The Pharmaceutical stability and formulation of plasmid DNA

MISSING IMAGE

Material Information

Title:
The Pharmaceutical stability and formulation of plasmid DNA
Physical Description:
xii, 116 leaves : ill. ; 29 cm.
Language:
English
Creator:
Poxon, Scott William, 1971-
Publication Date:

Subjects

Subjects / Keywords:
Research   ( mesh )
Plasmids -- chemistry   ( mesh )
DNA -- chemistry   ( mesh )
Drug Stability   ( mesh )
Transfection   ( mesh )
Endotoxins   ( mesh )
Department of Pharmaceutics thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Pharmacy -- Department of Pharmaceutics -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1999.
Bibliography:
Bibliography: leaves 107-114.
Statement of Responsibility:
by Scott William Poxon.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030253967
oclc - 51638680
System ID:
AA00022318:00001

Full Text












THE PHARMACEUTICAL STABILITY AND FORMULATION OF PLASMID DNA









By



SCOTT WILLIAM POXON


















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


UNIVERSITY OF FLORIDA



1999








































This dissertation is dedicated to my wife,

Stephanie Poxon, without whose understanding and

patience I could not have accomplished this task. As

well, I would like to thank my parents and the other

students in the Hughes laboratory for their help and

support.



















ACKNOWLEDGMENTS

I would like to acknowledge my committee members,

Dr. Jeffrey Hughes, Dr. Gayle Brazeau, Dr. Edwin

Meyer, Dr. Michael Schwartz, and Dr. Ian Tebbett for

their critical reading of this manuscript, support and

contribution to this project. The Parenteral Drug

Association Foundation Biotechnology Grant, the

National Institutes of Health (NIH P01-AG 10485), and

the University of Florida Gene Therapy Center also

financially supported these studies.






















iii
















TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ........................................... iii

LIST OF FIGURES ................................... ...... vi

ABSTRACT ................................................. xi

1 INTRODUCTION ............................................ 1

Specific Aims and Hypotheses............................ 1
Background and Significance............................. 2
Chemical and Physical Stability of DNA.................. 5
Background ........................................... 5
Glycosidic Bond ...................................... 7
Phosphodiester Bond ......... ......... ........ ...... 9
Steric Effects .......................... ............. 10
Oxidative Damage ....................................... 10
Lyophilization of DNA................................. 11
Endotoxin Contamination of DNA......................... 13
Conclusions ............................................15

2 PHYSICAL AND CHEMICAL STABILITY OF PLASMID DNA ......... 16

Background ...... .......................... ........... 16
Materials and Methods..................................21
Plasmid Purification ............................... 21
Liposome Preparation ............................... 22
In vitro Coupled Transcription-Translation ..........23
Transfection Efficiency Assay ......................23
DMED Assay .......................................... 25
H Solution Stability Study ......................... 25
Aemperature Stability Study ........................26
Statistical Analysis ................................ 27
Results and Discussion................................. 27

3 FORMULATION OF PLASMID DNA: THE EFFECT OF
LYOPHILIZATION ON PLASMID DNA STABILITY................ 34

Introduction. .............................. ............ 34

iv









Materials and Methods.................................. 38
Plasmid Purification ................................ 38
Liposome Preparation .............................. 39
Transfection Efficiency Assay ...................... 40
Differential Scanning Calorimetry Analysis .......... 41
DMED Assay ..................... ... ......... ............... 42
Analysis of Hyperchromic Effect ............. ....... 42
Oxidative Analysis ...................... .. ....... 43
Circular Dichroism Analysis ........................ 44
Statistical Analysis ................................ 44
Results and Discussion................................. 44
Conclusion............................................. 61

4 CHARACTERIZATION OF ENDOTOXIN AND CATIONIC LIPOSOME
INTERACTION ........................................... 64

Introduction ........................................... 64
Materials and Methods.................................. 67
Plasmid Purification ............................... 67
Liposome Preparation ................................ 68
Transfection Efficiency Assay ...................... 68
Anisotropy Assay ............................... .. .. 70
Endotoxin Dephosphorylation ........................ 71
Endotoxin Assay .......................... .. ......... 71
Cell Viability Assay ................................ 72
Statistical Analysis ................................ 73
Results................................................ 73
Discussion ............................................. 81

5 FOAM FRACTIONATION AS A METHOD TO SEPARATE ENDOTOXIN
FROM RECOMBINANT BIOTECHNOLOGY PRODUCTS................ 88

Introduction............................................. 88
Methods ................................................ 91
Plasmid Purification ............................. ... 91
Foam Fractionation .................................. 92
Plasmid and FITC-Endotoxin Gel Analysis .............92
Surface Activity Determination ...................... 93
Particle Size Analysis ............................. 93
Endotoxin Assay ...................... .... .......... 93
Protein Analysis ..................................... 94
Statistical Analysis ........................... ....... 94
Results and Discussion...... .............. ............ 94
Conclusions ................................ ............. 103

LIST OF REFERENCES ...................................... 107

BIOGRAPHICAL SKETCH .................................... 115

v











LIST OF FIGURES


Figure page

1. Major sites for chemical degradation of DNA.
Filled arrows point to potential oxidative damage
sites, while open arrows point to potential
hydrolytic damage sites. ..............................6

2. Glycosidic cleavage of cytosine to an aldehyde
under alkaline or acidic conditions followed by
beta elimination to cleave the phosphate
backbone. .......................... ................... 7

3. Comparison of cotransfection (A) and single
plasmid (B) concentration vs activity. Mean +
SEM, n=4. .............................................28

4. Effect of incubation at elevated temperature for
3 weeks on the functional activity of plasmid DNA
as measured via coupled transcription-
translation. RLU+SEM, n=3, 370C is significantly
different than 75 and 950C (p<0.01) via Fisher's
PLSD. ................................................. 29

5. Effect of incubation at various pHs for 3 weeks
on the functional activity of plasmid DNA as
measured via coupled transcription-translation.
RLU+SEM, n=6, *=p<0.05 via Fisher's PLSD
compared to pH 7 ......................................30

6. Effect of incubation in pH 3, 2.5 mM citrate
buffer on plasmid DNA in a 0.8% agarose gel,
stained with ethidium bromide. .......................31

7. Effect of citrate buffer concentration on pRL-
CMV plasmid degradation rate at 500C. Mean +
SEM, n=5, r2>0.98 for all fit lines. ..................32

8. Effect of lyophilization (FD) on plasmid DNA
activity, with various amounts lyoprotectant to
DNA (w/w). Average + SEM. n=5 for all
treatments, = FD DNA significantly lower than
Control DNA. p<0.05 via Schefe's multiple
comparison T-test. ...................... ............46


vi









9. Representative thermal analysis scan of salmon
sperm DNA ............................................ 49

10. Effect of lyoprotectants on the melting
temperature of salmon-sperm DNA. Mean + SEM,
n=3, = significantly different than unprotected
control DNA (p < 0.05) via Fisher's PLSD. 50

11. Effect of lyophilization and subsequent
rehydration on the melting of plasmid DNA. Mean +
SEM, n=3 ..............................................51

12. Lyophilized plasmid DNA samples, rehydrated in
DI water and run an agarose gel. Samples:
marker, HindII digested lambda phage marker; FD,
lyophilized DNA without DMED treatment; FD DMED,
lyophilized DNA with DMED treatment; pH 3, DNA
incubated in pH 3.0, 2.5 mM citrate buffer for 15
minutes at 500C with no DMED treatment; pH 3
DMED, DNA incubated in pH 3.0, 2.5 mM citrate
buffer for 15 minutes at 500C with DMED treatment
before running on gel .................................52

13. Effect of lyophilization on plasmid form. Mean +
SEM, n=3. p > 0.05 via ANOVA ........................53

14. Effect of lyophilization on plasmid DNA
absorbance at 260 nm. Average + SEM. n=4, FD
DNA significantly higher than lyoprotected DNA
p<0.05 via ANOVA ..................................... 54

15. Wavelength circular dichroism scan of
lyophilized plasmid DNA (FD) compared to A-form
and B-form plasmid DNA. Each scan is the average
of three separate baseline subtracted scans. ...........55

16. Effect of lyoprotection on plasmid DNA
conformational change. Panels each signify a
separate protectant; A: lactose, B: glucosamine,
C: glucose-l-phosphate, and D: glucose. For each
panel Protectant:DNA 1, 5, 10, 20 and 40 or 1,
2, and 8 w/w ratios. Cntrl S "protectant" is
protectant:DNA solution (40:1 w/w) non
lyophilized. Cntrl FD and Cntrl S "without
protectant" are no protectant, lyophilzed or non
lyophilized respectively. n=10, mean + SEM. ......... 56



vii









17. Kinetics of plasmid conformational change after
rehydration. mean + SEM, n=3.. ..................... 58

18. Effect of rehydration time on plasmid DNA
activity. Mean + SEM. n=6. No significant
difference between 0 and 24 hours by T-test. ......... 59

19. HPLC analysis of hydrolyzed pRL-CMV plasmid DNA.
Panel A is a UV analysis. Peaks by retention
time: 2.76, solvent front, 3.79 cytosine, 7.85
guanine, 11.15 thymine, 13.00 unidentified, 17.89
adenine. Panel B is an ECD analysis. Peaks by
retention time 2,67 3.21 solvent front, 3.53 5-
OH cytosine, 3.82 cytosine, 4.78 plasticizer,
7.35 guanine, 7.92 8-OH guanine, 16.99 8-OH
adenine and adenine. ...................................60

20: Effect of lyophilization of plasmid DNA on
oxidative damage. Positive control was Fe+3
catalyzed oxidation. Mean + SEM, n=5, except
control n=3. = p < 0.05 via Fisher's PLSD
versus control. ........................................ 61

21. Effect of tube type on plasmid DNA bioactivity
after lyophilization. Cntrl = non-lyophilized
DNA, PP = polypropylene, SP = siliconized
polypropylene, PE = polyethylene. Mean + SEM,
n=5, = p < 0.05 via Fisher's PLSD versus
control, X = p < 0.05 versus PP and SP. ..............62

22. Enzyme activity corrected for total cellular
protein after transfection of luciferase plasmid
(lpg) in the presence of endotoxin, with and
without DOTAP : DOPE cationic liposomes. 0 with
lipid (2 ug/ml), U without lipid. RLU+SEM, n=4,
p<0.05 via one way ANOVA for effect of endotoxin
in the presence of cationic lipid. ................... 74

23. Effect of increasing DOTAP : DOPE liposomes
concentration with FITC-endotoxin held constant
(1000 EU) on anisotropy (r). 0 DI Water, m 0.5 M
NaC1, A 1.0 M NaC1, X 2.0 M NaCI. Inset: Free
FITC (22 ng) 1.0 M NaC1. Change in r+SEM, n=3,
p<0.05 using two way ANOVA for increase in r with
increasing lipid under all conditions. ............... 76




viii









24. Effect of increasing luciferase plasmid
concentration on anisotropy (r) with constant
FITC-endotoxin (1000 EU) and DOTAP : DOPE
liposomes (5 gg). Change in r+SEM, n=3, p<0.05
using two way ANOVA for decrease in r with
increasing plasmid. ..................................77

25. Effect of increasing lipid or dendrimer
concentration with FITC-endotoxin held constant
(1000 EU) on anisotropy (r). U Dendrimer
(Generation 2), 0 Dendrimer (Generation 4), X
lecithin liposomes (0.2 pm), A lecithin liposomes
(0.8 gm), Change in r+SEM, n=3, p<0.05 using two
way ANOVA for increase in r with increasing
concentration under all conditions. ..................78

26. Effect of increasing DOTAP : DOPE liposome
concentration with FITC-Endotoxin held constant
(10 gg) .............................................. 78

27. Effect of increasing incubation time of alkaline
phosphatase and unlabeled endotoxin (33 EU) on
anisotropy (r) with constant and NBD labeled
DOTAP : DOPE liposome (10 Vg). Change in r + SEM,
n=4, p>0.05 via one way ANOVA for decrease in
anisotropy over time ................................ 80

28. Effect of increasing lipid concentration with
endotoxin held constant (0.6 EU) on endotoxin
activity, EU/ml + SEM, n=3, p>0.05 via one way
ANOVA for effect of cationic lipid on endotoxin
activity. ............................................. 81

29. Effect of increasing cationic lipid (panel A)
and endotoxin (panel B) on COS-1 cell survival
via MTT assay. mean + SEM, n=5, p < 0.05 via
ANOVA for DOTAP : DOPE, p > 0.05 for endotoxin
via ANOVA. ................. ......................... 82

30. Effect of foam fractionation on FITC labeled
endotoxin levels (initial 100 pg/ml n=3) and BSA
(initial 910 gg/ml, n=3) in a plasmid DNA
solution, (25 ig/ml), mean + SEM. p>0.05 using
one way ANOVA for differences in endotoxin
concentration over time. p<0.05 for differences
in BSA concentration over time. .....................96


ix









31. Effect of foam fractionation on endotoxin
activity. Mean + SEM, p > 0.05 via ANOVA ............97

32. Effect of foam fractionation on physical
stability of plasmid DNA, molecular weight marker
X phage Hind III digest in lane 1; lanes 2 8,
fractionation time points 1 gg pDNA per lane, in
a 0.8% agarose gel stained with ethidium bromide. ....98

33. Surface tension (dynes/cm) of water, endotoxin,
FITC-endotoxin, and BSA (all 1 mg/ml), n=4, mean
+ SEM, = p<0.05 using Fisher's PLSD versus
water .................................................99

34. Volume weighted particle size of FITC-endotoxin
(1 mg/ml) at increasing HC1 concentration. Data
are shown as mean+SEM., p<0.05 via ANOVA, *=
p<0.05 using Fisher's PLSD verus 1 pM HC1 ........... 100

35. Surface tension (dynes/cm) of FITC-endotoxin (1
mg/ml) at various HC1 concentrations, n=4, mean +
SEM, = p<0.05 using Fisher's PLSD versus 0 pm
HC1 .................................................. 101




























x
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy


THE PHARMACEUTICAL STABILITY AND FORMULATION OF PLASMID DNA


By



Scott W. Poxon



May, 1999



Chairman: Jeffrey Hughes, Ph.D.
Major Department: Pharmaceutics


This research examined the stability and formulation

of plasmid DNA from a pharmaceutical perspective. Plasmid

DNA has the potential to be used as a drug, which could

replace missing or damaged proteins that cause disease. As

well, plasmid DNA could be used as the basis for new types

of vaccines. However, the potential for this new drug can

not be realized without research into the formulation and

stability of plasmid DNA. These studies examined the



xi









stability of plasmid DNA with respect to environmental

factors such as temperature and pH, showing that plasmid

DNA is exceedingly stable under many conditions.

Furthermore, this research compared the stability of

plasmid DNA in both solution and solid lyophilized states.

These studies demonstrated that the lyophilization process

damages DNA through what appears to be a conformationally

induced denaturation, although the lyophilized plasmid DNA

is more stable at elevated temperature than plasmid DNA in

solution. Finally, it was ascertained that a potential

contaminant of plasmid DNA, endotoxin, can decrease DNA

transfection efficiency through a competitive electrostatic

interaction with a common DNA delivery vector, cationic

liposomes. It was also established that the endotoxin

contaminant can not be removed from plasmid DNA using foam

fractionation. However, the foam fraction method can

successfully be used to separate endotoxin from amphiphilic

protein products. This research should help to support the

future pharmaceutical development of plasmid DNA as a

therapeutic modality.










xii















CHAPTER 1
INTRODUCTION


Specific Aims and Hypotheses

The overall goal of this project was to determine the

impact of environmental factors on plasmid DNA stability.

Using structural and biofunctional assays, the stability

and potential mechanisms of plasmid DNA instability were

investigated. This information was then assessed with

respect to the pharmaceutical development of plasmid DNA in

a therapeutic modality.

Four specific hypotheses were investigated by this

project. The first hypothesis was that biofunctional assays

are more sensitive than structural assays, since damage at

one base in the encoding region would not be easily

detected by conventional structural methods, but could

result in a functionally inactive product. Secondly, it

was hypothesized that environmental factors will affect the

stability of plasmid in solution. The third hypothesis was

that the stability of lyophilized plasmid DNA to

environmental factors will be higher compared to plasmid

DNA in solution. However, the lyophilization process may


1






2


damage plasmid DNA via conformational strain caused by the

removal of the DNA hydration sphere. Finally, this project

tested the hypothesis that increasing levels of endotoxin

contamination would result in reduced functionality of

plasmid DNA in tissue culture models.

To test these hypotheses, the following were the

specific aims of this work:

1. To establish and validate a sensitive and specific assay

to quantitate the structural and biofunctional stability

of plasmid DNA.

2. To determine the optimum storage conditions for plasmid

DNA in solution.

3. To determine the influence of the hydration sphere on

optimum storage conditions for lyophilized plasmid DNA.

4. To determine the effect of plasmid DNA endotoxin

contamination on functionality using specific assays.


Background and Significance

DNA, or deoxyribonucleic acid, is the genetic material

of life. It is used to pass information from one

generation to the next, supplying the set of genes needed

for the manufacture of further structures in the organism.

DNA functions via its capacity to encode a large variety of

proteins, where specific sequences of nucleotides in DNA






3


encode different proteins. Structurally, DNA consists of

two antiparallel strands whose sequences are made up of

chemically linked subunits, consisting of a nitrogenous

base (purine or pyrimidine) attached to a pentose sugar

linked together by a phosphate backbone. Each nucleic acid

contains one of four types of nitrogenous base: two

purines, adenine and guanine, and two pyrimidines, cytosine

and thymine. In mammals, chromosomes make up the discrete

unit of the genome. Each chromosome consists of a long

duplex DNA complexed with protein.

Another form of DNA, plasmid DNA, is an autonomous

unit that can exist inside a bacterial cell's cytoplasm.

These extrachromosomal, double stranded, circular molecules

of DNA are self-replicating. During the 1970s, many

artificial plasmids were constructed in the laboratory

utilizing fragments from naturally occurring plasmids, and

are commonly used as vectors for recombinant DNA work and

gene therapy.

Since DNA is the genetic material, mutations to DNA

can potentially have serious consequences. A change in the

sequence of DNA causes an alteration of the coded protein,

which may cause mutational inactivation. Mutations may

either be inherited or induced. In either case,

replacement of the mutated DNA via gene therapy can result





4


in the production of competent protein. This makes gene

therapy one of the most exciting and rapidly advancing

areas of medicine. Great strides have been recently made

in the development of DNA as a therapeutic agent including

the successful injection and expression of plasmid DNA into

animals (Nabel et al. 1989; Nabel et al. 1990; Nabel et al.

1992; Stewart et al. 1992). These initial successes have

quickly led to human clinical trials (Nabel et al. 1993;

Caplen et al. 1994).

Although there has been some initial investigation

into the potential production methods for pharmaceutical-

grade DNA (Horn et al. 1995; Durland, and Eastman 1998),

little information is available concerning plasmid DNA

stability in a pharmaceutically acceptable dosage form.

For instance, shelf life has not been determined in any

potential dosage form, and it is commonly assumed that DNA

is stable as either an ethanol precipitate or a frozen

buffered solution at neutral to slightly basic pH (Maniatis

et al. 1982). However, neither postulate has been

thoroughly investigated with respect to the pharmaceutical

development of plasmid DNA.






5


Chemical and Physical Stability of DNA


Background

The reactivity of the phosphate backbone, the

deoxyribose, and the nitrogenous bases that make up the

individual components of DNA has been well documented

(Shabarova, and Bogdanov 1994). Initial research in this

area determined how bases may react with exogenous

electrophilic agents, resulting in heterocyclic

halogenation or nitration. Other electrophilic reactions

include methylation and oxidation. Reactions with amine

containing nucleophilic reagents may also result in

substitutions under basic conditions (Shabarova, and

Bogdanov 1994). Of these reactions, oxidation is most

likely of concern in long-term DNA storage since the

plasmid isolation process should not result in the addition

of exogenous electrophiles (Finnegan et al. 1996).

However, plasmid isolation may result in the addition of

activated oxygen species through air saturated buffers or

the use of organic solvents for extraction resulting in

subsequent oxidative damage (Finnegan et al. 1996).

Although these previous reactions have generally been

characterized experimentally using nucleic acid monomers,

it is possible that the reactivity of plasmid DNA may





6

increase or decrease significantly from the monomers due to

steric factors associated with the primary and secondary

structures of plasmid DNA. Several specific areas of the

DNA have been extensively studied for their lability under

physiologic conditions. The sites susceptible to

hydrolytic attack and oxidative damage in plasmid DNA can

be seen in Figure 1.





Cytosine Guanin /
SH2 "........' NA -O-

5"' "'NI N INN'N

-o0 ......H2N 9

"........ H2 PN
9 5
-0- N 0 Adenine
0- Thymine



3'

Figure 1: Major sites for chemical degradation of DNA.
Filled arrows point to potential oxidative damage sites,
while open arrows point to potential hydrolytic damage
sites.






7


Glycosidic Bond

The glycosidic bond is highly stable under neutral or

basic conditions, but is extremely sensitive to acid

hydrolysis. Acid hydrolysis of the glycosidic bond on

nucleotides occurs rapidly with purines (k=10-4sec-1) and

more slowly with pyrimidines (k=10-8sec-1) at pH 1.0, 370C,

since the purines are better leaving groups than

pyrimidines (Lindahl 1993). Under physiologic conditions,

it has been predicted that mammalian cells undergo 2,000 to

10,000 cases of hydrolytic depurination followed by repair

every day (Lindahl 1993). In pharmaceutical storage of

plasmid DNA, however, no repair mechanisms exist, bringing

concern about potential degradation of the glycosidic bond.

NH12
5' 5' 5'

ON O --- HOH 0 ----0 O


O-6 H H -O





Figure 2: Glycosidic cleavage of cytosine to an aldehyde
under alkaline or acidic conditions followed by beta
elimination to cleave the phosphate backbone.

After depurination by cleavage of the glycosidic bond,

an aldehyde forms at the C1 of deoxyribose, which can then






8


undergo beta elimination as shown in Figure 2 (McHugh, and

Knowland 1995). This aldehyde form exists in equilibrium

with the cyclic depurinated hemiacetal form, with about 1%

of the base-free sugar residue in the aldehyde form at any

one time. Even without chemical catalysis, the weakened

DNA chain would undergo the elimination process within a

few days, rendering the stored plasmid DNA non-functional

by blocking DNA polymerase.

It has been shown that deamination of the bases may

also occur in alkaline conditions, resulting in potential

loss of the encoded protein's functionality. This

deamination process occurs mainly via acid-catalysis under

physiologic conditions (Lindahl 1993). This, however, is a

slow reaction, with the half-life of an individual cytosine

residue in single stranded DNA extrapolated to 200 years

(Frederico et al. 1990). The hydrolytic deamination

reaction slows further with double stranded DNA to a half-

life of about 30,000 years for each cytosine residue

(Frederico et al. 1990). Deamination occurs at a slower

rate than depurination, and is therefore less likely to be

the rate-determining step in plasmid degradation during

storage.






9


Phosphodiester Bond

During prolonged exposure to high temperatures, DNA

will progressively melt via first order kinetics followed

by heat-induced hydrolysis of the phosphodiester bond with

an expected 3,000 fold increase in DNA decay at 1000C as

compared to 370C (Lindahl 1993). At temperatures above

100C the DNA is very unstable because of both its chemical

nature to hydrolyze and the problem of retaining the

hydrogen bonding between the two DNA strands at high

temperatures. It has also been suggested that increased

pressure could help stabilize DNA, since the melting

temperature of the double helix is 100C higher at 5,000

atmospheres than at 1 atmosphere (Lindahl 1993).

The phosphodiester bond is another potential site for

damage in plasmid DNA. It may be broken by beta

elimination as well as by acid hydrolysis at a pH below 3.

The phosphodiester backbone can also be cleaved by

nucleases and by oxidative degradation, but is otherwise as

stable at neutral and basic pH as the glycosidic bond and

the amine groups on the bases. This explains the reasoning

behind current DNA storage paradigms of freezing DNA at

neutral to slightly basic pH.






10


Steric Effects

Positive supercoiling can act as a barrier to

deamination, depurination and subsequent beta elimination.

Protection via supercoiling is due to a change in the

plasmid's tertiary structure that results in torsional

twisting of the DNA structure. This twisting causes the

duplex to cross itself in space and is responsible for the

increased pH needed to denature circular DNA as compared to

linear DNA. Supercoiling can also protect against thermal

DNA degradation (Marguet, and Forterre 1994), suggesting

that condensation of plasmid DNA may increase stability

during storage.


Oxidative Damage

Oxidative damage of plasmid DNA may be a problem

inherent in long-term storage, particularly if the DNA was

exposed to hydroxyl or superoxide radicals (Lindahl 1993).

Exposure to activated oxygen radical is ubiquitous inside

cells, where DNA repair enzymes exist that reduce the

damage to cellular DNA. In long-term storage of plasmid

DNA, these repair enzymes will not exist, so oxidative

damage can accrue resulting in non-functional protein

product. Damage can potentially occur during plasmid DNA

isolation with solvents such as phenol and chloroform, and






11


air-saturated buffers (Lindahl 1993; Finnegan et al. 1996).

Exposure to activated oxygen can cause the formation of

formamidopyrimidines, purines with opened imidazole rings,

that are non-coding residues. As well, hydroxyl radicals

can react with guanine, forming 8-hydroxyguanine, which

base pairs with adenine rather than cytosine. Therefore,

transversion mutations will be generated after replication.

Another common oxidative-induced DNA damage is the

generation of ring saturated pyrimidines. In this case,

losing the 5,6 double bond causes the loss of planar ring

structure leaving a non-coding base residue (Lindahl 1993).


Lyophilization of DNA

Lyophilization of plasmid DNA may be a preferred form

of storage, potentially imparting a longer shelf life and

greater stability against heat-induced degradation. /

Experimental findings support this by suggesting that

biological macromolecules demonstrate an increase in

stability from the frozen to the lyophilized state (Crowe

et al. 1990). The exact mechanism underlying this

difference has not been fully elucidated, but most likely

relates to the nonfreezable water associated with most

biomolecules, including nucleic acids. It is well known

that the structure and conformational states of DNA are






12


critically dependent on hydration level (Umrania et al.

1995; La Vere et al. 1996). Several forms of DNA secondary

structure are known and, under normal physiological

conditions, DNA is found in the B-form. This form has a

major and a minor groove, with 10 base pairs per turn. The

A-form is more compact, with 11 base pairs per turn, and is

seen when the presence of 2'hydroxyl groups on the ribose

prevents formation of the B-form during complexation with

RNA, and under conditions of high salt concentration, or

low humidity (<75%), as expected under lyophilized

conditions. While the A and B-forms are both right handed

helixes, the Z-form is a left handed helix and has the most

base pairs per turn of any of the forms, and only one

groove. This form has been seen under high salt

concentrations and probably does not exist in vivo but

could potentially be seen in low concentrations under the

high salt conditions seen with lyophilization. The

secondary structure of the DNA may affect stability and

functionality.

Lyophilization causes the removal of the hydration

sphere around a molecule. For DNA, it appears that there

are approximately 20 water molecules per nucleotide pair

bound most tightly to DNA that do not form an ice-like

structure upon low temperature cooling. Upon DNA






13


dehydration at 0% relative humidity only five or six water

molecules remain (Falk et al. 1963; Lith et al. 1986; Tao,

and Lindsay 1989). Lyophilization may increase the

stability of DNA under long-term storage, but may also

cause some damage upon the initial lyophilization process,

potentially through changes in the DNA secondary structure.

Agents that can substitute for nonfreezable water, such as

trehalose, can demonstrate cryoprotective properties for

DNA and other molecules during lyophilization of intact

bacteria (Rudolph et al. 1986; Israeli et al. 1993). Other

cryoprotective agents, such as polyols, amino acids, sugars

and lyotropic salts are preferentially excluded from

contact with protein surfaces but are also capable of

stabilizing enzymes during lyophilization by undefined

mechanisms (Carpenter, and Crowe 1989). It is possible

that agents that act as cryoprotectants to proteins may

also act to stabilize nucleic acids.


Endotoxin Contamination of DNA

Since plasmid DNA is typically produced by bacteria,

endogenous bacterial products can potentially contaminate

the final plasmid DNA product. Typical problem compounds

in plasmid DNA preparations include endotoxin, a cell wall

component that is pyrogenic in man, and DNase, a ubiquitous






14


bacterial enzyme that can degrade the final plasmid

product. Endotoxin from the cell wall of Gram-negative

bacteria, such as the E. coli typically used to prepare

plasmid DNA, may be the most serious concern for gene

therapy because of its extreme toxicity. Several plasmid

DNA preparation methods have been examined for level of

endotoxin contamination (Weber et al. 1995).

Endotoxin has a relatively low lethal dose for 50% of

tested animals (LDso) in rats of 3 mg per kg (Shibayama et

al. 1991) and an LD50 in dogs of only 1 mg per kg (Fletcher,

and Ramwell 1980). The endotoxin, or lipopolysaccharide, of

E. coli consists of a polysaccharide component and a

covalently bound lipid component termed lipid A. The lipid

A portion on the endotoxin is biologically active and can

cause a number of pathophysiological effects including

fever, hypotension, intravascular coagulation and death

(Fletcher, and Ramwell 1980; Aida, and Pabst 1990;

Rietschel et al. 1993; Xing et al. 1994). The endotoxins

are acute inflammatory mediators, elevating levels of

cytokines (Xing et al. 1994). Furthermore, an increasing

level of endotoxin contamination has been shown to decrease

transfection efficiency using DOTAP : DOPE (dioleoyl

glycero trimethylammonium propane : dioleoyl glycero

phosphoethanolamine) cationic liposomes to deliver plasmid





15


DNA (Weber et al. 1995). Besides having an easy way to

detect endotoxin, there are several ways to remove

endotoxin from DNA samples. Phase separation using Triton

X-114 has been used to reduce endotoxin levels for both

protein and plasmid DNA (Aida, and Pabst 1990; Manthorpe et

al. 1993). The use of these methods to reduce endotoxin

contamination may be of importance in the potential

usefulness of plasmid DNA.


Conclusions

It is important to understand the effects of the

formulation parameters such as pH, temperature, and buffer

composition on plasmid DNA stability. Furthermore, the

differences between solution and lyophilized plasmid

stability must be taken into account to determine the

optimal plasmid DNA storage conditions. This research will

address the stability of plasmid DNA by examining plasmid

structural integrity and biofunctionality. This

distinction is important, since current structural methods

will not always detect changes in the functionality of the

final protein product that can be produced by damage to the

plasmid DNA. Using the information provided by these

studies, it should be possible to develop a rational basis

for the plasmid DNA manufacturing process.















CHAPTER 2
PHYSICAL AND CHEMICAL STABILITY OF PLASMID DNA


Background

Before the benefits of gene therapy can be realized on

a large scale, the pharmaceutical community needs to

ascertain the optimal storage conditions for plasmid DNA so

as to prolong shelf life to the greatest possible extent.

While a great deal of work has gone into the study of

individual nucleotides, little has been accomplished to

date regarding the bioactivity of plasmid DNA, or the

ability of plasmid DNA to code for a functional protein, as

a marker for shelf life. Other methods have been used to

examine DNA stability, including agarose gel

electrophoresis, HPLC, Southern blotting, and PCR (Niven,

Pearlman et. al 1998; Strege and Lagu 1991; Thierry,

Lunardi-Iskandar et. al 1995). However, a bioactivity based

method will detect changes not only in physical stability

of plasmid DNA but also will detect cellular changes in the

DNA handling, including, but not limited to, transport,

transcription, and translation.




16






17


The stability of the subunits of plasmid DNA, the

individual purine and pyrimidine bases, and their

attachment to the ribose sugar and phosphate backbone are

fairly well understood. The glycosidic bond is highly

stable under neutral or basic conditions, but is extremely

sensitive to acid hydrolysis. Acid hydrolysis of the

glycosidic bond on nucleotides occurs rapidly with purines

(k=10-4sec-) and more slowly with pyrimidines (k=10-ssec-1)

at pH 1.0, 37C, since the purines are better leaving

groups than pyrimidines (Lindahl 1993). Under physiologic

conditions, it has been predicted that mammalian cells

undergo 2,000 to 10,000 cases of hydrolytic depurination

followed by repair every day (Lindahl 1993). In

pharmaceutical storage of plasmid DNA, however, no repair

mechanisms exist, bringing concern about potential

degradation of the glycosidic bond.

After depurination by cleavage of the glycosidic bond,

an aldehyde forms at the Cl of deoxyribose, which can then

undergo beta elimination as shown in Figure 2 on page 7

(McHugh, and Knowland 1995). This aldehyde form exists in

equilibrium with the cyclic depurinated form, with about 1%

of the base-free sugar residue in the aldehyde form at any

one time. Even without chemical catalysis, the weakened

DNA chain will undergo the elimination process within a few






18


days. This would render the stored plasmid DNA non-

functional by blocking DNA polymerase.

It has been shown that deamination of the bases may

occur in alkaline conditions, resulting in potential loss

of the encoded protein's functionality. This process can

also occur via acid hydrolysis under physiologic

conditions, where the half-life of an individual cytosine

residue in single stranded DNA has been extrapolated to 200

years (Frederico et al. 1990). This hydrolytic deamination

reaction slows further with double stranded DNA to a half-

life of about 30,000 years for each cytosine residue

(Frederico et al. 1990). This reaction occurs at a slower

rate than depurination and is therefore less likely to be

the rate-determining step in plasmid degradation during

storage. Depurination rates may be effected by changes in

the formulation of plasmid DNA.

During prolonged exposure to high temperatures, DNA

will progressively melt via first-order kinetics followed

by heat-induced hydrolysis of the phosphodiester bond, with

an expected 3,000 fold increase in DNA decay at 1000C as

compared with 370C (Lindahl 1993) This rate increase

essentially follows the 100C rule with a 3-fold rate

increase per 100C temperature increase. At temperatures

above 1000C the DNA is very unstable because of both its





19


chemical nature to hydrolyze and the problem of retaining

the hydrogen bonding between the two DNA strands at high

temperatures. It has also been suggested that increased

pressure could help stabilize the DNA, since the melting

temperature of the double helix is 100C higher at 5,000

atmospheres than at 1 atmosphere (Lindahl 1993).

The phosphodiester bond is another potential damage

site for plasmid DNA. It may be broken by beta elimination

as well as by acid hydrolysis at a pH below 3. The

phosphodiester backbone can also be cleaved by nucleases

and by oxidative degradation, but is otherwise as stable at

neutral and basic pH as the glycosidic bond and the amine

groups on the bases. This explains the reasoning behind

current DNA storage paradigm of frozen DNA at neutral to

slightly basic pH.

Positive supercoiling can act as a barrier to

deamination, depurination and subsequent beta elimination.

This supercoiling protection is due to a change in the

plasmid's tertiary structure that results in torsional

twisting of the DNA structure. This twisting causes the

duplex to cross itself in space and this conformationally

induced steric protection is responsible for the increased

pH needed to denature circular DNA as compared to linear

DNA. Supercoiling can also protect against thermal DNA






20


degradation (Marguet, and Forterre 1994). If supercoiling

exerts a protective effect, then it follows that DNA

delivery systems that further condense supercoiled plasmid

DNA may additionally increase stability during storage. As

well, supercoiled plasmid DNA may consequently be more than

the monomers that were originally used to determine DNA

stability.

While it is possible to examine plasmid DNA stability

through examination of any of the above individual

degradation pathways, a better way to examine DNA stability

may be through the examination of plasmid DNA bioactivity,

by assaying for enzymatic activity of the plasmid's encoded

gene. Assaying for activity of the plasmid DNA will allow

a comprehensive determination of degradation that will

include all types of damage that affect the expressed gene

rather than just the several specific types of damage

previously mentioned. The activity of the plasmid can be

determined through one of several bioactivity assays.

First, plasmid DNA can be assayed using an in vitro coupled

transcription-translation system. The in vitro system

allows a one tube determination of activity that is not

dependent on any particular type of cell system for

expression. Rather, the in vitro system includes all the

necessary enzymes and cofactors to transcribe the plasmid






21


DNA to mRNA and to further translate the mRNA to protein,

which can then be assayed. This method's main disadvantage

is due to variability induced by the 'stop-time" of the

reaction. If the reaction is stopped at different times,

the resulting enzymatic activity will be effected.

The second method for determining the bioactivity of

the plasmid is using a mammalian cell based tissue culture

transfection assay. This method requires that the plasmid

DNA be complexed with a delivery agent before it is

transducted into a mammalian cell line. The cell line will

then express the protein of interest, which can then be

assayed. Several main types of DNA damage are expected, as

discussed previously. This study will examine the

stability of plasmid DNA using both types of activity

assays.


Materials and Methods


Plasmid Purification

Vectors pGL3 plasmid, pSPluc+, and pRL-CMV (Promega,

Madison, WI), respectively encoding for photinus

luciferase, photinus luciferase and renilla luciferase,

were grown in E. coli DH5a cells (Promega, Madison, WI).

The transformed bacteria were cultivated in Lura Bertina

(LB) broth containing 100 ug/ml ampicillin to select for 3-






22


lactamase encoding plasmid. The plasmids were isolated via

an alkaline lysis method and purified using a silica slurry

column (Wizard Plus Megapreps, Promega, Madison, WI). The

plasmid was stored in TE buffer (10 mM Tris-HCl, 1 mM EDTA,

pH 7.4).


Liposome Preparation

Lipids were obtained from Avanti Polar Lipids

(Alabaster, AL). Cationic DOTAP : DOPE (dioleoyl glycero

trimethylammonium propane : dioleoyl glycero

phosphoethanolamine) liposomes were prepared using the

hand-shaking method (New 1990). DOTAP: DOPE (10 mg DOTAP,

10 mg DOPE) was added to 10 ml of chloroform and introduced

into a 250 ml round bottom flask. The chloroform was

evaporated using a rotary evaporator at 600C. The lipid

film was hydrated with 10 ml of distilled water and shaken

for 30 minutes at 600C. Lipids were sized by extrusion, six

times through 600 nm polycarbonate filters (Poretics Corp,

Livermore, CA). Sizes were confirmed with a laser light

scattering particle sizer using volume-weighted

distribution (Nicomp 380ZLS, Santa Barbara, CA).





23


In vitro Coupled Transcription-Translation

The coupled transcription-translation assay was

carried out using a commercially available kit (Promega,

Madison, WI). Essentially, pGL3 plasmid DNA (100 pg) was

incubated for 60 minutes at 300C with a proprietary mixture

containing wheat germ extract, RNA polymerase, amino acids,

and a ribonuclease inhibitor. The resulting protein

product, luciferase, was then assayed for functionality

using a luminescence spectrophotometer (Monolight 2010,

Analytical Luminescence Laboratory, San Diego, CA). The

light output was measured for 10 seconds and the results

integrated to yield the activity.


Transfection Efficiency Assay

Untreated pGL3 (0.05 pg/ml) and treated pRL-CMV (0.5

pg/ml) were combined with cationic lipid mixture (DOTAP :

DOPE 1:1, 2 pg/ml each) and incubated for 15 minutes to

allow interaction of the anionic plasmid with the cationic

liposomes. The 10:1 concentration ratio was as suggested

by the manufacturer to achieve a linear relationship

between treated renilla luciferase plasmid concentration

and renilla luciferase / photinus luciferase activity ratio

by decreasing crosstalk between the promoters. These

mixtures were then co-transfected into COS-1 cells (ATCC,






24


Rockville, MD) that had been plated 24 hours earlier at

3x104 cells/ml in 24 well tissue-culture dishes. All

transfections were effected using serum free DMEM

(Dulbecco's Modified Eagle's Medium) supplemented with

penicillin G (100 units/ml) and streptomycin (100 pg/ml).

The cells were subsequently incubated for 5 hours in a

humidified, 5% CO2 incubator at 37C. The media was

aspirated and the cells washed with PBS (phosphate buffered

saline) before being replaced by DMEM supplemented with 10%,

FBS (Fetal Bovine Serum), penicillin G (100 units/ml) and

streptomycin (100 pg/ml). The cells were incubated for an

additional 19 hours, washed with 1 ml PBS, and then lysed

with Passive Lysis buffer (200 pL, Promega, Madison, WI).

Photinus luciferase and renilla luciferase activities

were determined by analyzing 5 pL of the lysate using the

Dual Luciferase Assay kit (Promega, Madison, WI) and a

luminescence spectrophotometer (Monolight 2010, Analytical

Luminescence Laboratory, San Diego, CA). The light output

was measured for 10 seconds and the results integrated to

yield the activity. All pRL-CMV, renilla luciferase

plasmid, measurements were standardized to the ratio of

renilla luciferase (rluc treated) activity to photinus

luciferase (luc control) activity. Essentially, the ratio






25


of rluc/luc is linear to the concentration of rluc added to

COS-1 cells. Therefore, a linear regression of rluc/luc

ratio at various concentrations of rluc and inactivated

rluc (cut with restriction enzymes), with luc held

constant, allows a determination of the apparent

concentration of the treated pRL-CMV plasmid.


DMED Assay

Treated pGL3 DNA was incubated with N,N'-

dimethylethylenediamine (100 mM, pH 7.4, 30 minutes at

37C), cleaving abasic sites via 3-elimination (McHugh, and

Knowland 1995). The resultant DNA was then visualized

using a 0.8% agarose gel. The gel was electrophoresed at 1

V/cm for 16 hours to affect good separation between the

supercoiled, relaxed circular and linear forms of the

plasmid DNA. The gel was then stained with ethidium

bromide and photographed on an UV light box using a DC40

digital camera (Kodak, Rochester, NY).


pH Solution Stability Study

The effect of pH upon plasmid DNA bioactivity was

studied using a multiple component buffer (Schrier et al.

1993). Poly-B buffer, equimolar sodium citrate, sodium

succinate, Tris, HEPES, imidazole, histidine, and glycine

at various pHs, was added to pSP6-luc plasmid DNA to a






26


final concentration of 10 mM each salt and 0.25 mg/ml

plasmid DNA. This mixture was incubated at room

temperature for three weeks, with the assumption that pH

did not change over time. The plasmid was then

precipitated using 70% ethanol, and the pellet rehydrated

in TE buffer (10 mM Tris-HC1, 1 mM EDTA, pH 7.4) before

analysis by the coupled transcription-translation method.

The potential for buffer catalysis of plasmid DNA was

examined using a citrate buffer at four different

concentrations, all at pH 3, 500C. Plasmid, pRL-CMV, was

incubated with the various buffers, with aliquots withdrawn

over a course of five hours. The aliquots were returned to

neutral pH by the addition of excess TE buffer, pH 7.4.

The pH was verified to be greater than 7.0 using litmus

paper. These samples then were assayed using the

cotransfection method.


Temperature Stability Study

The effect of temperature upon plasmid DNA bioactivity

was examined after a three-week incubation at four

temperatures: 25, 37, 75 and 950C. Plasmid samples, pSP6-

luc, were stored in the humidified incubator at 1 mg/ml in

TE buffer in sealed 1 ml cryogenic screw-top vials. After






27


the three-week period, all samples were cooled and assayed

using the coupled transcription-translation method.


Statistical Analysis

Statistical analysis between the various treatments

was conducted using analysis of variance and Fisher's

protected least significant difference (PLSD) post-hoc T-

tests where appropriate (StatView v4.5, Abacus Concepts,

Berkley, CA), with p<0.05 considered statistically

significant.


Results and Discussion

The luciferase cotransfection method used two plasmids

coding for two different enzymes (fixed concentration

photinus and varying concentration renilla luciferase) to

help control for variability by comparing the ratio of the

two enzymes activity after transfection. The raw data was

analyzed by comparing only the activity of the renilla

luciferase. The luciferase cotransfection method of

normalizing transfection efficiency was compared to single

transfection, raw, non-normalized data shown in Figure 3.

It can be seen that the normalized cotransfection data

shows a better fit and shows less variability than the

single plasmid raw data. The percent coefficient of

variation (CV) and percent bias seen with both methods at






28


each measured concentration can be seen in Table 1. The

trend toward lower CV at plasmid concentrations above 0.13

gg / ml with the cotranfection suggests that the luciferase

cotransfection method is more sensitive than the single

plasmid method. Another trend is systematic bias below the

fit line, suggesting that the normalized method is more

reproducible than the single plasmid luciferase methods.

A 5 B o _
R2 0.9982 R' 0.9575
4 48



I 02--


0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
pRL-CMV (ug) pRL-CMV (ug)

Figure 3: Comparison of cotransfection (A) and single
plasmid (B) concentration vs activity. Mean + SEM, n=4.


Table 1: Comparison of variance and bias between the
luciferase cotransfection and single plasmid methods


% CV % Bias
jg DNA Cotransfection Single Cotransfection Single
0.03 42 27 21 8
0.06 57 40 20 25
0.13 21 29 4 26
0.25 21 20 3 27
0.50 4 38 0 9


Extended incubation of pSP6-luc plasmid DNA for three

weeks at elevated temperature can result in a significant

decrease in the bioactivity of the treated plasmid at





29


elevated temperatures (Figure 4). This degradation would

be expected through heat-induced hydrolysis of the

phosphodiester bond. The bioactivity assay demonstrated a

three order of magnitude decrease in plasmid DNA activity

after incubation at 950C, which is in line with an expected

3,000 fold increase in DNA decay at 1000C compared to 370C

as previously reported in the literature (Lindahl 1993).



106


105


j 104


103


10


0
25 37 75 95
Storage Temperature (C)
Figure 4: Effect of incubation at elevated temperature for
3 weeks on the functional activity of plasmid DNA as
measured via coupled transcription-translation. RLU+SEM,
n=3, 370C is significantly different than 75 and 95C
(p<0.05) via Fisher's PLSD.

The effect of pH on pSP6-luc plasmid DNA bioactivity

was also examined (Figure 5). It can be seen that the





30


plasmid DNA remained exceedingly stable across a wide pH

range. Degradation was seen only at pH 1 and 2. This is

not suprising, since the glycosidic bond is reported in the

literature to be highly stable under neutral or basic

conditions, but is reported to be extremely sensitive to

acid hydrolysis (Lindahl 1993). Furthermore, the

phosphodiester bond may be broken by beta elimination as

well as by acid hydrolysis at a pH below 3. The loss of

bioactivity seen at low pH is most likely due to a

combination of the two degradation pathways.



106

105


104

n, 103

102

10



0
1 2 4 6 7 8 10
Storage pH
Figure 5: Effect of incubation at various pHs for 3 weeks
on the functional activity of plasmid DNA as measured via
coupled transcription-translation. RLU+SEM, n=6, *=p<0.05
via Fisher's PLSD compared to pH 7





31


Using the DMED assay to examine the mechanism of

acidic plasmid DNA degradation, it can be seen that nearly

100% of the plasmid DNA is depurinated at one or more sites

per plasmid molecule after 15 minutes (Figure 6). This

degradation rate is reasonable compared to published rates

with a T1/2 of 16 minutes at pH 1.0, 370C for depurination

(Lindahl 1993).


Standard DMED




OC
NL



SC


X 0 15 30 60 120 240 X 0 15 30 60 120 240

Minutes Minutes

Figure 6: Effect of incubation in pH 3, 2.5 mM citrate
buffer on plasmid DNA in a 0.8% agarose gel, stained with
ethidium bromide.

It was also interesting to note that besides

undergoing a pH dependent degradation, plasmid DNA may also

be subject to buffer catalysis. Statistically different

degradation rates were seen when incubating plasmid DNA,






32


pRL-CMV, in a 500C, pH 3.0 solution, buffered with various

amounts of citrate (Figure 7). Degradation rates appear to

be first-order exponential and increase with citrate buffer

concentration through an undetermined mechanism.


3

2.5

2 ~*.5 mM
S0 01 mmM
C 1.5 A A2mM

1 \

0.5 1..
0 5 10 15 20 25 30
Incubation Time (min.)


Figure 7: Effect of citrate buffer concentration on pRL-CMV
plasmid degradation rate at 500C. Mean + SEM, n=5, r2>0.98
for all fit lines.

In conclusion, plasmid DNA is exceedingly stable in

solution over a wide range of pH and temperature extremes.

However, the choice of buffer is of more importance than

has typically been considered in the past. Based on the

literature review and the data presented here, it is most

advisable to formulate plasmid DNA at neutral to slightly

basic conditions, using previously tested buffers. With

buffers that exhibit protective effects, a high

concentration of buffer would be preferable. With buffers

that that exhibit buffer catalysis, a low buffer






33


concentration that will minimize any potential buffer

effects should be used.















CHAPTER 3
FORMULATION OF PLASMID DNA: THE EFFECT OF LYOPHILIZATION ON
PLASMID DNA STABILITY


Introduction

Gene therapy is one of the fastest growing areas in

therapeutics. While rapid progress has occurred in this

field, the pharmaceutical community has not addressed all

concerns in detail. Of particular interest is the final

dosage form. It is well understood that naked DNA

introduced into a patient's circulatory system does not

reach enough of the appropriate cells and therefore has

little chance of affecting most disease processes

(Friedmann 1997). This has led to the development of a

number of gene delivery vectors, such as cationic

liposomes. Liposomes are considered one of the more

promising systems for use in gene delivery (Xu, and Szoka

1996). Unfortunately, cationic liposomes complexed with

plasmid DNA are not stable for long-term storage,

undergoing aggregation over time (Anchordoquy et al. 1997).

There have been several studies with non-ionic

liposomes, which may apply to the stability of the



34





35


DNA/cationic liposome mixture during the lyophilization

process. Lyophilization of non-ionic liposomes in the

presence of carbohydrates has been cited as one of the most

promising ways to keep the liposome stable under long term

storage (Williams, and Polli 1984). The lyophilization

process, without lyoprotectants, can lead to the leakage of

the inner aqueous phase due to liposomal fusion and phase

separation of liposomal membranes during drying and

rehydration (Crowe et al. 1988). Using carbohydrates as a

lyoprotectant will prevent mechanical rupture of the

liposomal membrane, caused by ice crystals, during the

freezing process and will prevent membrane disruption

during drying and rehydration by maintaining the membrane

in a flexible state (Ozaki, and Hayashi 1997).

Although DNA is complexed with cationic liposomes, not

entrapped, it has been shown that the liposome/DNA complex

requires lyoprotectants to maintain transfection efficiency

after lyophilization (Anchordoquy et al. 1997). Moreover,

other gene delivery systems containing polyethylenimine,

polylysine, and adenovirus particles require cryoprotection

to maintain transfection efficiency (Talsma et al. 1997).

However, it is unclear if this effect is due only to

stabilization of the liposomal component of the mixture or

if the plasmid DNA component may itself be damaged by






36


lyophilization in the absence of carbohydrate

lyoprotectants.

Lyophilized plasmid DNA may be a preferred form of

storage, potentially imparting a longer shelf life and

greater stability against heat-induced degradation, since

biological macromolecules demonstrate an increase in

stability from the frozen to the lyophilized state (Crowe

et al. 1990). The exact mechanism underlying this

difference has not been fully elucidated, but most likely

relates to the nonfreezable water associated with

biomolecules, including nucleic acids. It is well known

that the structure and conformational states of DNA are

critically dependent on hydration level (Umrania et al.

1995; La Vere et al. 1996).

Several forms of DNA secondary structure are known and

under normal physiological conditions most DNA is found in

the B-form. The B-form has a major and a minor groove with

10 base pairs per turn. The A-form is more compact with 11

base pairs per turn. It is observed when DNA is complexed

with RNA, due to the presence of 2'hydroxyl groups on the

ribose preventing formation of the B-form during

complexation. The A-form is also seen under conditions of

low humidity (<75%) or high salt concentration (Umrania et

al. 1995; La Vere et al. 1996). This A-form secondary





37


structure is therefore observed under lyophilized

conditions. While the A and B-forms are both right handed

helixes, the Z-form is a left handed helix and has the most

base pairs per turn of any of the forms and only one

groove. This form has been seen under high salt

concentrations and probably does not form in vivo but could

potentially be seen in low concentrations under the high

salt conditions seen with lyophilization. The secondary

structure of the DNA may effect stability and

functionality.

Lyophilization causes the removal of the hydration

sphere around a molecule. For DNA, it appears that there

are approximately 20 water molecules per nucleotide pair

bound most tightly to DNA that do not form an ice-like

structure upon low temperature cooling. Upon DNA

dehydration at 0% relative humidity only five or six water

molecules remain (Falk et al. 1963; Lith et al. 1986; Tao,

and Lindsay 1989). Lyophilization may increase the

stability of DNA under long-term storage because of

decreased water activity, but may also cause some damage

upon the initial lyophilization process, potentially

through changes in the DNA secondary structure. Agents

that can substitute for nonfreezable water, such as

trehalose, can demonstrate cryoprotective properties for






38


DNA and other molecules during lyophilization of intact

bacteria (Rudolph et al. 1986; Israeli et al. 1993). Other

cryoprotective agents, such as polyols, amino acids, sugars

and lyotropic salts are preferentially excluded from

contact with protein surfaces but are also capable of

stabilizing enzymes during lyophilization by undefined

mechanisms (Carpenter, and Crowe 1989). It is possible

that agents that act as cryoprotectants to proteins may

also act to stabilize nucleic acids.


Materials and Methods


Plasmid Purification

Both pGL3 plasmid and pRL-CMV (Promega, Madison, WI),

respectively encoding for photinus luciferase and renilla

luciferase, were grown in E. coli JM109 cells (Promega,

Madison, WI). The transformed bacteria were cultivated in

Lura Bertina (LB) broth containing 100 pg/ml ampicillin to

select for P-lactamase encoding plasmid. The plasmids were

isolated via an alkaline lysis method and purified using a

silica slurry column (Wizard Plus Megapreps, Promega,

Madison, WI). The plasmid was stored in TE buffer (10 mM

Tris-HCl, 1 mM EDTA, pH 7.4). Plasmid, pRL-CMV, that was to

be lyophilized was aliquoted (50 ul) into 1.5 ml






39


polypropylene tubes (Sarstadt) and frozen at -800C. The

frozen samples were then lyophilized for 24 hours under

vacuum using a Savant Centrivap and a -1050C cold trap.

Lyophilized samples were rehydrated with distilled water

immediately before use.


Liposome Preparation

Lipids were obtained from Avanti Polar Lipids

(Alabaster, AL). Cationic DOTAP : DOPE (dioleoyl glycero

trimethylammonium propane : dioleoyl glycero

phosphoethanolamine) liposomes were prepared using the

hand-shaking method (New 1990). DOTAP: DOPE (10 mg DOTAP,

10 mg DOPE) was added to 10 ml of chloroform and introduced

into a 250 ml round bottom flask. The chloroform was

evaporated using a rotary evaporator at 600C. The lipid

film was hydrated with 10 ml of distilled water and shaken

for 30 minutes at 600C. Lipids were sized by extrusion 6

times through 600 nm polycarbonate filters (Poretics Corp,

Livermore, CA). Sizes were confirmed with a laser light

scattering particle sizer using volume-weighted

distribution (Nicomp 380ZLS, Santa Barbara, CA).





40


Transfection Efficiency Assay

Untreated pGL3 (0.05 pg/ml) and treated pRL-CMV (0.5

pg/ml) were combined with cationic lipid mixture (DOTAP :

DOPE 1:1, 2 pg/ml each) and incubated for 15 minutes to

allow interaction of the anionic plasmid with the cationic

liposomes in TE buffer (Tris HC1, 10 mM; EDTA, 1 mM, pH

7.4). The mixtures were then co-transfected into COS-1

cells (ATCC, Rockville, MD) that had been plated 24 hours

earlier at 3x104 cells/ml in 24 well tissue-culture dishes.

All transfections were effected using serum free DMEM

(Dulbecco's Modified Eagle's Medium) supplemented with

penicillin G (100 units/ml) and streptomycin (100 gg/ml).

The cells were subsequently incubated for 5 hours in a

humidified, 5% CO2 incubator at 370C. The media was

aspirated and the cells washed with PBS (phosphate buffered

saline) before being replaced by DMEM supplemented with 10%

FBS (Fetal Bovine Serum), penicillin G (100 units/ml) and

streptomycin (100 pg/ml). The cells were incubated for an

additional 19 hours, washed with 1 ml PBS, and then lysed

with Passive Lysis buffer (200 pL, Promega, Madison, WI).

Photinus luciferase and renilla luciferase activities

were determined by analyzing the 5 pL of the lysate using

the Dual Luciferase Assay kit (Promega, Madison, WI) and a






41


luminescence spectrophotometer (Monolight 2010, Analytical

Luminescence Laboratory, San Diego, CA). The light output

was measured for 10 seconds and the results integrated to

yield the activity. All pRL-CMV, renilla luciferase

plasmid, measurements were standardized to the ratio of

renilla luciferase (rluc treated) activity to photinus

luciferase (luc control) activity. Essentially, the ratio

of rluc/luc is linear to the concentration of rluc added to

COS-1 cells. Therefore, a linear regression of rluc/luc

ratio at various concentrations of rluc and inactivated

rluc (cut with restriction enzymes), with luc held

constant, allows a determination of the apparent

concentration of the treated pRL-CMV plasmid.


Differential Scanning Calorimetry Analysis

Salmon sperm DNA (Sigma Chemical Company, St. Louis,

MO), was lyophilized in the presence of lyoprotectant (1:1

w/w). Aliquots of 12-15 mg DNA were then placed in crimped

sample pans and the melting profiles determined using a

Seiko 220C Differential Scanning Calorimeter (Tokyo,

Japan). Scans were carried out over a -100C to 1500C

temperature range at a heating rate of 50 per minute.

Nitrogen was used as a purge gas at a flow rate of 100

ml/min. Only a single scan could be accomplished per






42


sample, due to thermal degradation of the DNA/carbohydrate

mixtures. Onset of melt was determined using the

manufacturers software.


DMED Assay

Treated pRL-CMV DNA was incubated with N,N'-

dimethylethylenediamine (100 mM, pH 7.4, 30 minutes at

370C), cleaving abasic sites via p-elimination (McHugh, and

Knowland 1995). The resultant DNA was then visualized

using a 0.8% agarose gel in TBE buffer (50 mM Tris, 50 mM

borate, 1 mM EDTA, pH 7.8). The gel was electrophoresed at

1 V/cm for 16 hours to affect good separation between the

supercoiled, relaxed circular and linear forms of the

plasmid DNA. The gel was then stained with ethidium

bromide (0.5 gg/ml) and photographed on a UV light box using

a DC40 digital camera (Kodak, Rochester, NY).


Analysis of Hyperchromic Effect

Hyperchromic effect was examined using a UV/Vis

spectrophotometer at 260 nm (Lambda 3, Perkin Elmer).

Analysis consisted of comparing the change in absorbance at

260 nm between lyophilized and non-lyophilized samples at

the same 1 mg/ml concentration of plasmid DNA.





43


Oxidative Analysis

Oxidative analysis of plasmid DNA was determined using

an isocratic HPLC method. Plasmid DNA, 100 gg, was first

chemically hydrolyzed in 1 ml of 60% formic acid at 1500C

under vacuum for 45 minutes. Samples were then lyophilized

to remove all formic acid and rehydrated in 2 ml of mobile

phase (50 mM sodium acetate, 1 mM EDTA, 2% methanol, pH

5.5). The samples were then analyzed by HPLC. The HPLC

conditions consisted of a 50 il loop, a 0.8 ml/min flow

rate, and a 270C column temperature. The column used was a

reverse phase C18 column, 25 cm, 5 pm (Microsorb MV, Rainin,

Woburn MA), with a C18 guard column. Detection was

accomplished using a BAS UV-8 detector (West Lafayette, IN)

at 254 nm for non-oxidized bases and a BAS LC-4B

amperometric detector with a range of 10 nA at 750 mV,

using a Ag/AgCl electrode with a glassy carbon working

electrode, for oxidized bases.

Control oxidized DNA was prepared using a Fenton type

iron-catalyzed reaction. Ferric chloride (1 nm) and

hydrogen peroxide (3%) were incubated with pRL-CMV plasmid

(final concentration 1 gg/pl) on ice for 15 minutes. The

plasmid was then purified by ultrafiltration using a 1000

MWCO spin filter (Amicon).






44


Circular Dichroism Analysis

Circular dichroism was analyzed using a Jacso J-500C

spectrapolarimeter (Easton, MD). Scans were performed

using a 1 nm bandwidth, automatic slit-width, 50 mdeg/fs

sensitivity, at 20 nm/min, over a 300-250 nm wavelength

range. All scans were baseline subtracted against DI water

and each data point was reported as the average of at least

two scans. Plasmid DNA conformational change was observed

at 270 nm. Samples were analyzed at 0.5 mg/ml pRL-CMV

plasmid DNA in a 1 cm cuvette, comparing various

weight/weight ratios of pRL-CMV plasmid DNA to

lyoprotectant. Lyophilized samples were rehydrated

immediately before analysis.

Statistical Analysis

Statistical analysis between the various treatments

was conducted using analysis of variance and Fisher's

protected least significant difference (PLSD) post-hoc T-

tests where appropriate (StatView v4.5, Abacus Concepts,

Berkley, CA), with p<0.05 considered statistically

significant.


Results and Discussion

A biofunctional assay was devised to detect damage due

to lyophilization of plasmid DNA. This method is a






45


variation on the conventional cell transfection method,

using two plasmids rather than one. By cotransfecting both

control and treated plasmid the variation in transfection

efficiencies can be taken into account. The treatment

plasmid was the pRL-CMV construct (Promega, Madison, WI),

containing the renilla luciferase gene driven by a

cytomegalovirus promoter. This version of luciferase was

isolated from the sea pansy, a species of bioluminescent

coral. The control plasmid was the pGL3 construct

(Promega). The construct encodes the photinus luciferase

gene driven by a simian virus promoter. The two versions

of luciferase each require different cofactors and can be

assayed in the same test tube, since the conditions

appropriate for the renilla luciferase reaction quench the

photinus luciferase reaction. The two plasmid constructs

were transfected into COS1 cells, an African green monkey

kidney cell line, that expresses the large T antigen

transcription factor needed to drive the SV40 promoter.

This transcription factor was capable of upregulating the

CMV promoter, as well (Soneoka et al. 1995). The

transfections were liposome mediated with the optimal ratio

of lipid to pRL-CMV to pGL3 for detection as suggested by

the manufacturer. This ratio was empirically based upon

the linearity of concentration to activity standard curves






46



(Figure 3) and the need to prevent cross talk between the


different promoters present in the two vectors. The cells


were transfected in serum free media and were incubated for


,0.7- ---- 0.7
0.6- 0.6
0.5 0.5
E0.4 E 0.4
< 0.3 40.3
z *
S0.2 0 02
0.1- 0.1
0 0
Cntrd 4 2 1 0.1 Cntd 4 2 1 0.1
Glucose:DNA (wlw) Glucose-l-Phosphate:DNA (wtw)



0.7 0.7-
0.6 0.60

E 0.5 2 0.4
E 0.4- E 0.4 -
0.3 0.3
S0.2 00.2 *
0.1- M 0 0.1

Cntri 4 2 1 0.1 Cntr 4 2 1 0.1
Glucosamine:DNA (wlw) Sucrose:DNA (wfw)



0.7- 0.7
0.6 0.6



0.5 0.1
0o.l I0 i I A

0 I .0.0
Cntrl 4 2 1 0.1 Cntr 4 2 1 0.1
Lactose:DNA (wlw) Urea:DNA (w/w)

Figure 8: Effect of lyophilization (FD) on plasmid DNA
activity, with various amounts lyoprotectant to DNA (w/w).
Average + SEM. n=5 for all treatments, = FD DNA
significantly lower than Control DNA. p<0.05 via Schefe's
multiple comparison T-test.


5 hours. The media was then replaced with serum containing


media and incubated a further 48 hours before being lysed.






47


Both forms of luciferase were then assayed using a

luminescence spectrophotometer (Monolight 2010).

The biofunctionality assay demonstrated a loss of more

than 75% of plasmid DNA activity after lyophilization

(Figure 8). Furthermore, this loss of activity could be

prevented by the use of lyoprotectant carbohydrates

(approximately 4:1 carbohydrate to DNA w/w), including

glucose, glucose-1-phosphate, glucosamine, lactose and

sucrose). These carbohydrates were chosen to study the

effect of charge and size on the effectiveness of the

lyoprotection.

The presence of lyoprotectant did not significantly

alter the apparent plasmid DNA concentration either before

(data not shown) or after freeze-drying. The protective

effect of the glucose-1-phosphate was unexpected if the

lyoprotective effect was due to the replacement of the

sphere of hydration by the carbohydrate, as it might be

assumed that this anionic carbohydrate would be

electrostatically repulsed from the major groove by the

anionic phosphate backbone. Furthermore, glucosamine,

which would be expected to interact strongly with plasmid

DNA, did not exert a statistically significant protective

effect at a 4:1 w/w ratio. If the lyoprotectant

carbohydrates are protective of plasmid DNA due to






48


replacement of the water sphere of hydration, then the use

of a water destabilizing compound such as urea should

decrease DNA activity. However, as seen in Figure 8, urea

did support a protective effect through an undetermined

mechanism.

Disaccharides demonstrated a greater protective effect

on a molar basis than did monosaccharides, suggesting that

the size of the protective compound may exert and

influence. Potentially, the effect could be due to a non-

interactive covering of the plasmid DNA that inhibits any

conformationally induced denaturation. This hypothesis is

supported by the protective effect seen upon "saturating"

concentrations of all lyoprotectants tested (4000:1 w/w).

The extent of interaction between DNA and the

lyoprotectant carbohydrates was assessed using thermal

analysis to determine onset of melting temperatures.

First, behavior in the solid phase was examined using

differential scanning calorimetry, a sample DSC scan can be

seen in Figure 9. Equal weights of carbohydrate and

salmon-sperm DNA were mixed and lyophilized. The

lyophilized cake was then assayed to determine the onset of

melting. Second, the effect of lyophilization upon melting

temperature after rehydration was analyzed using UV

spectroscopy. In the dried state, there was a general






49


trend towards increased onset of melting temperature in the

presence of lyoprotectants Figure 10, with a change in the

onset of melting suggesting an interaction between the

carbohydrates and the DNA. However, glucose-l-phosphate

did not show a significant increase in Tm, but did protect

against loss of bioactivity, which may suggest that a

physical interaction is not necessary for a protective

effect.


5





51.4 C -
12.68 iln
-3. O0 -




-1



-10
-7.S\ -10





1G-I .l5 -15
-11.7 5.8 23.3 40.8 58.3 75.8 93.3 110.8 128.3 145.8
University of Flortil-MSE TEMP C (Heating)



Figure 9: Representative thermal analysis scan of salmon
sperm DNA.


The effect of rehydration of lyophilized plasmid DNA

on the DNA melting profile, as measured via absorbance at

260 nm, can be seen in Figure 11. Lyophilized-rehydrated






50


plasmid DNA showed an earlier onset of melting than plasmid

DNA that was kept in solution. This appears to agree with

the lower onset of melting seen in the solid state without

lyoprotection. It can be seen that the previously

lyophilized DNA solutions showed an increased absorption

over the 25-800C interval, indicative of the presence of a

small amount of single-stranded DNA in the DNA solutions

(Lindhal and Nyberg 1972).

80


60


E 40


20


0





Figure 10: Effect of lyoprotectants on the melting
temperature of salmon-sperm DNA. Mean + SEM, n=3, =
significantly different than unprotected control DNA (p <
0.05) via Fisher's PLSD.

To identify a potential mechanistic explanation for

the results obtained by the biofunctionality assay, several

structural assays were also utilized. Standard agarose gel

electrophoresis (0.8% agarose) is capable of separating the

conformations of plasmid such as supercoiled, relaxed






51


circular and linear, by utilizing size fractionation in the

agarose gel via an electric field. The DNA is negatively

charged because of its phosphate backbone, and will migrate

toward the cathode when an electric field is applied. No

apparent change in the ratio of supercoiled to relaxed

plasmid DNA was observed using this method, Figure 12 and

Figure 13, suggesting that the lyophilization process did

not cause any gross conformational changes to the plasmid

DNA.

0.6

0.5

0.4
v i-FD
N 0.3
.3 J-g-Sin

0.2

0.1

0.0
o.o -M fs 9 '-------- -
25.0 35.0 45.0 55.0 65.0 75.0 85.0
Temperature (C)
Figure 11: Effect of lyophilization and subsequent
rehydration on the melting of plasmid DNA. Mean + SEM, n=3

The standard agarose gel method can not differentiate

between undamaged supercoiled and damaged supercoiled

plasmid DNA. This damaged supercoiled plasmid DNA could

potentially be seen after glycosidic bond cleavage at






52




















Figure 12: Lyophilized plasmid DNA samples, rehydrated in
DI water and run an agarose gel. Samples: marker, HindII
digested lambda phage marker; FD, lyophilized DNA without
DMED treatment; FD DMED, lyophilized DNA with DMED
treatment; pH 3, DNA incubated in pH 3.0, 2.5 mM citrate
buffer for 15 minutes at 500C with no DMED treatment; pH 3
DMED, DNA incubated in pH 3.0, 2.5 mM citrate buffer for 15
minutes at 500C with DMED treatment before running on gel.

abasic sites. Abasic sites are one of the most common DNA

lesions and can be produced in at least two ways (McHugh,

and Knowland 1995): via spontaneous hydrolysis of the

glycosyl bond between deoxyribose and the purines, or at a

slower rate between deoxyribose and pyrimidines. These

modifications can be accelerated by base modifications such

as the alkylation of purines, saturation of C5-C6 bond of

the pyrimidines, and fractionation of the heterocyclic ring

(Talpaert-Borle, and Liuzzi 1983). In addition, they can

be detected by the use of the DMED assay, which cleaves the






53


phosphate backbone by a beta elimination reaction in the

presence of an abasic site (Figure 2 on page 7).

100%


80%


60%
o
S 40%


20%


0%
0 1 2 3
Lyophilization Cycles
Figure 13: Effect of lyophilization on plasmid form. Mean +
SEM, n=3. p > 0.05 via ANOVA

After incubation in the presence of DMED damaged DNA

would be expected to show a lower percentage of supercoiled

DNA. The percentage of supercoiled DNA did not change

after lyophilization and subsequent DMED treatment (Figure

12). This suggests that the lyophilization process did not

cause an increase in plasmid DNA abasic sites.

Analysis by UV/Vis spectroscopy showed a hyperchromic

effect for lyophilized plasmid DNA, suggesting potential

denaturation or conformational change. Any process which

increases the interaction of purine and pyrimidine rings,

such as a contraction of the plasmid DNA macromolecule or






54


restriction of internucleotide rotation by the formation of

hydrogen bond stabilized helical structures, would result

in a hyperchromic effect at 260 nm, as would separation of

the two anti-parallel strands (denaturation) (Michelson,

1963). Lyophilization of plasmid DNA resulted in a

significant increase in absorbance at 260 nm. This effect

can be alleviated by the use of lyoprotectant carbohydrates

at a 50 mM concentration (Figure 14). These results

suggested that the lyophilization process causes a

conformational change or denaturation in plasmid DNA that

can be prevented by the use of carbohydrates as

lyoprotectants, since these data agree with the UV melting

profile which also suggests a denaturation.

20


O 15
0
x
E
o 10






0




Figure 14: Effect of lyophilization on plasmid DNA
absorbance at 260 nm. Average + SEM. n=4, FD DNA
significantly higher than lyoprotected DNA p<0.05 via ANOVA






55


To further identify possible conformational changes

circular dichroism analysis was utilized. Lyophilized

plasmid DNA was compared to "normal" B-form DNA (pH 7.4 TE

buffer) and to high-salt, A-form DNA (5 M NaC1) (Bailleal

et al. 1984; Nishimura et al. 1985). As can be seen in

Figure 15, the high salt conditions seen during

lyophilization cause a conformational change that may be

indicative of a change from the B-form to the A-form.



6

?5 FD

4 A




1

0
250 260 270 280 290 300
Wavelength (nm)


Figure 15: Wavelength circular dichroism scan of
lyophilized plasmid DNA (FD) compared to A-form and B-form
plasmid DNA. Each scan is the average of three separate
baseline subtracted scans.

The use of lyoprotectants can exert further

conformational changes during the lyophilization process

Figure 16. The smallest uncharged protectant, glucose,

appeared to undergo the least conformational change during

the lyophilization process. Increasing the size of the





56

protective molecule, by utilizing lactose as a protectant

causes a slight decrease in conformational shift, but not

nearly as great a shift as the smaller glucose monomer.
When charged moieties were added to the glucose protectant,

a further shift in conformation was noted. Both anionic

and cationic compounds caused a shift in conformation

a 3.0 3.0
2.5- A g 25 B
2. I I 20I


tI 1 111 1. 1 I
to is






0s5 05 Df





Figure 16: Effect of lyoprotection on plasmid DNA
conformational change. Panels each signify a separate
protectant; A: lactose, B: glucosamine, C: glucose-1-
phosphate, and D: glucose. For each panel Protectant:DNA
1, 5, 10, 20 and 40 or 1, 2, and 8 w/w ratios. Cntrl S
"protectant" is protectant:DNA solution (40:1 w/w) non
lyophilized. Cntrl FD and Cntrl S "without protectant" are
no protectant, lyophilzed or non lyophilized respectively.
n=10, mean + SEM.

towards the A-form, with the anionic glucose-1-phosphate
having a two-fold increase in CD shift over the cationic
glucosamine, causing a conformational change even in the
glucosamine, causing a conformational change even in the





57


solution control state. Furthermore, there is a lack of a

clear dose-response relationship between the amount of the

carbohydrate and the conformational shift for all compounds

except glucose-l-phosphate. It seems that the anionic

glucose-l-phosphate requires a 5:1 (w/w) protectant:DNA

ratio before the protectant exerts a greater conformational

change upon the DNA. This increase in CD measurement is

typical of a change in both the helix winding angle and the

base pair twist of the plasmid DNA. An increase in the CD

measurement signals a shift towards the A-form of the

plasmid DNA, with an a decreased winding angle, angle

between two adjacent bases, leading to more base pairs per

turn of the double helix (Johnson et al. 1981).

The kinetics of the conformational change were

examined using CD. Essentially, DNA was lyophilized and

consequently rehydrated. CD spectra were recorded for

multiple time points after rehydration (Figure 17). A

first order fit of the data suggests that most of the

plasmid returns to the B-form in approximately five hours.

At 24 hours after rehydration the measured CD was not

statistically different from control non-lyophilized DNA

(data not shown), this suggests that the conformation of

the plasmid DNA may return to the B-form within 24 hours.






58


It was hypothesized that this initial conformational

change could be responsible for the apparent loss of

plasmid DNA activity after lyophilization. Since it was

shown that the plasmid returns to the B-form in a

relatively short period of time, plasmid DNA was

lyophilized and rehydrated immediately before transfection

or rehydrated 24 hours before transfection to determine the

A-form effect on plasmid DNA activity (Figure 18). There

was no significant difference between 0 and 24 hours

rehydration, suggesting that this potential conformation of

the plasmid DNA does not have a direct impact on biological

activity.


0.8
0.7
0 0.6
0.5 -
0.4
0.3 -
S0.2

0.1
0.0 ,
0.00 0.08 0.16 0.25 0.50 1.00 24.00
Hours Rehydrated



Figure 17: Kinetics of plasmid conformational change after
rehydration. Mean + SEM, n=3.






59


Another mechanism of damage may be oxidative damage of

the plasmid DNA during the lyophilization process. To

examine the possibility of this mechanism, plasmid DNA was

lyophilized up to six times and then chemically hydrolyzed

to the individual bases. The bases were then analyzed by


0.7
0.6
0.5 -
c 0.4
E
< 0.3 -
0.2
0.1 -
0.0
0 24 Cntrl
Rehydration Tine (Hours)



Figure 18: Effect of rehydration time on plasmid DNA
activity. Mean + SEM. n=6. No significant difference
between 0 and 24 hours by T-test.

HPLC using parallel UV and ECD detection to quantitate both

normal and oxidized bases, specifically guanine and 8-OH-

guanine. The 8-OH-guanine is a commonly used marker for

oxidative damage of DNA. The lyophilized samples showed no

significant increase in 8-OH-guanine per mole of guanine

and all lyophilized samples were significantly less damaged

than the positive control (Figure 20).

The type of tube used for lyophilization of plasmid

DNA could potentially change the biological activity of





60


plasmid DNA (Figure 21). If the loss of biological

activity was due to the interaction of plasmid DNA and the

container used for lyophilization, then the use of





A B





\l







Figure 19: HPLC analysis of hydrolyzed pRL-CMV plasmid DNA.
Panel A is a UV analysis. Peaks by retention time: 2.76,
solvent front, 3.79 cytosine, 7.85 guanine, 11.15 thymine,
13.00 unidentified, 17.89 adenine. Panel B is an ECD
analysis. Peaks by retention time 2,67 3.21 solvent
front, 3.53 5-OH cytosine, 3.82 cytosine, 4.78 plasticizer,
7.35 guanine, 7.92 8-OH guanine, 16.99 8-OH adenine and
adenine.

siliconized tubes would be expected to decrease DNA loss.

However, there is no significant difference in DNA activity

between siliconized and non-siliconized polypropylene

tubes, suggesting that plasmid DNA does not interact with

the polypropylene tube during the lyophilization process.

This is supported by gel electrophoresis, which showed no

significant loss of plasmid DNA after cycles of






61


140

0 120

E 100
C
o 80



040
o*
E 20 *


0 2 4 6 Cntrl
Freeze-drying Cycles

Figure 20: Effect of lyophilization of plasmid DNA on
oxidative damage. Positive control was Fe+3 catalyzed
oxidation. Mean + SEM, n=5, except control n=3. = p <
0.05 via Fisher's PLSD versus control.

lyophilization. There did appear to be a significant

interaction between polyethylene and plasmid DNA during the

lyophilization process, suggesting that polyethylene tubes

are not an appropriate choice for storage of plasmid DNA.


Conclusion

In conclusion, lyophilization of plasmid DNA causes a

loss of plasmid DNA functionality that can be prevented by

the use of carbohydrates as lyoprotectants. The mechanism

behind this loss of activity does not appear to be due to a

gross structural change, as would be evidenced via agarose

gel electrophoresis. Nor does the damage appear to be due

to an increase in plasmid DNA abasic sites, as measured by





62


the DMED assay, increase in oxidative damage, as measured

via HPLC, or interaction with the enclosure used for


E 0.7-

-0.6

0 0.5
< 0.4
S*
P.0.3

0.2
"X
10.1


Cntri PP SP PE
Figure 21: Effect of tube type on plasmid DNA bioactivity
after lyophilization. Cntrl = non-lyophilized DNA, PP =
polypropylene, SP = siliconized polypropylene, PE =
polyethylene. Mean + SEM, n=5, = p < 0.05 via Fisher's
PLSD versus control, X = p < 0.05 versus PP and SP.

lyophilization. However, this change in plasmid DNA

activity does correlate to a change in plasmid hyperchromic

effect, as evidenced by a change in absorbance at 260 nm,

and a change in the melting temperature, as evidenced by

DSC and UV measurements. These data suggest that plasmid

DNA damage after lyophilization is mediated by some sort of

conformational change but the conformational of the plasmid

DNA does not directly affect plasmid activity, as suggested

by circular dichroism spectra. This conformational change

is not prevented by all carbohydrates tested, but all





63


protectants were effective against the hyperchromic effect.

This suggests that lyoprotectants may decrease a

conformationally-induced denaturation of the plasmid DNA

caused by the lyophilization process.















CHAPTER 4
CHARACTERIZATION OF ENDOTOXIN AND CATIONIC LIPOSOME
INTERACTION


Introduction

Gene therapy is one of the fastest growing areas in

therapeutics. While rapid progress has occurred in this

field, the pharmaceutical community has not adequately

addressed safety concerns. In particular, there exists a

potential toxicity from plasmid DNA contaminates that could

be exacerbated by conventional non-viral gene delivery

methods. Since plasmid DNA is typically produced by gram-

negative bacteria E. coli, endogenous bacterial products

can potentially contaminate the final plasmid DNA product

(Weber et al. 1995). Typical problematic agents in plasmid

DNA preparations include endotoxin (also known as

lipopolysaccharide or LPS), a cell wall component that is

pyrogenic in man, and DNase, a ubiquitous bacterial enzyme

that can degrade the final plasmid product and act as an

immunogen.

Endotoxin ranks among the most serious limitations in

the development of gene products for gene therapy because



64





65


of its extreme toxicity. E. coli endotoxin consists of a

polysaccharide component and a covalently bound lipid

component, lipid A. Lipid A is the biologically active

component responsible for pathophysiological effects

including fever, hypotension, intravascular coagulation and

potentially death (Fletcher, and Ramwell 1980; Aida, and

Pabst 1990; Rietschel et al. 1993; Xing et al. 1994).

Endotoxin is an extremely potent toxin with an LD5o of 3

mg/kg and 1 mg/kg in rats and dogs, respectively (Fletcher,

and Ramwell 1980; Shibayama et al. 1991). It has been

shown that as little as 150 pg of endotoxin can be lethal to

a horse (Bottoms 1982). Therefore, the Food and Drug

Administration (FDA) has established a guidance on human

maximal endotoxin dose permissible for parenteral products

(F.D.A 1985). This limit is based on endotoxin activity

(EU), and can be measured via the LAL (Limulus amebocyte

lysate) assay (Levin, and Bang 1964; Levin, and Bang 1968;

Iwanaga 1993). Non-intrathecal parenteral drug products

are limited to 5 EU kg-1 hr-1. In the case of an intrathecal

parenteral product, the injection limit drops to 0.2 EU kg-1

hr-1. These allowable levels are of concern in gene

therapy, where the extent of endotoxin contamination in

plasmid DNA preparations after treatment to remove

endotoxin has been shown to range from 3-15 EU/mg plasmid





66


DNA, with 1 EU = 1 ng pure endotoxin (Cotton et al. 1994).

Determinations in our laboratory have shown that before

specific removal, endotoxin contamination of plasmid DNA

can typically approach 5,000 EU/mg, while the literature

reports endotoxin concentrations exceeding 15,000 EU/mg DNA

(Montbriand, and Malone 1996).

It was previously shown that increasing levels of

endotoxin contamination decrease transfection efficiency

when using DOTAP : DOPE (dioleoyl glycero trimethylammonium

propane : dioleoyl glycero phosphoethanolamine) cationic

liposomes (Weber et al. 1995) or adenovirus particles

(Cotton et al. 1994) to deliver plasmid DNA. However, the

mechanism underlying this decreased transfection efficiency

has not been clearly defined. It can be predicted that

endotoxin interacts with cationic liposomes

electrostatically due to the positive charges on the

cationic liposomes and the negatively charged

phosphorylated glucosamine residues of the lipid A moiety

(Rietschel et al. 1993). There also may be other

lipophilic methods of interaction with cationic lipids due

to the saturated fatty acid chains on lipid A (Rietschel et

al. 1993). Upon this interaction of endotoxin, DNA and

cationic liposomes, the liposome complex would be expected

to be taken into the cell via endocytosis with subsequent






67


destabilization of the endosome and release of the DNA and

endotoxin into the cytoplasm (Xu, and Szoka 1996).

It will be of importance for formulation scientists to

understand the potential modes of interaction between

endotoxin and cationic liposomes. With this knowledge it

should be possible to develop newer methods of endotoxin

removal from these products. In this report we have

focused on endotoxin, similar concerns could be raised with

other anionic compounds.


Materials and Methods


Plasmid Purification

A pGL3 plasmid (Promega, Madison, WI) encoding for

luciferase driven by the SV-40 promoter was propagated in

E. coli JM109 cells (Promega, Madison, WI). The

transformed bacteria were cultivated in Lura Bertina (LB)

broth containing 100 gg/ml ampicillin to select for 0-

lactamase encoding plasmid. The pGL3 plasmid was isolated

via an alkaline lysis method and purified using an anionic

exchange column (Qiagen, Chatsworth, CA). The plasmid was

stored in TE buffer (Tris HC1, 10 mM; EDTA, 1 mM, pH 7.4).

Endotoxin content of the plasmid DNA before endotoxin

removal was approximately 4,000 EU/mg, as determined by the

LAL assay method (QCL-1000, BioWhittaker, Walkersville,





68


MD). Endotoxin was removed by the Triton X-114 method of

Manthorpe, et al (Manthorpe et al. 1993).


Liposome Preparation

Lipids were obtained from Avanti Polar Lipids

(Alabaster, AL). Cationic DOTAP : DOPE, NBD labeled DOTAP

: DOPE, and zwitterionic lecithin liposomes were prepared

using the hand-shaking method (New 1990). DOTAP : DOPE (10

mg DOTAP, 10 mg DOPE), NBD labeled DOTAP : DOPE (10 mg

DOTAP, 10 mg DOPE, 0.1 mg l-palmitoyl-2-[12-[(7-nitro-2-

1,3-benzodiamino]dodecanoyl]-sn-glycero-3-phosphate), or

lecithin (10 mg) was added to 10 ml of chloroform and

introduced into a 250 ml round bottom flask. Chloroform

was evaporated using a rotary evaporator at 600C. The lipid

film was hydrated with 10 ml of distilled water and shaken

for 30 minutes at 600C. Lipids were sized by extrusion 6

times through 200, 600, or 800 nm polycarbonate filters

(Poretics Corp, Livermore, CA). Sizes were confirmed with

a laser light scattering particle sizer using volume-

weighted distribution (Nicomp 380ZLS, Santa Barbara, CA).


Transfection Efficiency Assay

Phenol purified, unlabeled endotoxin (0-50,000 EU/ml),

E. coli serotype 055:B5 (500 EU/.g, Sigma Chemical Company,






69


St. Louis, MO) was added to a pGL3 luciferase plasmid :

cationic lipid mixture (DOTAP : DOPE 1:1 w/w) with a ratio

of 1 gg plasmid to 2 gg DOTAP. The mixtures were incubated

for 15 minutes to allow interaction of the anionic plasmid

to the cationic liposomes, followed by transfection into

COS-1 cells (ATCC, Rockville, MD) that had been plated 24

hours earlier at 3x104 cells/ml in 24 well tissue-culture

dishes. All transfections were carried out using serum

free DMEM (Dulbecco's Modified Eagle's Medium) supplemented

with penicillin G (100 units/ml) and streptomycin (100

gg/ml). They were subsequently incubated for 5 hours in a

humidified, 5% CO2 incubator at 370C. The media was

aspirated and the cells washed with PBS (phosphate buffered

saline) before being replaced by 10% FBS (fetal bovine

serum) containing DMEM and antibiotics. The cells were

incubated for an additional 19 hours, washed with 1 ml PBS,

and then lysed with lysis buffer (200 gL, 0.1 M potassium

phosphate, 1% Triton X-100, 1 mM DTT, 2 mM EDTA, pH 7.8).

Luciferase activity was determined using a

luminescence spectrophotometer (Monolight 2010, Analytical

Luminescence Laboratory, San Diego, CA). Lysate (20 iL) and

the assay buffer (100 iL, 30 mM tricine, 3 mM ATP, 15 mM

MgSO4, 10 mM DTT, pH 7.8) were added to the sample cuvette.






70


D-luciferin (100 pL, 1 mM, pH 6.5) was injected to initiate

the reaction. The light output was measured for 10 seconds

and the results integrated to yield the luciferase

activity. All luciferase measurements were standardized

against protein concentration as measured by the

bicinchoninic acid protein assay (BCA Protein Assay,

Pierce, Rockford, IL).


Anisotropy Assay

A constant amount (1000 EU) of flourescein

isothiocyanate (FITC) labeled endotoxin, E. coli serotype

055:B5 (100 EU/pg, Sigma Chemical Company, St. Louis, MO),

and varying amounts of unlabeled cationic DOTAP : DOPE

liposomes (600 nm, 0-16 ig), zwitterionic lecithin liposomes

(200 and 800 nm, 0-16 pg,), PAMAM (polyamidoamine)

dendrimer (generation 4, 0-16 ig, in TE buffer, pH 7.4,

Aldrich Chemical, Milwaukee, WI), or plasmid DNA (pGL3 in

DI water), with a final volume of 1 ml, were allowed to

equilibrate together for 15 minutes at 250C. Fluorescent

anisotropic measurements were conducted at 250C, 487 nm and

525 nm excitation and emission wavelengths, respectively

using a luminescence spectrophotometer (LS50B, Perkin

Elmer, Oak Brook, IL).






71


Endotoxin Dephosphorylation

Unlabeled endotoxin (30 EU) was incubated for various

time from 0 to 120 minutes with 1 mU alkaline phosphatase

(Promega, Madison, WI) at 370C following the methods of

Poelstra, et al (Poelstra et al. 1997a; Poelstra et al.

1997b) with minor variation in the assay buffer: 0.5 M

Tris, 10 mM MgCl2, 1 mM ZnCl2, pH 7.8.


Endotoxin Assay

Endotoxin activity was quantitatively measured using a

commercially available chromogenic LAL test method (QCL-

1000, BioWhittaker, Walkersville, MD) according to the

manufacture's instructions. In brief, endotoxin (0.6 EU)

was incubated for 15 minutes with an increasing

concentration of DOTAP: DOPE cationic liposomes (0-100 ng)

and brought to a final volume of 50 gL with endotoxin free

water. The LAL reagent (50 pL) was added to the reaction

vessel and the mixture was further incubated for 10 minutes

at 370C. Chromogenic substrate solution (100 gL) was added

and the sample incubated for 6 minutes before the addition

of stop reagent (100 RL, 25% glacial acetic acid in water).

Absorbance of the p-nitroanaline product was read at 405 nm

using a UV/Vis spectrophotometer (Lambda 3, Perkin Elmer,

Oak Brook, IL)






72


Cell Viability Assay

The MTT (dimethylthiazol diphenyltetrazolium bromide)

assay (Freshney 1994) was used to assess the effect of the

endotoxin and liposome delivery systems on cell viability.

COS-1 cells were plated in a 24 well tissue-culture plate

at a density of 3x104 cells/well in 1 ml DMEM media

containing 10% FBS and incubated for 12 hours in a 370C,

humidified, 5% CO2, incubator. The serum containing media

was aspirated, washed with PBS, and replaced with serum

free media. Varying amounts of lipid (0-170 gg), endotoxin

(0-50,000 EU), and lipid: endotoxin combinations (5 pg : 0-

50,000 EU) were added to each well and incubated for 5

hours. The media was then changed back to 10% FBS DMEM and

the incubation continued until 24 hours following the

addition of test formulations. The cells were then fed

with 1 ml of fresh media and MTT (250 pL, 5 mg/ml) and were

incubated for 5 hours and then the media was removed.

Dimethylsulfoxide (DMSO, 1 ml) and glycine buffer (250 pL,

0.1 M glycine, 0.1 M NaC1, pH 10.5) were added and the

absorbance was immediately read at 570 nm. Untreated cells

and buffer alone were used as positive and negative

controls, respectively.






73


Statistical Analysis

Statistical analysis between the various treatments

was conducted using analysis of variance and Fisher's

protected least significant difference (PLSD) post-hoc T-

tests where appropriate (StatView v4.5, Abacus Concepts,

Berkley, CA), with p<0.05 considered statistically

significant.


Results

The effects of endotoxin contamination upon

transfection efficiency were determined by adding increased

levels of endotoxin to pGL3 luciferase plasmid : cationic

lipid mixture (DOTAP : DOPE 1:1 w/w) with a ratio of 1 pg

plasmid to 2 pg DOTAP before transfection into COS-1 cells

(Figure 22). A 1:2 ratio of plasmid to cationic lipid

results in a neutral to slightly negative net charge for

the overall complex. In the presence of DOTAP : DOPE

cationic liposomes (2 pg/ml), low levels of endotoxin (50

EU/ml) appeared to increase the variability of luciferase

reporter activity, a marker of the plasmid DNA

functionality. Higher concentrations of endotoxin resulted

in a corresponding decrease in the luciferase reporter

activity. Activity decreased more than 90% at 5000 EU/ml






74


endotoxin, however the cells appeared viable under direct

observation.

50

40

3c 30
-D Liposome
t-- No Liposome
o)20



0

0 50 500 5,000 50,000
Endotoxin (EU/mL)
Figure 22: Enzyme activity corrected for total cellular
protein after transfection of luciferase plasmid (lpg) in
the presence of endotoxin, with and without DOTAP : DOPE
cationic liposomes. with lipid (2 ug/ml), U without
lipid. RLU+SEM, n=4, p<0.05 via one way ANOVA for effect of
endotoxin in the presence of cationic lipid.

The extent of interaction between endotoxin and

cationic liposomes (DOTAP : DOPE) was determined using

fluorescence anisotropy. The potential contribution of

electrostatic versus lipophilic interactions was

investigated by varying the ionic strength of the

incubation solution (NaC1, 0-2 M). There was a correlative

increase in the anisotropic measurement when endotoxin was

incubated with additional cationic lipid, suggesting

formation of a complex (Figure 23). Maximum binding

occurred at a 1:2 (w/w) ratio of total lipid to endotoxin.






75


The interaction decreased with increasing ionic strength,

but the size of the complex increased significantly, as

monitored by change in anisotropy, even under high ionic

strengths (2 M NaC1), suggesting the presence of

electrostatic in addition to other interactions. Free FITC

was used as a control to ensure the interaction detected

was due to endotoxin and not a consequence of the FITC

label (Figure 23 inset). Furthermore, the spectral

properties of the free FITC and the conjugated FITC appear

similar as determined by excitation and emission maxima.

The potential for competition between plasmid DNA and

endotoxin for DOTAP : DOPE cationic liposome was compared

using a similar experimental paradigm. DOTAP : DOPE (5 gg)

and FITC conjugated endotoxin (1000 EU) were held constant

at the maximal binding ratio seen in Figure 23. As plasmid

DNA levels were increased, there was a decrease in

anisotropy readings, suggesting increased competition

between endotoxin and plasmid DNA for interaction with

DOTAP : DOPE. This results in the displacement of

endotoxin (Figure 24).

To further characterize the mechanism of endotoxin and

cationic liposome interaction, additional fluorescence

anisotropy studies were carried out. Two sizes of PAMAM

dendrimers, a cationic cascade polymer, were studied. The






76


0.12

0.12
0.10- 0.10 FRe Frrc
CL>. o.o0
I 0.06
0.04
0.08- 0.02
o.oo -- DI Water
C. 0 4 8 12 16 -w 0.5 M NaCI

S0.06 -- 2 M NaCI





on anisotropy (r) DI Water, 05 NaC, A 1.0 M NaCI
X 2.0 M NaC. Inset: Free FITC (22 ng 1.0 M NaC. ChangeaCI
0.024



0.002

0in rS2 4 6 8 10 12 14 16
DOTAP:DOPE (PAg)

Figure 23: Effect of increasing DOTAP : DOPE liposomes
concentration with FITC-endotoxin held constant (1000 EU)
on anisotropy (r). DI Water, N 0.5 M NaCl, A 1.0 M NaCl,
X 2.0 M NaCl. Inset: Free FITC (22 ng) 1.0 M NaCl. Change
in r+SEM, n=3, p<0.05 using two way ANOVA for increase in r
with increasing lipid under all conditions.


dendrimers would be expected to interact with FITC labeled

endotoxin though electrostatic forces since the dendrimers

only exposed functional groups are amines. Also, two sizes

of zwitterionic lecithin liposomes (0.2 and 0.8 pm) were

examined to determine the importance of the other

interactions between the lipids. The greatest change in

anisotropy occurred with the cationic PAMAM dendrimers

(Figure 25). Smaller, though statistically significant,

changes in anisotropy were seen at the same concentration

of lecithin, suggesting that the prevalent mechanism of






77






0.12


0.10


0.08


0.06


0.04


0.02-


0.00 I I '- I-' I1 I- I *- I-
0 2 4 6 8 10 12 14 16
pDNA (9g)

Figure 24: Effect of increasing luciferase plasmid
concentration on anisotropy (r) with constant FITC-
endotoxin (1000 EU) and DOTAP : DOPE liposomes (5 gg).
Change in r+SEM, n=3, p<0.05 using two way ANOVA for
decrease in r with increasing plasmid.

interaction is electrostatic rather than lipophilic. In

the case of both sizes of dendrimers and liposomes, the

larger sized particles led to a greater change in

anisotropy as expected, due to the additive nature of

anisotropy. Further evidence for interaction between

endotoxin and cationic liposomes was demonstrated using a

gel retardation assay (Figure 26), which confirmed the





78


FITC-endotoxin cationic lipid complex by retarding the

progress of the endotoxin band in an agarose gel.

0.16

0.14

0.12

S0.10
>-+ Dendrimer G4
o. -- Dendrimer G2
008 -- Uposome 0.8pm
S-*- ULposome 0.2tim
< 0.06

0.04

0.00

0 5 10 15
pg added
Figure 25: Effect of increasing lipid or dendrimer
concentration with FITC-endotoxin held constant (1000 EU)
on anisotropy (r). U Dendrimer (Generation 2), Dendrimer
(Generation 4), X lecithin liposomes (0.2 pm), A lecithin
liposomes (0.8 pm), Change in r+SEM, n=3, p<0.05 using two
way ANOVA for increase in r with increasing concentration
under all conditions.


Well


LPS







0 3 6 8 10 12 14 16
DOTAP:DOPE (lg)

Figure 26: Effect of increasing DOTAP : DOPE liposome
concentration with FITC-Endotoxin held constant (10 Vg).






79


The effect of dephosphorylation of endotoxin on

cationic liposome interaction was also examined (Figure

27). Unlabeled endotoxin (30 EU) was incubated for varying

time with calf intestinal alkaline phosphatase. The

dephosphorylated endotoxin was allowed to interact with

fluorescent NBD labeled DOTAP : DOPE cationic lipid (10 gg,

600 nm). Increased incubation time with the alkaline

phosphatase led to a trend towards decreased anisotropy

(p=0.2 via ANOVA). The anisotropy signal seen after 120

minutes of endotoxin incubation was statistically

equivalent to labeled lipid alone. The small change in

anisotropy was due to the label attached to the liposome

and to the high ionic strength of the reaction buffer,

resulting in a small size change upon interaction.

The chromogenic LAL assay was used to determine the

effect of interaction of endotoxin and cationic lipid upon

the toxicity of the lipid A moiety. All determinations

were compared to a standard curve of known endotoxin

concentrations. No endotoxin activity was evident with the

cationic lipid alone. The LAL assay showed no significant

change in the endotoxin activity when free endotoxin was

compared with cationic lipid-endotoxin complex (Figure 28).






80






0.025


0.020


0.015


S0.010 -


0.005


0.000 I I I II
0 20 40 60 80 100 120
Time (min)


Figure 27: Effect of increasing incubation time of alkaline
phosphatase and unlabeled endotoxin (33 EU) on anisotropy
(r) with constant and NBD labeled DOTAP : DOPE liposome (10
pg). Change in r + SEM, n=4, p>0.05 via one way ANOVA for
decrease in anisotropy over time.

Literature reports suggest that cationic lipids are

toxic in vitro and in vivo (Scheule et al. 1997); however

the toxicity of the cationic lipid endotoxin complex is

not known. The MTT assay was used to determine the effect

of the complex on COS-1 cell viability. Using this method,

a dose-response curve was first determined for the DOTAP :

DOPE cationic lipid. Dose-response curves were then

examined using endotoxin alone and endotoxin incubated with






81


relatively nontoxic concentrations of lipid (<5 pg/ml).

While the DOTAP : DOPE

1.0



S0.8
E


S0.6


C
c
o 0.4
0


0.2



0.0 -' 1 i. *j i i. I i ii *I 1'3I3
0 .1 1 10 100
DOTAP:DOPE (ng)
Figure 28: Effect of increasing lipid concentration with
endotoxin held constant (0.6 EU) on endotoxin activity,
EU/ml + SEM, n=3, p>0.05 via one way ANOVA for effect of
cationic lipid on endotoxin activity.

cationic lipid was toxic, no significant loss of COS-1 cell

viability was detected at endotoxin concentrations up to

5000 EU/ml in the presence of lipid and no change in

viability was seen with free endotoxin (Figure 29).


Discussion

The interactions of endotoxin with non-viral gene

delivery systems and the resultant decreases in






82


transfection efficiency are of potential concern in the

success of gene delivery. If a fundamental understanding

of the factors that influence endotoxin-delivery system

interaction is gained, vectors might be developed that

limit the impact of endotoxin in these formulations.

120"
A B
1001

80

60

40

20


1 10 100 5 50 500 5000
DOTAP:DOPE (ng/ml) Endotoxin ( gg/ml)

Figure 29: Effect of increasing cationic lipid (panel A)
and endotoxin (panel B) on COS-1 cell survival via MTT
assay. mean + SEM, n=5, p < 0.05 via ANOVA for DOTAP :
DOPE, p > 0.05 for endotoxin via ANOVA.

Fluorescent anisotropy results demonstrate endotoxin

interacts with DOTAP : DOPE cationic liposomes.

Fluorescent anisotropic measurements are well suited for

observing interactions between molecules since the reporter

fluorescent probe is sensitive to its environmental

conditions. Fluorophores have a defined orientation and

preferentially absorb light that is vectored in that






83


orientation. By excitation using polarized light, it is

possible to selectively excite individual molecules. As

molecules undergo faster rotational diffusion in solution,

the time of fluorescence emission and anisotropy decreases.

As the molecular volume increases with interaction, there

is a correlative decrease in rotational diffusion and thus

an increase in the value of the anisotropic measurement.

This anisotropic methodology has been applied to observing

oligonucleotide hybridization in solution (Murakami et al.

1991) and to examine oligonucleotide-dendrimer interaction

(Poxon et al. 1996).

The structure and physical properties of endotoxin

contribute to the interaction with cationic DOTAP : DOPE

liposomes. For instance, the fatty acid chains present on

the lipid A moiety contributes to the amphipathicity of

endotoxin. Endotoxin's amphipathic nature coupled with its

low solubility causes aqueous preparations of endotoxin to

exist in mainly an aggregated state (Takayama et al. 1995).

The prevalence of the aggregated state results in the

clustering of the negative charge. This cluster of charge

should help to contribute to the electrostatic interaction

between the cationic liposomes and the anionic endotoxin.

The LAL assay provided information on the effects of

the endotoxin-cationic lipid interaction on potential






84


biological activity. The interaction of the endotoxin and

cationic DOTAP liposomes did not alter the activity of the

endotoxin as measured by the LAL assay. This implies

either that the interaction does not hinder the toxic lipid

A moiety or that the complex is not stable.

The MTT assay demonstrated the effects of endotoxin-

cationic lipid complex on COS-1 cell viability. Increasing

levels of endotoxin-cationic lipid complex and free

endotoxin had no significant effect upon COS-1 cell

viability. This lack of selective toxicity in an

immortalized cell line has previously been reported,

including HeLa, Vero, 3T3, K562, WI-38, SV1, TX-4, CHO,

P3U3, and R-393 cells (Epstein et al. 1990, ; Cotton et al.

1994). Immortalized cell lines have shown no change in

growth as determined by doubling time, plating density and

confluent density measurements. Rather, the toxic response

occurs in primary cell cultures cells (Epstein et al. 1990;

Cotton et al. 1994). In any case, the effect of endotoxin

in a more complex in vivo system is extremely toxic and

should not be dismissed.

The mechanism of endotoxin-cationic lipid interaction

is thought to be mainly electrostatic, as evidenced by the

trend to decreased interaction after endotoxin

dephosphorylation and ionic strength experiments. However,






85


other forces may be involved due to the inability of high

screening-ion concentrations to completely inhibit the

interaction phenomena and the ability of zwitterionic

liposomes to interact with endotoxin. It should be noted

that there were a different number of particles with a

different exposed surface area at similar weights of

lecithin liposomes and cationic dendrimers. This fact

could change the rate of endotoxin interaction with each

compound. However, the greater change in anisotropy

observed with the cationic dendrimers, which should undergo

solely electrostatic interaction, is most likely indicative

of a greater electrostatic interaction.

Structural flexibility of the vesicular liposome

versus the more static, solid dendrimer may also have an

impact on the interactions. When endotoxin undergoes an

electrostatic interaction with cationic liposomes, this

interaction can result in lower transfection efficiencies,

as measured by enzymatic expression. Furthermore, this

effect may not solely be due to a difference in the

delivery of the complex as supported by these studies.

Previous studies with transiently transfected CHO cells

show that exposure to endotoxin after transfection results

in a stimulation of the transfected product at 10 ng/ml

endotoxin but decreases production at higher levels






86


(Epstein et al. 1990). A trend towards a bimodal response

can be seen in Figure 22, but was not statistically

significant in this study, resulting only in increased

variability.

These results suggest that the effect of endotoxin on

established cell lines is not through a cytotoxic mechanism

but rather a difference in the delivery of the plasmid-

endotoxin-cationic lipid complex. The inclusion of

endotoxin into the cationic liposome complex may alter the

morphological form and overall net charge of the complex,

thereby altering the delivery and subsequent transgene

expression. The displacement of DNA from cationic liposomes

by endotoxin is most likely responsible for the decreased

transfection efficiency in established cell lines, since

lowered delivered plasmid levels would result in decreased

expression.

From the present study it is clear that several

factors besides the plasmid and delivery vector will

influence the activity of non-viral gene delivery systems.

Endotoxin contamination can potentially impact transfection

efficiency via competition with plasmid DNA for cationic

liposome binding, but this would not be expected with

typical GLP or GMP preparations used in clinical studies

where the endotoxin levels range from 10 1000 EU/mg DNA.






87


At these reduced endotoxin levels, the competition between

endotoxin and DNA for the cationic liposome is at best

marginal. An effect can be seen, with increased

transfection variability, at 50 EU/ml. This is a level of

endotoxin contamination that can occur with small scale

plasmid preparations used for in vitro cell transfections,

and potentially affects the results seen in many in vitro

studies.















CHAPTER 5
FOAM FRACTIONATION AS A METHOD TO SEPARATE ENDOTOXIN FROM
RECOMBINANT BIOTECHNOLOGY PRODUCT


Introduction

Biotechnology based therapy, using products isolated

from E. coli, need to be concerned of a potential for

toxicity from the contaminant endotoxin (Weber et al.

1995). Endotoxin, also known as lipopolysaccharide or LPS,

is a gram negative bacterial cell wall component commonly

co-isolated with plasmid DNA and recombinant proteins. It

consists of a polysaccharide component and a covalently

bound lipid component, lipid A. Lipid A is biologically

active and can cause a number of pathophysiological effects

including fever, hypotension, intravascular coagulation and

death (Fletcher, and Ramwell 1980; Aida, and Pabst 1990;

Rietschel et al. 1993; Xing et al. 1994).

The removal of plasmid DNA endotoxin contamination can

be difficult on several accounts (Cotton et al. 1994).

First, the negative charges associated with lipid A will

cause endotoxin to mimic DNA on anion exchange resins.

Second, the large size of endotoxin molecule aggregates



88