF operation. A) 2-D map. B) The correlation between molecular weight and electrophoretic mobility show a linear trend. C) 3-D plot of A showing the signal Intens ity of the proteins. ChannelsTime (s) 5 10 15 20 25 20 40 60 80 100 120 R2 = 0.9928 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 0.0020.0040.0060.0080.01Electrophoretic mobility (mm2 V-1 s-1)Log molecular weight A B C BSA Ovalbumin Trypsin Inhibitor
105 Figure 4-21. 2-Dimensional separations of three proteins (BSA, Ovalbumi n and Hemoglobin) in 3-10 pH during IEF operation. A) 2-D tim e map. B) The correlation between molecular weight and electrophoretic mobility show a linear trend. C) Time profile plot for all 29 channels as recorded by the detector. R2 = 0.9979 0 20000 40000 60000 80000 050100150 Time (s)Molecular weight (Da) Lanes Time (s) 5 10 15 20 25 20 40 60 80 100 120 3000 3200 3400 3600 3800 4000 4200 Trypsin Inhibito r Carbonic Anh y drase Hemoglobin 0 20 40 60 80 100 120 140 0 0.5 1 1.5 2 2.5 3 3.5 4 Hemoglobin CA Trypsin A B C
106 Figure 4-22. 2-Dimensional sepa ration proteins for pr oteins BSA (1), carbonic anhydrase (2), ovalbumin(3) and trypsin (4). The pH gradie nt is 3-10 for IEF. The second dimension is carried out with 185 v/cm and the protei ns got separated within 2 minutes.A) 2-D map. B) Correlation between molecular weight and time required to reach the detection point. C) The time profile plot for all 29 channels as recorded by the detector at 2 cm from cross channel. The data is normalized with maximum intensity value. R2 = 0.9347 0 25000 50000 75000 557595115 Time (s)Molecular weight 0 20 40 60 80 100 12 0 0 0.5 1 1.5 2 2.5 3 3.5 4 A B C ChannelsTime (s) 5 10 15 20 25 10 20 30 40 50 60 70 80 90 100 110 120 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 1 2 3 4
107 Figure 4-23. 2-D map for BSA, ovalbumin hemoglobin and carbonic anhydrase in 10% acrylamide gel. The experiment is done at lower electric field (120 V/cm ) and hence has taken more time for proteins to reach th e detection point at 6.5 mm from the cross channel. A) 2-D map B) The time correlati on with molecular wei ght. C) 3-D plot of A showing the signal Intensity of the proteins. LanesTime (s) 5 10 15 20 25 50 100 150 200 250 300 350 400 Carbonic Anhydrase Ovalbumin Hemoglobin BSA R2 = 0.97 20000 40000 60000 80000 150200250300 Time (s)Molecular weight (Da) A B C BSA Hemoglobin Ovalbumin CA
108 CHAPTER 5 CONCLUSION AND FUTURE DIRECTIONS 5.1 Conclusion The main objective of this research is to miniaturize the two el ectrophoretic separation mechanism namely isoelectric focusing and SDSPAGE and finally integr ate them in a single microfluidic device. A whole channel imaging de tection is assembled for detecting IEF. This system provides vital information about the dynami cs of protein separation both in spatial and temporal mode and eliminates the need for m obilization of protein pe aks after IEF, thereby preventing loss in resolution and time. IEF in microchannel proved to be quite successful in 6channel devices. Proteins with as low as 0.1 pH difference in their pI could be separated. Resolution of IEF was found to be almost independe nt of channel length wh en the electric field was kept constant. This allo ws for miniaturization of device and rapid analysis. One way of evaluating the performance of the separation system is to measure the peak capacity of the device. Peak capacity is a nondime nsional quantity and is defined as a ratio of channel length and protein peak width. Since the peak capacity of a 2-dimensional separation system is the product of peak cap acities of an individual separati on in each dimension, the total peak capacity is higher and the system can be used for performing complex protein separation. The peak capacity of IEF in 6 channel device is around 125 (5 cm channel at 100 V/cm). The peak capacity of IEF in the cross channel in 2-D device, however, came down to around 90. The SDS-PAGE peak capacity in paralle l channels is 68. Combined resu lts are better than reported in literature49, 109. These peak capacities, however, depend on the peak width of the protein chosen. Different proteins have different peak width a nd the resultant peak capacity will vary. The most important component of 2-D separation is th e presence of gel in the second dimension
109 separation. The gel effectively prev ents proteins from diffusing in to parallel channel before IEF is performed. The resultant 2-D se paration is clean and precise. 5.2 Future Direction There are still some issues of small amount of proteins leak ing into extreme right parallel channels when an electric field is applied for IEF. This is mainly due to curved electric field in the junction of parallel channels with cross ch annel. This can be reduced by using a lower electric field at the beginning of experiment (IEF) and increasi ng the field when proteins are almost focused. Also redesigning the device such that higher concen tration of gel is present near the cross channel which will further reduce the di ffusion. Figure 5-1 shows a possible method to achieve the goal in the present design. The mask si ze is increased such that after polymerization, at least 500 m on both the sides of the cross channel is not polymerized. The contents of the cross channel are then sucked out and higher co ncentration of monomer solution is pumped in. Since it is impossible to suck out the unpolymeri zed monomer solution in parallel channel, the only way to increase the con centration is by diffusion. The new monomer solution with a higher con centration in the cross channel is then allowed to diffuse into the non-polymerised zone of parallel channel. Now the second polymerization is carried out with a mask with smaller trace width. The previously unpolymerized region in the parallel channels will now be polymerized. The device geometry can also be optimized by studying the diffusion phenomenon for different geometries in CFDACE. This will prevent costly and ti me consuming manufacturing steps. In addition, the cross channel in the 2-D device is quite sm all in terms of length. The reason behind decreasing the length is to re duce separation time, sma ller footprint and lower electric field requirements. But as the channel length is reduced electrosmotic and hydrostatic
110 flow starts playing a bigger role in the separa tion mechanism. Though it is difficult to accurately predict the right length of the cross channel for IEF, estimation can be done based on previous experimental results. Assuming the target separation resolution to be 0.1 pH for each parallel channels, total length required for resolving the full pH gradient (3-10 fo r our case) is 25.2 mm. But it is seen from experiments that both in 6 channel device and 2-D devices, the IEF has suffered extreme pH gradient compression (up to 60%). The length of cross channel should be increased to around 6.3 cm to take care of the pH gradient compression. The hydrodynamic flow is also of important co ncern in cross channel. This hydrodynamic flow will drive the focused proteins out from th e cross channel. The continuous flow will distort the IEF and will move the proteins from its isoelectric point. This flow usually happens due to unbalanced catholyte and anolytes in the reserv oir. Assuming a variation of around 10% in pipetting the liquid, 2 mm diameter reservoir will have a resulting variatio n of 0.2 mm in height. This will in turn l ead to a flow of 5 m/s in cross channel (1.5 cm) from higher pressure to lower pressure region. IEF in first dimension is achie ved in 180 seconds. This will lead to shift of peaks by about 900 m from its original isoelectric poin t. Increasing the channel length will reduce this speed ( L V1) and hence reduce the peak shift. Now taking into account the pH gradient compression a cross channel of 6.3 cm will also reduce this peak shift variation due hydrostatic imbalance by more than 75% (1.2 m/s). As a result the channel length for IEF is suggest ed to increase to about 6.5 cm in order to take care of these problems. Th e main advantages of miniaturiz ing 2-D separation is the shorter time required to complete the operations. For a fl at bed gel system it takes almost 10-12 hours to finish IEF and another 1 hr to fi nish SDS-PAGE. The same operati on can be finished in less than
111 10 minutes. This is a huge improvement over the conventional system and at the same time it reduces the space requirement and labor intensive operations. Figure 5-1. The mask size is increased such that 500 m on both the sides of cross channel is left unpolymerized after UV exposure. The monomer solution from cross channel is replaced with another one with higher c oncentration. This higher concentration monomer solution will diffuse in to stil l unpolymerised part of parallel channel. Second exposure to UV with thinner mask w ill create a layer of higher concentration gel on both the sides of cross channel. New gel A B
112 APPENDIX TRANSIENT SOLUTION OF IEF Standard transient electrophoret ic process can be defined as43, 78, 92, 110 2 2) ( ) ( ) ( ) ( x t x C D x x u t x C t t x C [A-1] The transient state of the Equation [A-1] provi des information about time and distance. It will define the velocity profile of macromolecule s in pH gradient. The macromolecules here are assumed as spherical and non-spinning in nature. For a charged spherical particle moving in an electric field, the force balance yields field electric to due Force Drag Viscous [A-2] Then from Stokes result for drag on solid sphere. r x u Drag ) ( 6 [A-3] where is the viscosity, r is the radius of the sphere and u(x) is the velocity of sphere. The force on the particles (macromolecules) du e to external electric field is E x q ) (, where ) ( x q is the charge on the sphere and E is the electric field. Hence E x q x u r ) ( ) ( 6 [A-4] So the velocity of the particle is directly dependent on charge of particle in the solution. E r x q x u 6 ) ( ) ( [A-5] From Equation  of Chapter 1, the particle velocity is the product of ep and electric field. Comparing with Equation [A-5 ], mobility can be written as r x q 6 ) ( [A-6]
113 The charge distribution in IEF is non-linear with respect to x. However the nonlinearity exists near the end of the channels. This non linear charge will give rise nonlinear distribution of mobility at different pH. The effective mobility for ionic species can be described as111, 112 2 2 11 c pK pH effae [A-7] where 1 and 2 are the effective mobilities at low a nd high pH values respectively; c is a constant and pKa is the dissociation constant. Since in IEF it is assumed that a pH gradient with respect to distance is established. So replacing pH by distance x in Equation [A-7], we have. 2 2 11 c pK x effae [A-8] So the mobility is now a function of distance. The charge distribution can be seen from Figure A-1 Equation [A-1] can be solved as a moving gaussian pulse in x-direction96, 97. ) ( 2 ) ( ) ( exp ) ( ,2 2t s t f x u x t A t x C [A-9] where A(t) is the amplitude, s(t) is the half width of the pulse and [()()] uxft will give the position of center of mass of gaussian pulse. At t=0, the analyte is assumed to have an in itial concentration as a Gaussian distribution. ) ( 2 exp 2 0 ,2 2 0t x c x C [A-10] So comparing the initial conditions 2 00c A ; 0 s ; 0 0 f [A-11]
114 Now in Equation [A-1], c(x,t) and u(x) can be used to find out the constants A(t) s(t) and f(t) It is a lengthy procedure and the final form of s(t) and f(t) can be found out as: dx du t dx du t f exp 1 [A-12] 1 2 exp2 2t dx du dx du D t s [A-13] where c pK x c pK x c E dx dua aexp exp 12 1 2 [A-14] The functions found above is very complicated, with f(t) and s(t) having dependency on both x and t The charge distribution, if carefully analyzed can be assumed to be essentially linear with non-linearity near the two extreme ends. This linear assumption makes the functions much simpler to evaluate. Figure A-2 shows such linear dependency of charge on distance. So for the linear case the mobility can be assumed as:97 r q x x xpI pI 60 [A-15] where xpI is the isoelectric point of that particle. pI pIx q r E x x x u06 ) ( ; [A-16] Denoting E r q u60 0 [A-17] u0 is the initial velocity due to an applied field E and initial charge is q0
115 Equation [A-16] can be rewritten as pI pIx u x x x u0) ( [A-18] Assigning a typical time scale fo r electrophoretic process as t0 0 0u x tpI [Assuming constant charge, t0 is the time required to reach xpI] [A-19] Solving the Equation [A-1] as before using Equati on [A-9], functions A(t), f(t) and s(t) can be found out as: 0 01 1 ) exp( ) ( t t t t f [A-20] 2 ) ( 1 ) ( t s t A [A-21] D t t t Dt t s0 2 0 01 2 exp 1 ) ( [A-22] where is the initial gaussian distri bution of the protein peaks. So the final solution of concentration with respect to time and distance is: 2 2() 1 ,exp 2() ()2 xxt Cxt st st [A-23] 0()1exppI x txtt [A-23] Here () x t is the location of center of mass of th e Gaussian pulse. The concentration plotted with respect to time and dist ance according to the Equation [A -23] is shown in Figure A-3 below.
116 Figure A-3 shows the IEF process. The x-axis is distance. The concentration is plotted at different times. The peak gradually sharpens as it approaches its pI point. The process asymptotically approaches towards completion. Equation [A-23] can be analy zed for different conditions At time t = 0, () st 0 ) ( t x, the center of mass for the CA/protein is at x=0 2 22 exp 2 1 0 x x C [A-25] At time t 0) ( Dt t g pIx t x ) ( So at very long time the CA/protein peaks center of mass will be exactly at their pI point. 0 2 02 exp 2 1 Dt x x Dt x CpI [A-26] Equation [A-24] has the spatia l location of center of mass of protein peak. Differentiating with respect to time will give the front veloc ity. The front speed thus can be defined as: 0 0exp ) ( t t t x t vpI [A-27] This front speed can be monitored in real time by the imaging system developed for whole channel viewing and the numerical results can be experimentally verified. As seen from the Figure A-4, higher field gi ves a higher starting speed in focu sing, but all of them approaches almost similar speed at about 1200s (20 minutes) from starting time. So higher voltage will ensure faster focusing and will have faster reduction of speed near its pI point. As the CA/protein s are gradually getting focused, the conductivity of the system drops. This drop in conductivity however may be used to predict the degree of completion, thereby eliminating need for c ontinuous monitoring of system dynamics by a CCD camera, if
117 final focused protein is the only requirement. So a theoretical development of conductivity model is required. The conductivity of the system can be found out in the following way. For a single ampholyte / protein species, the cu rrent density can be written as 0 ) ( ) ( x t x J t t x C [A-28] where j is the current density. For large number of proteins, the to tal concentration can be written asi i x x AllC C where i = 1,2,3 Introducing Equation [A-1] in 28, x C D x u t x C t x j ) ( ) ( ) ( [A-29] This can be analytically solved as ) ( ) ( 1 ) ( ) (2 0t s t x x D x x u t x C t x jpI [A-30] Since conductivity c is defined as current density per unit voltage. So c j E conductivity can be easily found out by di viding Equation [A-30] by E Now this conductivity is a function of both distan ce and time. In common IEF setup, usually the total voltage drop across the ch annel is measured. The conducti vity thus measured in our experiment is only a function of time. Hence inte gration of conductivity w ith respect to distance x leads to a function which is only a function of time. 1 01 () (,)L c cE tdx Axt [A-31] There is no closed formed analytical soluti on for this integration and hence has to be integrated numerically. This conductivity valu e however is only for a particular protein (macromolecule). Hence for all different proteins taken together the to tal conductivity can be
118 assumed to have very similar characteristic s with time though the absolute magnitude of conductivity will change. The Figure A-5 shows the relationship of conduc tivity with applied el ectric field according to Equation [A-30]. As the field is increased the CA are more focused and hen ce the conductivity drop is more for higher voltage. As the CAs are focused faster, IEF time will be reduced. Since the IEF process asymptotically approach es its completion, there must be a cut-off point say 95% of completion to define the time limit. Since from Equation [A-24], the position of the center of mass (C.M) of ampholyte / protein is known, it can be assumed the process is completed when front C.M is within 95% of it pI point. pI pIx t t x 95 0 ) exp( 10 [A-32] 3 ) 20 ln(0 t t 03 u x tpI [A-33] But from Equation [A-17], E u 0, so from Equation [A-33], E t 1 [A-34] Hence the time of completion is i nversely proportional to applied fi eld strength. So if the field strength is increased, the process can be achie ved faster, but the fiel d strength can not be increased indefinitely as it will have adverse e ffect of Joule heating and also increased electroosmotic flow.
119 Figure A-1. Assumed char ge distribution for different CA/ pr oteins in IEF focusing process for nonlinear case. 0 0.5 1 1.5 2 2.5 3 -6 -4 -2 0 2 4 6 DistanceCharge Figure A-2. Assumed char ge distribution for different CA/ proteins in IEF focusing process 0 1 2 3 4 5 6 7 8 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 DistanceCharge pI points
120 -0.5 0 0.5 1 1.5 2 0 2 4 6 8 10 12 Distance (cm) Concentration (a.u) Figure A-3. Evolution of Gaussian peak with time for a particular ampholyte. As time progresses, the peak gets narrower and slow s down and stops at pI point, exactly as observed in experiment. Using r ealistic diffusion constant D=10-7 cm2/s, E=500 V, u0=3x10-4cm/s, the protein front evolution is pl otted with time at different applied field.
121 0 50 100 150 200 250 300 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01 Time (s)Speed (cm/s) 1000 v 900 v 800 v 700 v 600 v Figure A-4. Numerical results for front speed of CA/ proteins with time at different. applied electric field. The front speed is higher at higher electric field strength and tends to reduce speed faster. Faster reduction of font speed indicates faster approach towards its isoelectric point and hen ce faster process completion.
122 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time (s)Conductivity 1000 v 900 v 800 v 700 v 600 v Figure A-5. Conductivity plotted wi th respect to time at different electric fields. The results shown is normalized. The length of the channe l is 5 cm. The diffusion constant is 1e7 cm2/s. As the electric field is increased, the initial current density is high, but drops faster with time.
123 LIST OF REFERENCES (1) Yazdi, N. A., F.; Najafi, K. Proceeding of the IEEE 1998, 86 1640-1659. (2) Schwarz, M. H., R.; Hosticka, B.J.; Huppertz, J., Kneip, T.; Kolnsberg, S.; Ewe, L.;Trieu, H.K.; Ieee Trans. Circuits sys. II 1999, 46 870-877. (3) De Gasperis, G. Y., Jun; Becker, Freder ick F.; Gascoyne, Peter R. C.; Wang, Xiao-Bo Biomedical Microdevices 1999, 2(1) 41-49. (4) Wang, X.-B. Y., Jun; Huang, Ying; V ykoukal, Jody; Becker, Frederick F.; Gascoyne, Peter R. C. Analytical Chemistry 2000, 72(4) 832-839. (5) Barrie, G. P., R.; Kaler, K. V. I. S.; Jones, T. B. Journal of Colloid and Interface Science 1995, 175(1) 97-107. (6) Giddings, J. C. Science 1993, 260 1456-1465. (7) Wang, M. M. T., Eugene; Raymond, Da niel E.; Yang, Joon Mo; Zhang, Haichuan; Hagen, Norbert; Dees, Bob; Mercer, Eli nore M.; Forster, Anita H.; Kariv, Ilona; Marchand, Philippe J.; Butler, William F. Nature Biotechnology 2005, 23(1) 83-87. (8) Yamada, M. S., Minoru Lab on a Chip 2005, 5(11) 1233-1239. (9) Terry, S. C.; Jerman, J. H.; Angell, J. B. Ieee Transactions on Electron Devices 1979, 26 1880-1886. (10) Shoji, S. S. B., V.; de Rooij, N.; Esashi, M.; Tech. Digest IEEE Transducers (San Fransisco, CA, 1991) 1991, 1052-1055. (11) Kim, J.-H. N., K-H.; Kang, C.J.; Jeon, D.; Kim, Y-S.; Microelectronic engineering 2004, 73-74 864-869. (12) Nguyen, N. T. S., S.; Richter, S.; Dotzel, W. Sensors and Actuators, A: Physical 1998, A69(1) 85-91. (13) Xie, J. S., Jason; Lin, Qiao; Yang, Bozhi; Tai, Yu-Chong Lab on a Chip 2004, 4(5) 495501. (14) Jackson, W. C. T., Hy D.; O'Brien, Mi chael J.; Rabinovich, Emmanuil; Lopez, Gabriel P. ournal of Vacuum Science & Technology, B: Microelectronics and Nanometer Structures 2001, 19(2) 596-599. (15) Yamahata, C. L., Frederic; Gijs, Martin A. M. Microelectronic Engineering 2005, 78-79 132-137.
124 (16) Jerman, J. H. Tech. Digest IEEE Transducers (San Fransisco, CA, 1991) 1991, 10451048. (17) Darhuber, A. A. V. J. P. T., S.M.; Wagner, S. Journal of microelect romechanical systems 2003, 12 873-879. (18) Stone, H. A. S., A.D.; Ajdari, A. Annu. Rev. Fluid Mech. 2004, 36 381-411. (19) Wang, P.-J. C., Chia -Yuan; Chang, Ming-Lang Biosensors & Bioelectronics 2004, 20(1) 115-121. (20) Olsson, A. S., Goeran; Stemme, Erik Sensors and Actuators, A: Physical 1996, A57(2) 137-143. (21) Olsson, A. S., Goeran; Stemme, Erik Sensors and Actuators, A: Physical 1995, A47(1-3) 549-556. (22) Sethu, P. M., C. H. Sensors and Actuators, A: Physical 2003, A104(3) 283-289. (23) Fuhr, G. H., R.; Muller, T.; Benecke, W.; Wagner, B.; Proc. MEMS 92 1992, 25-30. (24) Yao, S. H., D.E.; Mikkelsen, J.C.; Santiago, J.G. Journal of Colloid and Interface Science 2003, 268 143-153. (25) Sharp, K. V. A., R.J.; Santiago, J.G.; Molho, J.I. The MEMS handbook ; CRC Press, 2002. (26) Manz, A. G., N.; Widmer, H. M. Sensors and Actuators, B: Chemical 1990, B1(1-6) 244-248. (27) Dunn, M. J. Gel electrophoresis of proteins ; IOP Publishing Limited: Bristol, 1986. (28) Righetti, P. G.; Drysdale, J. W. Isoelectric Focusing First ed.; North-Holland Publishing: Amsterdam, 1976. (29) Westermeier, R. Electrophoresis in Practice ; VCH Publishers: Weinheim, Germany, 1993. (30) Mosher, R. A.; Thormann, W. Electrophoresis 2002, 23 1803-1814. (31) Saravis, C. A.; O'Brien, M.; Zamcheck, N. J Immunol Methods 1979, 29 97-100. (32) Saravis, C. A.; Zamcheck, N. J Immunol Methods 1979, 29 91-96. (33) Rosen, A.; Ek, K.; Aman, P. J Immunol Methods 1979, 28 1-11. (34) Righetti, P. G.; Macelloni, C. J Biochem Biophys Methods 1982, 6 1-15.
125 (35) Bjellqvist, B.; Ek, K.; Righetti, P. G.; Gianazza, E.; Gorg, A.; Westermeier, R.; Postel, W. J Biochem Biophys Methods 1982, 6 317-339. (36) Righetti, P. G.; Bossi, A. Anal Biochem 1997, 247 1-10. (37) O'Farrell, P. H. J Biol Chem 1975, 250 4007-4021. (38) Harrison, D. J. F., Karl; Seiler, Ku rt; Fan, Zhonghui; Effenhauser, Carlo S.; Manz, Andreas Science 1993, 261(5123) 895-897. (39) Fan, Z. H. H., D. J.; Analytical Chemistry 1994, 66 177-184. (40) McDonald, J. C.; Metallo, S. J.; Whitesides, G. M. Anal Chem 2001, 73 5645-5650. (41) Duffy, D. C. M., J. Cooper; Schuelle r, Olivier J. A.; Whitesides, George M. Analytical Chemistry 1998, 70(23) 4974-4984. (42) Boone, T. D.; Fan, Z. H.; Hooper, H. H.; Ricco, A. J.; Tan, H.; Williams, S. J. Anal Chem 2002, 74 78A-86A. (43) Tan, W.; Fan, Z. H.; Qiu, C. X.; Ricco, A. J.; Gibbons, I. Electrophoresis 2002, 23 36383645. (44) Hofmann, O.; Che, D.; Cruickshank, K. A.; Muller, U. R. Anal Chem 1999, 71 678-686. (45) Liu, Z.; Pawliszyn, J. J Proteome Res 2004, 3 567-571. (46) Duong, T. T.; Kim, G.; Ros, R.; Streek, M. ; Schmid, F.; Brugger, J.; Anselmetti, D.; Ros, A. Microelectronic Engineering 2003, 67-8 905-912. (47) Hatch, A. V. H., A.E.; Throckmorton, D.J.; Brennan, J.S.; Singh, A.S.; Anal Chem 2006, 78 4976-4984. (48) Chen, X.; Wu, H.; Mao, C.; Whitesides, G. M. Anal Chem 2002, 74 1772-1778. (49) Li, Y.; Buch, J. S.; Rosenberger, F.; DeVoe, D. L.; Lee, C. S. Anal Chem 2004, 76 742748. (50) Baker, D. R. Capillary Electrophoresis ; John Wiley and Sons Inc: New York: NY, 1995. (51) Crimaldi, J. P. Experiments in Fluids 1997, 23 325-330. (52) Eggeling, C. J. W., R. Rigler, C.A. M. Seidel Analytical Chemistry 1998, 70 2651. (53) Pettit, G. H. E., M.N.; Hahn, D.W.; Brinson, B.E.; Saurbrey, R.; Applied Physics A 1994, 58 573-579.
126 (54) Lukyanchuk, B. B., N.; Anisimov, S.; Arnld, N.; Bauerle, A.D. Applied Physics A 1996, 62 397-401. (55) Richards, D. P. S., C.; Polakowski, R.; Ahmadzadeh, H.; Dovichi, N. J. Journal of Chromatography., A 1999, 853 21-25. (56) Vogt, O.; Pfister, M.; Marggraf, U.; Neyer, A.; Hergenroder, R.; Jacob, P. Lab Chip 2005, 5 205-211. (57) McCormick, R. M.; Nelson, R. J.; Alonso-Amigo, M. G.; Benvegnu, D. J.; Hooper, H. H. Anal Chem 1997, 69 2626-2630. (58) Han, A.; Wang, O.; Graff, M.; Mohanty, S. K.; Edwards, T. L.; Han, K. H.; Bruno Frazier, A. Lab Chip 2003, 3 150-157. (59) Tan, W.; Fan, Z. H.; Qiu, C. X.; Ricco, A. J.; Gibbons, I. Electrophoresis 2002, 23 36383645. (60) Pugmire, D. L.; Waddell, E. A.; Haasch, R.; Tarlov, M. J.; Locascio, L. E. Anal Chem 2002, 74 871-878. (61) Lai, S.; Wang, S. N.; Luo, J.; Lee, L. J.; Yang, S. T.; Madou, M. J. Analytical Chemistry 2004, 76 1832-1837. (62) Juang, Y. J.; Lee, L. J.; Koelling, K. W. Polymer Engineering and Science 2002, 42 539550. (63) Liu, R. H.; Yang, J. N.; Lenigk, R.; Bonanno, J.; Grodzinski, P. Analytical Chemistry 2004, 76 1824-1831. (64) Li, Y.; Buch, J. S.; Rosenberger, F.; DeVoe, D. L.; Lee, C. S. Analytical Chemistry 2004, 76 742-748. (65) Xu, Y.; Vaidya, B.; Patel, A. B.; Ford, S. M.; McCarley, R. L.; Soper, S. A. Anal Chem 2003, 75 2975-2984. (66) Koh, C. G.; Tan, W.; Zhao, M. Q.; Ricco, A. J.; Fan, Z. H. Anal Chem 2003, 75 45914598. (67) Kameoka, J.; Craighead, H. G.; Zhang, H. W.; Henion, J. Analytical Chemistry 2001, 73 1935-1941. (68) Mela, P.; van den Berg, A.; Fintschenko, Y.; Cummings, E. B.; Simmons, B. A.; Kirby, B. J. Electrophoresis 2005, 26 1792-1799. (69) Kim, D. S.; Lee, S. H.; Kwon, T. H.; Ahn, C. H. Lab on a Chip 2005, 5 739-747.
127 (70) Esch, M. B.; Kapur, S.; Irizarry, G.; Genova, V. Lab on a Chip 2003, 3 121-127. (71) Rotting, O.; Ropke, W.; Becker, H.; Gartner, C. Microsystem Technologies 2002, 8 32 36. (72) McDonald, J. C.; Whitesides, G. M. Acc Chem Res 2002, 35 491-499. (73) Soper, S. A.; Ford, S. M.; Qi, S.; McCarley, R. L.; Kelly, K.; Murphy, M. C. Analytical Chemistry 2000, 72 642A-651A. (74) Becker, H.; Gartner, C. Electrophoresis 2000, 21 12-26. (75) Fredrickson, C., University of Florida, Gainesville, 2005. (76) Hofmann, O.; Che, D. P.; Cr uickshank, K. A.; Muller, U. R. Analytical Chemistry 1999, 71 678-686. (77) Macounova, K.; Cabrera, C. R.; Holl, M. R.; Yager, P. Analytical Chemistry 2000, 72 3745-3751. (78) Li, C.; Yang, Y.; Craighead, H. G.; Lee, K. H. Electrophoresis 2005, 26 1800-1806. (79) Cui, H.; Horiuchi, K.; Dutta, P.; Ivory, C. F. Anal Chem 2005, 77 1303-1309. (80) Xu, Y.; Zhang, C. X.; Janasek, D.; Manz, A. Lab on a Chip 2003, 3 224-227. (81) Herr, A. E.; Molho, J. I.; Drouvalakis, K. A. ; Mikkelsen, J. C.; Utz, P. J.; Santiago, J. G.; Kenny, T. W. Analytical Chemistry 2003, 75 1180-1187. (82) Raisi, F.; Belgrader, P.; Borkholder, D. A.; Herr, A. E.; Kintz, G. J.; Pourhamadi, F.; Taylor, M. T.; Northrup, M. A. Electrophoresis 2001, 22 2291-2295. (83) Wang, Y. C.; Choi, M. H.; Han, J. Anal Chem 2004, 76 4426-4431. (84) Chen, X. X.; Wu, H. K.; Ma o, C. D.; Whitesides, G. M. Analytical Chemistry 2002, 74 1772-1778. (85) Griebel, A.; Rund, S.; Schonfeld, F.; Dorner, W.; Konrad, R.; Hardt, S. Lab on a Chip 2004, 4 18-23. (86) Lee, H. G. J Chromatogr A 1997, 790 215-223. (87) Tian, H. J.; Landers, J. P. Analytical Biochemistry 2002, 309 212-223. (88) Shen, Y. F.; Smith, R. D. Journal of Microcolumn Separations 2000, 12 135-141.
128 (89) Li, Y.; DeVoe, D. L.; Lee, C. S. Electrophoresis 2003, 24 193-199. (90) Harrison, D. J.; Fluri, K.; Seiler, K. ; Fan, Z. H.; Effenhauser, C. S.; Manz, A. Science 1993, 261 895-897. (91) Fredrickson, C. K.; Fan, Z. H. Lab Chip 2004, 4 526-533. (92) Shimao Electrophoresis 1986, 7 121-128. (93) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: F undamentals and Applications ; John Wiley & Sons: New York, 1980. (94) Liu, Z. L., T.;Pawliszyn Journal of Proteome Research 2006, 5 1246-1251. (95) Svensson, H. Acta Chemica Scandinavica 1961, 15 325-341. (96) Zilberstein, G. V. B., Emmanuil M.; Bukshpan, Shmuel. Electrophoresis 2003, 24(21) 3735-3744. (97) Frumin, L. L. Z., G. V.; Peltek, S. E. Journal of Biochemica l and Biophysical Methods 2000, 45(2) 205-209. (98) Herr, A. E.; Molho, J. I.; Santiago, J. G.; Kenny, T. W.; Borkholder, D. A.; Kintz, G. J.; Belgrader, P.; Northrup, M. A., Hilton Head, SC 2000; 115-119. (99) Richards, D. P.; Stathakis, C.; Polako wski, R.; Ahmadzadeh, H.; Dovichi, N. J. Journal of Chromatography A 1999, 853 21-25. (100) Das, C.; Xia, Z.; Stoyanov, A.; Fan, Z. H. Instrumentation Science and Technology 2005, 33 379-389. (101) Baker, D. R. Capillary Electrophoresis ; John Wiley and Sons Inc: New York, NY, 1995. (102) Snyder, L. R.; Kirkland, J. J. Introduction to modern liquid chromatography 2d ed.; Wiley: New York, 1979. (103) Yang, C.; Zhu, G.; Zhang, L.; Zhang, W.; Zhang, Y. Electrophoresis 2004, 25 1729 1734. (104) Gao, L.; Liu, S. Anal Chem 2004, 76 7179-7186. (105) Lu, J. J.; Liu, S.; Pu, Q. J Proteome Res 2005, 4 1012-1016. (106) Das, C. F., Z.H.; Electrophoresis 2006, 27 3619-3626. (107) Viney, C. F., R.A. Eur. J. Phys. 1998, 19 575-580.
129 (108) Ewing, A. G. W., R.A.; Olefirowicz, T.M.; Analytical Chemistry 1989, 61 292-303. (109) Herr, A. E.; Molho, J. I.; Drouvalakis, K. A. ; Mikkelsen, J. C.; Utz, P. J.; Santiago, J. G.; Kenny, T. W. Anal Chem 2003, 75 1180-1187. (110) Macounova, K.; Cabrera, C. R.; Holl, M. R.; Yager, P. Anal Chem 2000, 72 3745-3751. (111) Poole, S. K.; Patel, S.; Dehri ng, K.; Workman, H.; Poole, C. F. J Chromatogr A 2004, 1037 445-454. (112) Chen, X. Z., J.; Zhai, H.; Chen, X; Hu, Z.; Food Chemistry 2005, 92 381-386.
130 BIOGRAPHICAL SKETCH Champak Das is currently pursuing his PhD in mechanical engineering from the University of FloridaGainesville. Prior to joining University of Florida in 2003 he finished his M.S in mechanical engineering from Florida Institute of Technology, Melbourne in 2003 and B.S in mechanical engineering from Ja davpur UniversityCalcutta, I ndia in 1998.After his B.S he joined Eveready Industries India Ltd. as a Develo pment engineer and worked there for 3 years. His current research activities involve microfluid ics, microscale electrophoresis and laser based diagnostics.