Evaluation of degradation pathways for plasmid DNA in pharmaceutical formulations via accelerated stability studies

Article abstract of DOI:10.1002/(SICI)1520-6017(200001)89:1<76::AID-JPS8>3.0.CO;2-U

The stability of highly purified supercoiled plasmid DNA formulated in simple phosphate or Tris-buffered saline solutions has been characterized to establish the overall degradation processes that occur during storage in aqueous solution. Plasmid DNA stability was monitored during accelerated stability studies (at 50°C) by measurements of supercoiled, open-circle, and linear DNA content, as well as the accumulation of apurinic sites and 8- hydroxydeoxyguanosine residues over time. The effects of formulation pH, demetalation, metal ion chelators, and ethanol (hydroxyl radical scavenger) on the supercoiled content of plasmid DNA during storage at 50°C were also determined. The results indicate that free radical oxidation may be a major degradative process for plasmid DNA in pharmaceutical formulations unless specific measures are taken to control it by the addition of free radical scavengers, specific metal ion chelators, or both. The generation of hydroxyl radicals in phosphate-buffered saline was confirmed by examining the hydroxylation of phenylalanine over time by reverse phase high-performance liquid chromatography. Ethanol was found to enhance plasmid DNA stability and to inhibit the hydroxylation of phenylalanine; both observations are consistent with the known ability of ethanol to serve as a hydroxyl radical scavenger. Moreover, the combination of ethylenediamine tetraacetic acid (EDTA) and ethanol had a synergistic enhancing effect on DNA stability. However, the metal ion chelator diethylenetriaminepentaacetic acid (DTPA) was as potent as the combination of EDTA and ethanol for enhancing the stability of plasmid DNA. By controlling free radical oxidation with EDTA and ethanol, the rate constants of plasmid DNA degradation by means of depurination and β-elimination were then determined, allowing accurate predictions of DNA storage stability as a function of formulation pH and temperature. The ability to predict plasmid DNA storage stability in the absence of free radical oxidation should prove to be a valuable tool for the design of stable pharmaceutical formulations of plasmid DNA. (C) 2000 Wiley-Liss, Inc.

Full text of DOI:10.1002/(SICI)1520-6017(200001)89:1<76::AID-JPS8>3.0.CO;2-U

Evaluation of Degradation Pathways for Plasmid DNA in
Pharmaceutical Formulations via Accelerated
Stability Studies
ROBERT K. EVANS, ZHENG XU, KATHRYN E. BOHANNON, BEI WANG, MARK W. BRUNER, DAVID B. VOLKIN
From the Department of Vaccine Pharmaceutical Research, Merck Research Laboratories, WP78-302, Sumneytown Pike,
West Point, Pennsylvania 19486
Received 11 August 1999; accepted 2 November 1999
ABSTRACT: The stability of highly purified supercoiled plasmid DNA formulated in
simple phosphate or Tris-buffered saline solutions has been characterized to establish
the overall degradation processes that occur during storage in aqueous solution. Plas-
mid DNA stability was monitored during accelerated stability studies (at 50°C) by
measurements of supercoiled, open-circle, and linear DNA content, as well as the ac-
cumulation of apurinic sites and 8-hydroxydeoxyguanosine residues over time. The
effects of formulation pH, demetalation, metal ion chelators, and ethanol (hydroxyl
radical scavenger) on the supercoiled content of plasmid DNA during storage at 50°C
were also determined. The results indicate that free radical oxidation may be a major
degradative process for plasmid DNA in pharmaceutical formulations unless specific
measures are taken to control it by the addition of free radical scavengers, specific
metal ion chelators, or both. The generation of hydroxyl radicals in phosphate-buffered
saline was confirmed by examining the hydroxylation of phenylalanine over time by
reverse phase high-performance liquid chromatography. Ethanol was found to enhance
plasmid DNA stability and to inhibit the hydroxylation of phenylalanine; both obser-
vations are consistent with the known ability of ethanol to serve as a hydroxyl radical
scavenger. Moreover, the combination of ethylenediamine tetraacetic acid (EDTA) and
ethanol had a synergistic enhancing effect on DNA stability. However, the metal ion
chelator diethylenetriaminepentaacetic acid (DTPA) was as potent as the combination
of EDTA and ethanol for enhancing the stability of plasmid DNA. By controlling free
radical oxidation with EDTA and ethanol, the rate constants of plasmid DNA degra-
dation by means of depurination and ␤-elimination were then determined, allowing
accurate predictions of DNA storage stability as a function of formulation pH and
temperature. The ability to predict plasmid DNA storage stability in the absence of free
radical oxidation should prove to be a valuable tool for the design of stable pharma-
ceutical formulations of plasmid DNA. © 2000 Wiley-Liss, Inc. and the American Pharma-
ceutical Association J Pharm Sci 89: 76–87, 2000
INTRODUCTION
tial clinical use in gene therapy and DNA vaccine
applications. In the area of vaccines in particular
the high stability of DNA, compared with live vi-
rus or protein-based vaccines, suggests that it
may be possible to produce vaccines that remain
relatively stable for years at room temperature.
However, although much is known about the
chemistry of DNA degradation in general, it is not
Plasmid DNA is becoming increasing important
as a pharmaceutical entity because of its poten-
Correspondence to: R. K. Evans (Tel: 215-652-4598. Fax:
15-652-5299. E-mail: robert_evans@merck.com)
2
Journal of Pharmaceutical Sciences, Vol. 89, 76–87 (2000)
©
2000 Wiley-Liss, Inc. and the American Pharmaceutical Association
6
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000
7

DEGRADATION PATHWAYS FOR PLASMID DNA
77
yet clear what, if any, unique obstacles exist for
the development of pharmaceutical formulations
of plasmid DNA with long-term storage stability.
DNA in aqueous solution is known to degrade
lation and metal ion chelators on plasmid DNA
stability. The results suggest that hydroxyl radi-
cals were produced in phosphate-buffered saline
(PBS) formulations containing highly purified
plasmid DNA and that they caused significant
degradation of the DNA over time in storage. Pre-
dictions of plasmid DNA stability were made on
the basis of values of k and k measured in PBS
containing ethylenediamine tetraacetic acid
(EDTA) and ethanol and compared with the ac-
tual stability of DNA in formulations of differing
pH. The results suggest that on control of free
radical oxidation by the addition of EDTA and
ethanol the depurination/␤-elimination pathway
becomes the rate-limiting degradation pathway
and defines the storage stability profile.
by the two-step process of depurination and
1
␤
-elimination, represented by their respective
rate constants k and k , shown in Scheme 1.
1
2
Because these steps are acid and base cata-
lyzed, respectively, the pH of the formulation is
expected to have a large effect on DNA stability.
However, raising the pH of the formulation to re-
duce the rate of depurination may also lead to
pain and necrosis of tissue on intramuscular in-
jection.2 Formulation pH may also affect the
rates of other degradative pathways or influence
the amount of extractable material from the con-
tainer/closure system.
1
2
,3
DNA bases and deoxyribose sugars are also
susceptible to degradation by free radical oxida-
tion, as indicated in Scheme 1, leading to the
MATERIALS AND METHODS
Materials
4
,5
production of oxidized bases and strand breaks.
However, the most common conditions required
to generate hydroxyl radicals, reduced transition
metal ions in the presence of hydrogen per-
oxide, would not exist in simple buffered solu-
tions of highly purified DNA. Therefore, it is not
clear whether hydroxyl radicals could be pro-
duced in these formulations in quantities suf-
ficient to affect plasmid DNA stability during
storage.
In this study we have examined the stability of
plasmid DNA under accelerated conditions to bet-
ter characterize the degradation pathways in
pharmaceutical formulations. The primary sta-
bility indicating assay involved measuring the
conversion of supercoiled plasmid DNA to the
open-circle form because the depurination/␤-
elimination and oxidative pathways each result
in the formation of strand breaks. In some formu-
lations, however, the accumulation of apurinic
Most of the stability experiments were conducted
with a single lot of V1JpHA plasmid DNA (6.6 Kb,
∼
95% supercoiled (SC) and 5% open-circle (OC)
6
DNA). In one experiment (50°C study, Fig. 2)
another lot of plasmid DNA was used; however,
both lots were purified by the same procedure and
were ∼ 95% SC. Ascorbic acid was purchased from
Fluka. Deoxyribonuclease 1, bacterial alkaline
phosphatase. and nuclease P1 were purchased
from Pharmacia. The 8-OH-dG standard was pur-
chased from Wako Co. Escherichia coli exonucle-
ase III was from Promega. To generate the data in
Figure 2, excipients from the following sources
were used: reagent source A (sodium phosphate-
dibasic, J. T. Baker; sodium phosphate-monobasic
and sodium chloride, Mallinckrodt); reagent
source B (all reagents from Spectrum Chemical
Co.); reagent source C (same as source A but dif-
ferent lots); reagent source D (all reagents from
Fisher Scientific, certified ACS grade). EDTA,
imidodiacetic acid (IDA), nitrilotriacetic acid
(
AP) sites and oxidized bases were also examined.
Because trace levels of transition metal ions are
known to catalyze the generation of hydroxyl
radicals, we also examined the effect of demeta-
(
NTA), inositol hexaphosphate (IHP), and dieth-
ylenetriaminepentaacetic acid (DTPA) were all
purchased from Sigma Chemical Co.
Methods
Most of the stability experiments were conducted
in PBS. PBS used at pH 7.2 contained 6 mM so-
dium phosphate and 150 mM NaCl, whereas PBS
at pH 8.0 contained 10 mM sodium phosphate
and 150 mM NaCl. The supercoiled (SC) DNA
Scheme 1.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

78
EVANS ET AL.
content was determined at several time points in
each stability study, even for studies in which the
data from only one time point are presented. The
results presented are from single experiments but
are typical of the results obtained from several
independent experiments.
and sterilized by vacuum filtration with a 0.22-
␮m filter and stored at 4°C until used.
Iron assays
The Ferrozine-based iron assay is a modified form
7
of a previously described method. Aliquots of
samples (340 ␮L) to be assayed were placed in
Supercoiled content assay
1
.5-mL polypropylene microcentrifuge tubes.
Then, 20 ␮L each of 1M HCl, 200 mM ascorbic
acid, and 20 mM Ferrozine (3-(2-pyrrolidyl)-5,6-
bis(4-phenyl-sulfonic acid)-1,2,4-triazine were
added. Samples were then heated to 60°C for 20
min and cooled to room temperature, followed by
measurements of the absorbance at 562 nm. Con-
trols containing deionized water and controls con-
DNA stability assays were performed by deter-
mining the SC, open-circle (OC), and linear DNA
content of samples after storage at either 50 or
3
0°C (in the absence of light) in 3-mL glass vials
equipped with Teflon-coated stoppers. The SC,
OC, and linear DNA content was determined by
agarose gel electrophoresis (1% agarose). The to-
tal mass of plasmid DNA (SC + OC + linear) ap-
plied to each lane of the gel, as measured by the
gel analysis, was nearly constant for samples
taken throughout the entire time course of each
stability study. Aliquots of stability samples (con-
taining 18 ng of DNA) were applied to each lane of
the agarose gel and electrophoresed for 90 min at
+
3
taining 40 ppb of added Fe were performed for
each assay. The absorbance, at 562 nm, of the
deionized water control was typically <0.0002,
+
3
whereas the 40 ppb Fe control sample had an
absorbance of 0.0192. The results are expressed
as averages of two to four assays performed on the
same day. For samples containing 1.0 mg/mL
plasmid DNA, HCl was added to a final concen-
tration of 50 mM. The acidified DNA sample was
then incubated at 100°C for 3 h. Then, 340 ␮L of
the degraded DNA was assayed as previously de-
scribed, keeping the final HCl concentration in
each sample at 50 mM. PBS controls and samples
of the same lot of V1JpHA plasmid DNA used for
most of the stability studies were also assayed by
graphite furnace atomic absorption spectroscopy
by Tektagen Inc. The lot-to-lot variation of iron
content in excipients from a single source was not
examined.
5
V/cm, followed by staining with ethidium bro-
mide. The stained gel was photographed under
ultraviolet illumination, and the negative was
scanned with a BioRad GS-700 densitometer at
6
00 dpi resolution. The amount of DNA in each
band on the gel was determined by comparison of
the band intensities with DNA standards of each
form (SC, OC, and linear) applied to the same gel,
using the BioRad Molecular Analyst software. A
standard curve (r Ն 0.95) was prepared for each
DNA form on each gel, on the basis of the band
intensities of SC, OC, and linear standards. The
precision of the agarose gel analysis (%RSD of SC
DNA content from gel to gel) is typically Յ2% and
Յ12% for samples containing between 11 and
8-OH-dG assay
The 8-OH-dG assay was developed on the basis of
8
,9
1
00% SC DNA or 0 and 10% SC DNA, respec-
two previously published methods.
Plasmid
tively.
DNA (4 ␮g DNA in 20 ␮L) was digested with
DNase 1 by adding 1 ␮L (2.8 units) of DNase and
4
3
␮L of 0.1M MgCl and incubating for 12 h at
7°C in a 1.5-mL microcentrifuge tube. After the
Demetalation of formulations
2
Buffers were demetalated by placing 5 g of Chelex
addition of 39 ␮L of 30 mM sodium acetate (pH
5.3), 3.9 ␮L of 20 mM zinc sulfate, and 1.5 ␮L of
nuclease P1 (1.2 units), the sample was then in-
cubated for an additional 6 h at 37°C. To complete
the digestion process 10 ␮L of 0.5M Tris-Cl (pH
8.3) and 6 ␮L (0.84 units) of bacterial alkaline
phosphatase was added to the reaction, followed
by 6 h of incubation at 37°C. Samples were then
stored at −20°C until analyzed by reverse-phase
high-performance liquid chromatography (HPLC)
and electrochemical detection (BAS 480 analyzer
1
00 (BioRad) resin in ∼ 75 mL of the buffer to be
demetalated. With the solution stirred at a mod-
erate rate on a magnetic stirrer, 1N HCl was
added to the slurry drop by drop, over ∼ 15 to 30
min until the pH of the slurry stabilized at the
desired pH. The slurry was then filtered, washed
with deionized water, and collected. The washed
resin was then placed in 250 mL of the buffer to
be demetalated and stirred slowly overnight at
4
°C. The demetalated buffer was then collected
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

DEGRADATION PATHWAYS FOR PLASMID DNA
79
equipped with two LC-4C amperometric detec-
tors, Bioanalytical Systems, Inc). An aliquot (20
containing 6 mM sodium phosphate, 150 mM
NaCl, 0.5 mM EDTA, and 1% (v/v) ethanol, pH
7.4. Aliquots (0.5 mL) were stored in sealed glass
ampules at 40, 50, 60, and 80°C. After time in
storage, the DNA solutions were removed from
the ampules and stored at −70°C until analyzed
to determine the SC DNA content and the num-
ber of AP sites per plasmid. The SC DNA content
was determined as described previously for the
stability samples. The number of strand breaks
per plasmid (SB, on average in a large population
of molecules) was then determined using the re-
lationship SB ס −ln f , where f is the fraction of
DNA in the SC form. The DNA backbone at the
site of each AP site was cleaved by treating the
plasmid DNA with E. coli exonuclease III, con-
verting SC plasmid DNA to the OC form. Each
35-␮L reaction contained 100 ng of plasmid DNA
and 0.25 units of exonuclease III in the supplied
buffer (Promega). Reactions were incubated for 30
min at 37°C then stopped by the addition of 15 ␮L
of 0.15M EDTA. After the addition of 5 ␮L of
sample buffer containing bromophenol blue in
50% glycerol, 10 ␮L of the sample was applied to
a 1% agarose gel and electrophoresed as described
previously. Control reactions were performed to
ensure complete cleavage of plasmid DNA con-
taining ∼ 1 AP site per plasmid and to ensure that
control SC plasmid DNA, lacking AP sites, was
not cleaved. The number of SBs per plasmid pro-
duced by AP site cleavage was based on the dif-
ference in SB per plasmid before and after AP site
␮
L) of each sample was injected onto a Supelcosil
LC-18s column (4.6 mm × 250 mm) equilibrated
in a mobile phase containing 8% methanol, 50
mM potassium phosphate (pH 4.5), and 1 mM
EDTA at flow rate of 1.5 mL/min. Detection of
8
-OH-dG was performed by the second of two elec-
trochemical detection cells (glassy carbon elec-
trodes) connected in series. A low voltage (0.1 V)
was applied to the first cell to reduce the back-
ground, whereas a higher voltage (0.6 V) was ap-
plied to the second cell to allow for detection of
1
1
1
1
8
-OH-dG (retention time of ∼ 9 min). Quantitation
of 8-OH-dG was performed by comparison of the
area counts (current output of the electrochemical
cell) for the samples to the area counts produced
by a series of 8-OH-dG standards used to gener-
ate a standard curve. The standard curve was lin-
ear (linear regression, r ס 0.999) from 10 to 1000
fmol per injection (0.47–0.94 ␮g DNA per injec-
tion).
Hydroxylation of phenylalanine
The phenylalanine hydroxylation assay was
1
0
based on a published method. Stability samples
containing 5 mM phenylalanine with and without
other excipients were incubated at 50°C in 3-mL
molded glass vials equipped with Teflon-coated
stoppers (West Co.). Tyrosine was detected and
quantitated by subjecting aliquots (20 ␮L) of the
stability samples to a reverse-phase HPLC sepa-
ration on a C18 column, with ultraviolet detection
at 274 nm. The HPLC assay was a 15-min isocrat-
ic separation in a mobile phase containing 90%
cleavage. The depurination rate constant (k , per
1
−1
purine base, units of s ) was determined at each
storage temperature by using the relationship k1
ס (SB + AP)/bp (t), where AP ס AP sites per
plasmid, bp ס base pairs or number of purine
bases in the plasmid (6,600), and t ס time, as
(
1
v/v) 5 mM potassium phosphate (pH 3.0) and
0% (v/v) methanol. Phenylalanine eluted with a
retention time of ∼ 8.5 min, whereas the ortho,
meta, and para forms of tyrosine eluted with re-
tention times of 7.9, 6.3, and 5.2 min, respec-
tively. Baseline separation was obtained between
each ultraviolet absorbance peak. The sum of the
integrated peak areas for ortho, meta, and para-
tyrosine (in milli-absorbance units, mAU) was
used as a measure of total tyrosine production.
The assignment of the absorbance peaks to the
ortho, meta, and para forms of tyrosine was made
on the basis of the comparison of retention times
to ortho, meta, and para-tyrosine standards.
1
2
previously described. The ␤-elimination rate
−
1
constant (k , per AP site, units of s ) was deter-
2
mined using the relationship k ס SB/AP (t), as
2
1
2
previously described.
Predictions of DNA supercoiled content over time
The method used to predict the SC content of
plasmid DNA, using k and k , is based on the
1
2
−
k1t
relationship SB ס PB[ 1 + (1/(k -k ))(k e
)] as previously described. A good ap-
proximation of this equation (SB ס k (bp)t −
−
1
2
2
−
k2t
13
k e
1
1
[
k (bp)t/(1 + k t)]) has been described else-
1 2
Depurination and ␤-elimination rate constants
1
2
where. The simpler form of the equation pro-
vided nearly identical results for predictions of SC
DNA content during storage because the rate of
Rate constants were determined using samples
containing 100 ␮g/mL plasmid DNA in a buffer
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

80
EVANS ET AL.
depurination in the pH ranges examined was very
slow (which results in PB remaining nearly con-
stant over time) and much smaller than the rate
of ␤-elimination. Predictions of the SC DNA con-
tent at any point in time were performed by first
determining SB; then the fraction of DNA in the
SC form was determined using the relationship f
1
−
SB
ס e . This model assumes that depurination
and ␤-elimination are the only significant degra-
dation pathways for the DNA. Predictions of SC
content in formulations having a pH greater than
Figure 1. Effect of pH on the SC content of plas-
mid DNA at 20 ␮g/mL. Photograph of the ethidium
bromide–stained gel showing the separation of the SC,
OC, and linear forms of the plasmid DNA after incuba-
tion in formulations varying from pH 6 to 9. DNA in
lanes 1 to 7 was formulated in 20 mM bis-Tris-propane
containing 150 mM NaCl, whereas DNA in lane 8 was
formulated in PBS, pH 7.1 at 50°C. The positions of SC,
OC, and linear DNA are shown on the left. The percent
of initial SC DNA remaining after 2 weeks of incuba-
tion at 50°C is shown at each pH. The initial SC content
of the plasmid DNA was 93% and was determined as
described in the “Methods.”
7
.4 required that the k and k rate constants be
1 2
adjusted for the change in pH. On the basis of
previous reports it was assumed that k1 de-
creased and k increased by a factor of 10 for each
2
1
4
unit increase in pH.
RESULTS AND DISCUSSION
Effect of pH and Metal Ions on Plasmid
DNA Stability
It is well known that double-stranded DNA in
aqueous solution degrades through a depurina-
tion/␤-elimination process and that the first step
tion, two formulation buffers were prepared and
demetalated by batch binding to Chelex 100 resin.
The results shown in Figure 3 indicate that the
stability of plasmid DNA in either of the demeta-
lated buffers was significantly higher than the
DNA in the nondemetalated buffers. These re-
1
,14
in this process is acid catalyzed.
Therefore,
raising the pH of DNA formulations would be ex-
pected to increase its stability by decreasing the
rate of depurination. To determine the effect of
pH on the stability of SC plasmid DNA, the SC
content of the DNA in several formulations of dif-
fering pH was determined after 2 weeks of incu-
bation at 50°C. The results (Fig. 1) indicate that
DNA stability increased with increasing pH, with
the maximum stability near pH 9. However, the
data also indicated that the stability of DNA in
the bis-Tris-propane/saline formulation at pH 7
was significantly lower than the DNA in PBS at
nearly an identical pH (pH 7.1). This result sug-
gested that the buffering species or impurities in
the reagents affected the stability. When the sta-
bility was determined in PBS prepared from re-
agents purchased from different vendors, the re-
sults (Fig. 2) indicated wide variations in stability
at 50°C and at 30°C, confirming that impurities
in the reagents were contributing significantly to
the rate of degradation.
Because DNA is known to be readily degraded
Figure 2. Supercoiled content of plasmid DNA in
PBS, pH 7.2, at 20 ␮g/mL during storage at 50°C (᭺, ᭢,
4
,5
by hydroxyl radicals, the production of which is
4
,17
often catalyzed by transition metal ions,
the
᭹) or 30°C (᭿, ᮀ). For each storage temperature two or
data suggested that trace levels of transition met-
als were responsible for the variation in plasmid
DNA stability. To determine whether trace levels
of metal ions were contributing to DNA degrada-
three different sources of excipients (sodium chloride
and sodium phosphate) were used to prepare the for-
mulation buffer. (᭺) Reagent source A; (᭢) source B;
(᭹) source C; (᭿) source D; (ᮀ) source C.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

DEGRADATION PATHWAYS FOR PLASMID DNA
81
Figure 3. Effect of demetalation on the SC content of
plasmid DNA at 50°C. PBS, pH 8.0, (᭹) nondemeta-
lated buffer, (᭢) demetalated buffer; 10 mM NaHCO3,
150 mM NaCl, 10 mM sodium succinate, pH 8.3, (᭺)
nondemetalated buffer, (᭞) demetalated buffer. DNA
concentration was 2 ␮g/mL.
sults strongly suggest that trace metal ions were
significantly affecting the stability of the DNA.
The effects of several metal ion chelators on
plasmid DNA stability are shown in Figure 4(A).
The results indicate that DTPA and NTA signifi-
cantly enhanced DNA stability compared with the
PBS control, whereas EDTA decreased DNA sta-
bility. Because DTPA, EDTA, and NTA each have
Figure 4. (A) Effect of metal ion chelators on the SC
content of plasmid DNA (20 ␮g/mL) in PBS, pH 8.0
after 2 or 4 weeks of storage at 50°C. Chelator concen-
tration was 200 ␮M. (B) Effect of EDTA and DTPA on
DNA stability in the presence of transition metal ions,
in PBS (pH 8.0) using 20 ␮g/mL plasmid DNA. The
final concentrations of metal ions and chelators were
+
3
+3
higher binding affinities for Fe (log K for Fe
a
1
5
ס 28, 25, and 16, respectively) than double-
+
3
18
stranded DNA (log K for Fe ס 11.4 and 12.7),
a
each chelator should be able to demetalate DNA.
These results suggest that the ability to stabilize
DNA in these formulations was not simply caused
by removing metal ions that were bound to the
DNA. However, because DTPA is known to
strongly inhibit hydroxyl radical production from
5
00 ppb and 200 ␮M, respectively. The percent of initial
SC plasmid DNA remaining was determined after 1
week of storage at 50°C. IDA ס imidodiacetic acid; IHP
ס inositol hexaphosphate; NTA ס nitrilotriacetic acid.
bound Fe , whereas EDTA does not,19–21 the
+3 19
data suggest that metal ion catalyzed hydroxyl
+
3
radical production from trace levels of Fe was at
least partly responsible for the observed plasmid
DNA degradation.
identity of the trace metal ions causing DNA deg-
radation in our formulations, EDTA or DTPA was
added (200 ␮M final concentration) to PBS formu-
lations (pH 8.0) containing 20 ␮g/mL plasmid
+
3
+2
DNA and either 500 ppb of added Fe , Fe ,
Identity of the Metal Ions Causing a Loss of SC
Plasmid DNA during Storage
+
2
+1
Cu , or Cu or no added metal ion (control). The
percent of initial SC DNA remaining after 1 week
of storage at 50°C is shown in Figure 4(B.) The
results clearly indicate that each of the metal ions
significantly increased the rate of DNA degrada-
tion. Moreover, the reduced form of iron was more
damaging to DNA than the oxidized form. The
Although several transition metal ions are known
to catalyze hydroxyl radical production through
Fenton-type reactions, iron and copper are likely
suspects because they are relatively common
trace contaminants. In an effort to confirm the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

82
EVANS ET AL.
results also indicate that DTPA enhanced the sta-
either iron assay) the ratio of iron atoms to mol-
ecules of plasmid DNA would be ∼ 16, assuming no
contribution of iron by the DNA. If only 10% of
these bound iron atoms produced an SB, the %SC
DNA would decrease from 100 to 20%. These re-
sults suggest that iron levels of <5 ppb in a for-
mulation buffer can lead to significant DNA deg-
radation over time in storage.
+
3
+2
bility of DNA in the presence of Fe , Cu , or
+
1
Cu but decreased its stability in the presence of
+
2
Fe . Although the data after 1 week of incuba-
tion suggest that DTPA only slightly decreased
+
2
the stability of the DNA in the presence of Fe ,
the SC content of the DNA 1 to 2 hours after the
addition of the DTPA was 40%, whereas the DNA
+
2
in the control containing Fe was 92% SC. These
data more dramatically show that DTPA greatly
The data in Figures 2 to 4 are consistent with
the generation of hydroxyl radicals by trace levels
of iron catalyzing Fenton-type reactions. How-
ever, the data also indicate that the oxidized
forms of both iron and copper can damage DNA,
suggesting that these ions can bind to DNA and
directly oxidize it, even in the absence of light.
During long-term storage, it may also be possible
for reducing agents leaching from the closure sys-
tem to provide reducing equivalents to trace
+
2
decreased DNA stability in the presence of Fe .
Because the addition of DTPA to the control
[
without added metal ion, Figure 4(A)] increased
+
2
DNA stability, these results suggest that Fe is
not the metal ion form leading to excessive deg-
radation in the PBS control formulations. The re-
sults in Figure 4(B) indicate that EDTA increased
the stability of plasmid DNA in the presence of
+
2
+2
+1
+3
+2
Fe , Cu , or Cu but slightly decreased DNA
amounts of Fe or Cu to drive Fenton-type re-
actions. Whether the damage to DNA is due to
direct oxidation (abstraction of electrons from
+
3
stability in the presence of Fe . Because the ad-
dition of EDTA to the control [without added
metal ion, Figure 4(A)] also decreased DNA sta-
bility, these results suggest that Fe is the metal
ion form leading to excessive degradation in the
PBS control formulations. However, these results
are certainly not definitive.
+
3
+3
DNA by bound Fe ) or to reduction of Fe fol-
lowed by catalysis of a Fenton reaction, only
+
3
+
3
minute quantities of Fe would be necessary to
cause DNA damage because the metal ion would
be acting as a catalyst. The stability data are con-
sistent with this hypothesis based on the dra-
matic impact of 500 ppb iron or copper on DNA
stability after only 1 week at 50°C (Fig. 4).
To further examine the potential for iron being
the problematic trace metal ion samples of the
PBS formulation buffer and plasmid DNA were
assayed for iron using a spectrophotometric assay
and a graphite furnace atomic absorption assay
as described in the “Methods.” The results, by
both methods, indicated <5 ppb iron in the formu-
lation buffers. However, both methods detected
the presence of approximately 20 to 40 ppb iron
Oxidation of Guanine Bases in Plasmid DNA
during Storage in PBS
To determine whether oxidation of DNA bases oc-
curred concomitantly with the loss of SC DNA,
plasmid DNA was formulated in PBS (pH 7.2) at
100 ␮g/mL and incubated at 50°C for up to 14
weeks. At various time points aliquots were ana-
lyzed for SC, OC, and linear DNA. Additional ali-
quots were analyzed for the presence of 8-hy-
droxydeoxyguanosine (8-0H-dG), one product of
(
in 1 mg/mL plasmid DNA) in the lot of plasmid
DNA used for most of the stability studies, includ-
ing the experiments for Figure 4. This corre-
sponds to roughly two iron atoms per plasmid
molecule, assuming no iron contribution from the
formulation buffer. If each of these bound iron
atoms caused only one SB per plasmid over time
in storage, the percent SC DNA would decrease
from 100 to 14% because of this degradative pro-
cess alone (based on the assumption that the frac-
4
DNA oxidation by hydroxyl radicals, HPLC, and
electrochemical detection. The results (Fig. 5) in-
dicated that 8-OH-dG was detectable after only 2
weeks of incubation, when the SC content had
decreased to ∼ 50%. The level of 8-OH-dG in-
creased with the time of incubation and reached
∼ 250 fmol of 8-OH-dG/␮g of DNA after 6 weeks of
incubation, just as the SC content of the plasmid
DNA approached zero. This level of 8-OH-dG cor-
responds to approximately one oxidized base per
plasmid, on average, and clearly indicates that
oxidation of the DNA was occurring as the OC
content of the plasmid DNA was increasing. Al-
though the level of dG oxidation was low, 8-OH-
−
SB
tion of SC DNA ס e ). If one considers the con-
tribution of iron by the formulation buffer, it be-
comes clear that only trace amounts of iron would
be needed to cause extensive degradation of plas-
mid DNA. For example, the data in Figure 2 show
significant differences in the stability of plasmid
DNA formulated with reagents from different
sources. If the level of iron in these formulation
buffers were only 4 ppb (below the sensitivity of
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

DEGRADATION PATHWAYS FOR PLASMID DNA
83
standards) consistent with the oxidation of phe-
1
0
nylalanine by hydroxyl radicals. With the total
production of tyrosine based on the sum of all
three forms, the results, shown in Figure 6, indi-
cated that the production of tyrosine was detect-
able after 4 weeks of incubation but effectively
eliminated by the addition of 1% ethanol, a known
hydroxyl radical scavenger. The results also indi-
cated that the addition of EDTA or 500 ppb Fe+3
greatly increased the production of tyrosine and
that the highest rate of tyrosine production was
observed in formulations containing both EDTA
+
3
and 500 ppb Fe . These results are consistent
with the capacity of EDTA/Fe to serve as a cata-
lyst for hydroxyl radical production and with the
data indicating that EDTA alone decreased the
stability of plasmid DNA in PBS [Fig. 4(A)].
Taken together these data suggest that free radi-
cal oxidation of plasmid DNA may be a major deg-
radative pathway unless measures are taken to
control it.
Figure 5. DNA composition and accumulation of
-OH-dG residues in plasmid DNA (100 ␮g/mL) stored
at 50°C in PBS, pH 7.2. (+) 8-OH-dG was detected by
reverse-phase HPLC followed by electrochemical detec-
tion. The SC (᭹), OC (᭺), and linear DNA (᭢) content
was determined by gel electrophoresis.
8
dG is only 1 of nearly 100 known base oxidation
4
products produced by hydroxyl radicals. There-
fore, it is possible that the DNA may have accu-
mulated other types of oxidized bases as well.
Moreover, hydroxyl radicals are known to ab-
stract hydrogen atoms from several sites on the
deoxyribose sugars, leading to SBs and to the con-
Model of Plasmid DNA Stabilization
by EDTA/Ethanol
If free radical oxidation of DNA is a major source
of degradation and is caused by trace metal ions,
4
,5
version of SC DNA to OC DNA. If the introduc-
tion of SBs by oxidation occurs at the same rate as
the accumulation of 8-OH-dG, this would result in
a dramatic decrease in DNA stability because the
introduction of one SB per plasmid (on average in
a large population of plasmid molecules) would
decrease the SC DNA content from 100 to 37%.
Therefore, these data are consistent with a model
in which the stability of DNA in PBS formulations
is significantly decreased by uncontrolled free
radical oxidation.
Hydroxylation of Phenylalanine during Storage
in PBS
To confirm that hydroxyl radicals can be gener-
ated in PBS, PBS solutions containing phenylal-
anine were incubated at 50°C for several weeks
and then assayed for tyrosine. Because hydroxyl
radicals cause the oxidation of phenylalanine to
tyrosine, we reasoned that the production of tyro-
sine in this formulation would be consistent with
the generation of hydroxyl radicals in PBS formu-
lations of DNA. The results of the reverse phase
HPLC assay (not shown) indicated the presence of
absorbance peaks corresponding to ortho, meta,
and para-tyrosine (based on o, m, and p-tyrosine
Figure 6. Hydroxylation of phenylalanine to tyrosine
at 50°C in a formulation buffer containing 50 mM so-
dium phosphate, 150 mM NaCl, pH 7.2. Tyrosine was
separated from phenylalanine by reverse-phase HPLC
and detected by its ultraviolet absorbance at 274 nm.
The amount of tyrosine produced was based on the in-
tegrated peak areas of the ortho, meta, and para forms
of tyrosine, expressed as milli-absorbance units (mAU).
(
᭹) Control (5 mM phenylalanine in formulation
buffer), (᭺) control with 1% (v/v) ethanol, (᭞) control
+
3
with 0.5 mM EDTA, (᭢) control with 500 ppb Fe , (᭿)
+
3
control with 500 ppb Fe and 0.5 mM EDTA.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

84
EVANS ET AL.
it seems certain that in the absence of an added
chelator the metal ions would be bound to the
DNA itself because DNA is known to have high
affinity for divalent and trivalent metal ions.1
Moreover, this would place the production of hy-
droxyl radicals on the “surface” of the DNA, mak-
ing it difficult to protect the DNA through the use
of hydroxyl radical scavengers such as ethanol.
This reasoning suggests that the presence of a
chelator that can demetalate DNA would serve to
move the production of hydroxyl radicals to the
bulk solution and off the “surface” of the DNA.
Theoretically, this would allow for more effective
protection of the DNA by a radical scavenger be-
cause the distance between the point of radical
production and the DNA is larger. The results of
DNA stability experiments on formulations con-
taining EDTA and ethanol support this model.
For example, the data in Figure 7 indicate that
the ability of ethanol to protect DNA was consid-
erably better in the presence of EDTA than in its
absence or in the presence of succinate, a much
was found to provide the best protection against
loss of SC DNA over time at 50 and 30°C. These
data suggest that the addition of DTPA or EDTA
and ethanol to the formulation increases DNA
stability by reducing free radical oxidation. More-
over, the results suggest that free radical oxida-
tion was completely inhibited in each formulation
because the level of plasmid DNA stability was
nearly equivalent in formulations containing ei-
ther DTPA or EDTA and ethanol.
8,23
Determination of the Depurination and
␤-Elimination Rate Constants
For DNA in aqueous solution its maximum sta-
bility at any given pH will be in a formulation in
which free radical oxidation is highly inhibited,
leaving depurination and ␤-elimination as the
most significant pathway for degradation (see
Scheme 1). For the evaluation of various formu-
lations of plasmid DNA, it would be useful to pre-
dict beforehand the maximum stability of the
DNA at the desired formulation pH. Then, by
comparing actual stability data with the pre-
dicted stability, the contribution of free radical
oxidation (or other degradation pathways) to the
overall DNA degradation process could be deter-
mined. To make more accurate predictions of
plasmid DNA stability, the depurination and
␤-elimination rate constants were determined in
+
3
weaker binding chelator for Fe . However, etha-
nol had no noticeable effect on DNA stability in
the presence of DTPA, presumably because the
metal ions bound to DTPA were rendered inactive
1
9
for hydroxyl radical generation. Although other
chelators were tested in combination with ethanol
(
including EGTA, IHP, gluconate, and salicylate)
to examine their effects on DNA stability, the
combination of EDTA and ethanol or DTPA alone
6
mM sodium phosphate buffer containing 150
mM NaCl, 0.5 mM EDTA, and 1% ethanol, pH 7.4
at 40, 50, 60, and 80°C, as described in the “Meth-
ods.” An Arrhenius plot of the data (not shown)
indicated linear plots for both depurination (least
squares fit, r ס 0.997) and ␤-elimination (least
squares fit, r ס 1.00). The results indicated that
the activation energies (E ) for the depurination
a
and ␤-elimination reactions were 28.4 kcal/mol
and 25.2 kcal/mol, respectively. These values are
in good agreement with the published activation
1
4,22
energies.
-elimination rate constants at 50°C, pH 7.4
based on the equations to the linear regression
However, the depurination and
␤
(
−
11 −1
lines for the Arrhenius plots) were 5.63 × 10
and 1.4 × 10 s , respectively, whereas the pub-
lished rate constants at the same temperature
and pH are 3.3 × 10
respectively.
s
−
6 −1
Figure 7. Effects of three metal ion chelators on the
ability of ethanol to enhance plasmid DNA stability. All
samples contained 20 ␮g/mL plasmid DNA in PBS (pH
−
10 −1
−6 −1
s
and 2.6 × 10 s ,
1
4,22
It seems possible that the differ-
ence between the rate constants previously deter-
mined and the present results could be due to
better control of free radical oxidation in the
present experimental samples.
8.0) and were incubated for 4 weeks at 50°C. The final
concentrations of ethanol, EDTA, DTPA, and succinate
were 1% (v/v), 0.5 mM, 200 ␮M, and 10 mM, respec-
tively.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

DEGRADATION PATHWAYS FOR PLASMID DNA
85
Predictions of Plasmid DNA Stability
during Storage
To determine whether the rate constants deter-
mined in the presence of EDTA and ethanol could
be used to accurately predict plasmid DNA stabil-
ity, the percent SC DNA remaining over time was
predicted, as described in the “Methods.” As the
results in Figures 8 and 9 show, this method of
predicting DNA stability is remarkably accurate
if supplied with rate constants determined in the
absence of free radical oxidation. It is important
to note that k and k were determined in one
1
2
formulation (PBS containing EDTA and ethanol)
then used to predict the stability of DNA in a
number of different formulations in which actual
stability data were generated. The data in Figure
8
indicate that the predicted stability closely
matched the actual stability data from formula-
tions containing EDTA and ethanol at pH 7.4, 7.7,
and 8.2. Moreover, the method accurately pre-
dicted the stability of formulations at pH 7.7 and
8
.5 over 1 year at 30°C [Fig. 9(A)]. The method
was not, however, able to predict the stability of
DNA in formulations lacking EDTA and ethanol
because an additional degradation pathway (pre-
Figure 9. Actual vs. predicted SC content of plasmid
DNA at 30°C. (A) Data were collected at pH 7.7 (᭹) and
at pH 8.5 (᭺). The predicted stabilities of the plasmid
DNA are indicated by the lines, pH 7.7 (solid line), pH
8
.5 (dashed line). The formulations used were the same
as those described in Figure 8, except that the Tris
formulation in this experiment was adjusted to pH 8.5
at 30°C. (B) Actual vs. predicted SC content of plasmid
DNA for 1.5 years in PBS at pH 7.2. One prediction
(
dashed line) was based on the use of only k and k ,
1 2
whereas the second prediction (solid line) included a
−
1
hypothetical rate constant (k , per base pair, units s
)
3
to account for free radical oxidation. The actual stabil-
ity data (solid circles) were collected using a formula-
tion containing 20 ␮g/mL plasmid DNA.
Figure 8. Comparison of actual vs. predicted SC con-
tent of plasmid DNA at 50°C. Data were collected at pH
sumably free radical oxidation) exists [Fig. 9(B)].
Furthermore, because the stability of DNA de-
pends on the level of trace impurities in the for-
mulation, which may vary from lot to lot of the
excipients or plasmid DNA, the rate constant for
the additional degradation pathway varies with
each preparation. Consistent with this conclusion
is our inability to predict the stability of plasmid
DNA in PBS using either the published rate con-
stants or rate constants derived from formula-
7.4 (᭹), pH 7.7 (᭺), and at pH 8.2 (᭢). The predicted
stabilities of the plasmid DNA are indicated by the
lines, pH 7.4 (solid line), pH 7.7 (short dashes), pH 8.2
(
long dashes). The pH 7.4 formulation was PBS con-
taining 0.5 mM EDTA, 1% (v/v) ethanol, and 100 ␮g/mL
DNA. The pH 7.7 formulation contained 10 mM sodium
phosphate, 150 mM NaCl, 100 ␮M EDTA, 1% (v/v)
ethanol, and 20 ␮g/mL DNA. The pH 8.2 formulation
contained 10 mM Tris-Cl, 150 mM NaCl, 100 ␮M
EDTA, 1% ethanol, and 20 ␮g/mL DNA.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

86
EVANS ET AL.
tions with or without EDTA and ethanol. Also
consistent with this conclusion are results show-
ing that the stability of plasmid DNA in PBS con-
taining EDTA and ethanol was not significantly
affected by the source of the excipients used to
prepare the formulation or by the addition of 500
conditions the supercoiled DNA content would
also be predicted to drop to ∼ 20%. To prevent the
supercoiled DNA content from decreasing below
50%, a pH of at least 7.7 would be needed. At this
pH the predicted number of accumulated AP sites
per plasmid would only be 0.048 after 2 years of
storage at 30°C. Actual measurements of AP sites
in plasmid DNA (20 ␮g/mL) stored at pH 7.7 in-
dicated the accumulation of 0.069 AP sites per
plasmid molecule after 2 years of storage at 30°C.
These data suggest that the potential for AP site
accumulation increases with increasing tempera-
ture. This prediction was confirmed by direct
measurements of AP site accumulation at 40, 50,
60, and 80°C, at pH 7.4 (as described in “Meth-
ods”), that confirmed a larger activation energy
for depurination than for ␤-elimination in formu-
lations containing EDTA and ethanol. However,
these data suggest that no significant AP site ac-
cumulation should occur for plasmid DNA formu-
lations stored for 2 years, at either 5°C (pH Ն7.2)
or 30°C (pH Ն7.7).
+
3
ppb Fe (data not shown).
To determine whether the stability of plasmid
DNA in PBS could be better predicted by a model
that includes one additional pathway of degrada-
tion, the model was modified to include a pseudo-
first order rate constant k , to account for the oxi-
3
dation pathway shown in Scheme 1. The calcula-
tion for the number of SBs per plasmid at any
point in time now becomes SB ס PB[1 + (1/
−
k1t
−k2t
(
k −k ))(k e
− k1e
)] + k (bp)t. In one long-
1
2
2
3
term stability experiment the best fit of the pre-
dicted stability data to the actual data was
−
12 −1
achieved with the value of k set to 4.1 × 10
s
3
(
per base pair). The results shown in Figure 9(B)
clearly indicate that the stability of plasmid DNA
in PBS can be accurately predicted once k , k
1
2,
and k are known. Moreover, on the basis of the
3
close fit of the actual and predicted stability data,
the value of k remained relatively constant for
3
CONCLUSIONS
1
.5 years at 30°C in this formulation.
In summary, these data strongly suggest that the
depurination/␤-elimination and free radical oxi-
dation pathways represent the two major sources
of DNA degradation for highly purified plasmid
DNA in phosphate or Tris-buffered saline formu-
lations. The data also suggest that trace levels of
transition metal ions catalyze the oxidation path-
way, but that the use of specific chelators and/or
hydroxyl radical scavengers can eliminate or
greatly reduce the rate of oxidation for at least 1
year in storage at 30°C. For formulations in which
free radical oxidation was controlled, predictions
of plasmid DNA stability closely matched the ac-
tual stability data, indicating that depurination/
␤-elimination were the only degradative pro-
cesses occurring, and therefore the predicted sta-
bility represents the maximum stability for DNA
in aqueous solution at the given pH.
The ability to predict the maximum storage
stability of plasmid DNA should be useful for the
design of room temperature stable plasmid DNA,
particularly for applications where EDTA and
ethanol may not be compatible with other formu-
lation components such as adjuvants or delivery
vehicles. Moreover, a comparison of the actual
and predicted plasmid DNA stability can be used
Accumulation of AP Sites in Plasmid DNA
during Storage
To have a more complete understanding of plas-
mid DNA stability in pharmaceutical formula-
tions, the presence of accumulated AP sites was
evaluated because they represent a form of DNA
damage that does not directly alter the SC plas-
mid DNA content. On the basis of the method
1
3
described previously for predicting the stability
of plasmid DNA, the number of accumulated AP
sites per plasmid DNA molecule, would be AP ס
−
k1t
−k2t
PB(k /(k −k ))(e
− e
). Using the values for
1
2
1
k and k established earlier, in the absence of
1
2
free radical oxidation, one would predict an accu-
mulation of only 0.029 AP sites per plasmid after
2
years of storage at 5°C. Direct AP site measure-
ments in plasmid DNA stored in PBS, pH 7.2, for
2
<
years at 2 to 8°C indicated the accumulation of
0.01 AP sites per plasmid. Therefore, there ap-
pears to be little cause for concern about the ac-
cumulation of AP sites in plasmid DNA at 5°C,
unless the pH of the formulation is less than 7.2.
For room temperature storage of DNA (assume
3
0°C at pH 7.2) the prediction suggests an accu-
mulation of 0.477 AP sites per plasmid after 2
years of storage. However, under these storage
to determine k at any point in time in which
3
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000

DEGRADATION PATHWAYS FOR PLASMID DNA
87
actual stability data exist. These data will indi-
cate whether the rate of free radical oxidation is
changing over time in storage and help guide the
selection of container/closure components and ex-
cipients to control oxidation over the entire shelf-
life. EDTA and ethanol were found to be ex-
tremely effective for controlling free radical oxi-
dation. For applications in which their inclusion
is acceptable, these excipients, along with careful
pH selection, should allow the preparation of
plasmid DNA formulations that maintain Ն90%
of their initial SC DNA content for 2 years at
room temperature. Taken together, these results
suggest that the identification of other pharma-
ceutically acceptable methods for controlling free
radical oxidation will be important for the devel-
opment plasmid DNA formulations that are
stable at room temperature and is the focus of
ongoing work in this laboratory.
suyama A, Tanooka H. 1986. Formation of 8-hy-
droxyguanine moiety in cellular DNA by agents
producing oxygen radicals and evidence for its re-
pair. Carcinogenesis 7:1849–1851.
1
1
0. Kaur H, Halliwell B. 1994. Aromatic hydroxylation
of phenylalanine as an assay for hydroxyl radicals.
Anal Biochem 220:11–15.
1. Drinkwater NR, Miller EC, Miller JA. 1980. Esti-
mation of apurinic/apyrimidinic sites and phos-
photriesters in deoxyribonucleic acid treated with
electrophilic carcinogens and mutagens. Biochem-
istry 19:5087–5092.
2. Middaugh CR, Evans RK, Montgomery DL, Ca-
simiro DR. 1998. Analysis of plasmid DNA from a
pharmaceutical perspective. J Pharm Sci 87:
130–146.
13. Freifelder D. 1982. Physical chemistry for students
of biology and chemistry. Boston: Science Books In-
ternational, Inc. p 365.
1
1
4. Lindahl T, Nyberg B. 1972. Rate of depurination of
native deoxyribonucleic acid. Biochemistry 11:
3
610–3617.
The authors thank Nirpal Dhanjal and Weigi Lu for
their support of these studies and Sangeetha Sagar for
preparation of the highly purified plasmid DNA.
1
5. Toyokuni S, Sagripanti J-L. 1992. Iron-mediated
DNA damage: Sensitive detection of DNA strand
breakage catalyzed by Iron. J Inorganic Biochem
4
7:241–248.
1
6. Chiu S, Xue L, Friedman LR, Oleinick NL. 1995.
Differential dependence on chromatin structure for
copper and iron ion induction of DNA double-
strand breaks. Biochemistry 34:2653–2661.
7. Nofre C, Cier A, Lefier A. 1961. Activation of mo-
lecular oxygen in aqueous solutions by metal ions
and their chelates. Bull Soc Chim France 530–535.
REFERENCES
1
. Lindahl T. 1993. Instability and decay of the pri-
mary structure of DNA. Nature 362:709–715.
. Oshida S, Degawa K, Takahashi Y, Akaishi S.
2
1
1
1
979. Physico-chemical properties and local toxic
effects of injectables. Tohoku J Exp Med 127:
01–316.
3
8. Netto LES, Ferreira AMD, Augusto O. 1991. Iron
3
. Gupta PK, Patel JP, Hahn KR. 1994. Evaluation of
pain and irritation following local administration
of parenteral formulations using the rat paw lick
model. J Pharm Sci Technol 48:159–166.
(III) binding in DNA solutions: Complex formation
and catalytic activity in the oxidation of hydrazine
derivatives. Chem Biol Interact 79:1–14.
1
2
9. Graf E, Mahoney JR, Bryant RG, Eaton JW. 1984.
Iron-catalyzed hydroxyl radical formation. J Biol
Chem 259:3620–3624.
0. Tullius TD. 1987. Chemical ‘Snapshots’ of DNA:
Using the hydroxyl radical to study the structure of
DNA and DNA-protein complexes. TIBS 12:297–
4
5
6
. Demple B, Harrison L. 1994. Repair of oxidative
damage to DNA. Annu Rev Biochem 63:915–948.
. Imlay JA, Linn S. 1988. DNA damage and oxygen
radical toxicity. Science 240:1302–1308.
. Montgomery DL, Shiver JW, Leander KR, Perry
HC, Friedman A, Martinez D, Ulmer JB, Donnelly
JJ, Liu MA. 1993. Heterologous and homologous
protection against influenza A by DNA vaccina-
tion: Optimization of DNA vectors. DNA Cell Biol
3
00.
1. Tullius TD, Dombroski BA.1986. Hydroxyl radical
footprinting”: High-Resolution information about
2
2
“
DNA-protein contacts and application to lambda
repressor and cro protein. Proc Natl Acad Sci 83:
1
2:777–783.
7
. Carter P. 1971. Spectrophotometric determination
of serum iron at the submicrogram level with a new
reagent (Ferrozine). Anal Biochem 40:450–458.
. Gehrke CW, McCune RA, Gama-Sosa MA, Ehrlich
M, Kuo KC. 1984. Quantitative reversed-phase
high-performance liquid chromatography of major
and modified nucleosides in DNA. J Chromatogr
5
469–5473.
2. Lindahl T, Andersson A. 1972. Rate of chain break-
age at apurinic sites in double-stranded deoxyribo-
nucleic acid. Biochemistry 11:3618–3623.
8
23. Sorokin VA, Valeev VA, Gladchenko GO, Sysa IV,
Blagoi YP, Volchok IV. 1996. Interaction of biva-
lent copper, nickel, manganese ions with native
DNA and its monomers. J Inorg Biochem 63:79–98.
3
01:199–219.
9. Kasai H, Crain PF, Kuchino Y, Nishimura S, Oot-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 89, NO. 1, JANUARY 2000