ABSTRACT
We have developed an
in vitro
DNA polymerase forward mutation assay using damaged DNA templates that contain
the herpes simplex virus type 1 thymidine kinase (HSV-
tk
) gene. The quantitative method uses complementary strand hybridization to
gapped duplex DNA molecules and chloramphenicol selection. This design ensures
exclusive analysis of mutations derived from the DNA strand produced during
in vitro
synthesis. We have examined the accuracy of DNA synthesis catalyzed by calf
thymus polymerase
[alpha]
-primase, polymerase
[beta]
and exonuclease-deficient Klenow polymerase. Using unmodified DNA templates, polymerase
[beta]
displays a unique specificity for the loss of two bases in a dinucleotide
repeat sequence within the HSV-
tk
locus. Treatment of the DNA template with
N
-ethyl-
N
-nitrosourea resulted in a dose-dependent inhibition of DNA synthesis concomitant with an increased
mutation frequency. Similar dose-response curves were measured for the three polymerases examined; thus
the identity of the DNA polymerase does not appear to affect the mutagenic
potency of ethyl lesions. The HSV-
tk
system is unique in that damage-induced mutagenesis can be analyzed both quantitatively and qualitatively
in human cells, in bacterial cells and in
in vitro
DNA synthesis reactions at a single target sequence.
An early step in the mechanism of chemical carcinogenesis is the production of
premutagenic DNA lesions in target tissues (
1
). The conversion of these lesions into mutations is modulated by the competing
rates of DNA repair and DNA replication. Traditional approaches to
understanding the mutagenic consequences of DNA damage have correlated specific
DNA lesions and/or their repair with specific types of mutations (
2
,
3
). We are interested in examining damage-induced mutagenesis from the perspective of the DNA polymerase as a
variable. DNA polymerases constitute a multi-membered family of enzymes that share a common structural motif for the
basic catalytic mechanism of phosphodiester bond formation (
4
,
5
). Human cells contain four known nuclear DNA polymerases ([alpha], [beta], [delta] and [epsilon]) that differ significantly in the kinetic parameters
for DNA and dNTP substrate binding, catalysis, processivity and mechanisms of
error discrimination (
6
,
7
). The recent description of a novel polymerase activity involved in DNA repair
in yeast (
8
) illustrates that the complexities of eukaryotic cells with respect to mutagenesis have yet to be fully realized. Variable
mutagenic responses dependent upon the DNA sequence context and identity of the
polymerase have been observed for a limited number of DNA adducts (
9
-
14
). The extent to which such variability affects the mutational potency and
specificity of various classes of carcinogens within the human genome is
unknown.
In vitro
DNA synthesis assays have shown that the frequency and specificity of
polymerase-induced errors on undamaged DNA templates is unique for each enzyme (
15
). Forward mutation assays of polymerase error rates have the ability to analyze
errors produced in many different DNA sequence contexts (
16
). Only a few forward mutagenesis studies have been carried out using damaged
DNA templates (
17
-
19
). We have developed an
in vitro
forward mutation assay to examine how different eukaryotic DNA polymerases
respond to the presence of DNA damage and process DNA lesions into mutations.
Our mutagenic target is the herpes simplex virus type 1 thymidine kinase (HSV-
tk
) gene, which we have previously used to study damage-induced mutations in
Escherichia coli
(
20
,
21
) and human cells (
22
-
24
). The assay allows for rapid, yet sensitive selection and DNA sequence analysis
of induced mutants by focusing on polymerase-mediated errors in the 5'-region of the HSV-
tk
target. Unlike previous forward mutational studies, which utilized uracil-substituted, damaged templates (
17
-
19
), the HSV-
tk
assay employs naturally occurring, damaged templates for DNA synthesis.
Moreover, a method has been developed which does not introduce the modified DNA
template into
E.coli
for selection of mutants, yet quantitatively recovers DNA synthesis fragments
containing polymerase-mediated errors. The experimental design is versatile and can be applied
to the study of any type of DNA damaging lesion (random or site-specific) and DNA polymerase combination. The HSV-
tk
system is unique in that mutagenesis at a single DNA target sequence can be
analyzed both quantitatively and qualitatively in human cells, in bacterial
cells or
in vitro
using purified DNA polymerases. Keeping the mutagenic target constant will
eliminate biases due to DNA sequence context (damage distribution) and protein
structure (mutant selection), thus allowing direct comparisons to be made among
the different levels of biological complexity.
Calf thymus DNA polymerase [alpha]-primase complex was a gift from Dr Fred Perrino (Wake Forest
University, Winston-Salem, NC). One unit is defined as the amount of protein incorporating 1
nmol [[alpha]-
32
P]dTMP in 60 min at 37oC using activated calf thymus DNA. Recombinant DNA polymerase [beta] was a gift from Dr Samuel Wilson (MIEHS, Research Triangle Park, NC). The 3' -> 5' exonuclease-deficient form (D355A,
E357A) of the Klenow fragment of
E.coli
DNA polymerase I (Exo
-
Klenow) was purchased from Amersham Life Science Inc. (Arlington Heights, IL).
N
-Ethyl-
N
-nitrosourea (ENU) and 5-fluoro- 2'-deoxyuridine (FUdR) were purchased from Sigma
Chemical Co. (St Louis, MO). The purity of the ENU lot used for these analyses
was determined to be 75% by absorbance at 399 nm ([epsilon] =110 M
-1
cm
-1
). All restriction endonucleases were supplied by Gibco BRL Life Technologies
(Gaithersburg, MD) and used according to manufacturer's instructions.
To construct a DNA template for the
in vitro
synthesis reactions, we subcloned the 5'-end of the HSV-
tk
gene (
Eco
RI-
Sph
I fragment of pETK;
20
) into the multiple cloning site of bacteriophage M13mp18. To make a unique
Eco
RV restriction site within the HSV-
tk
gene, the
Eco
RV site at position 577 of pETK was eliminated by oligonucleotide site-directed mutagenesis using uracil single-stranded (ss)DNA (
25
), forming M13TK3.5. The 630 bp
Eco
RI-
Sph
I fragment from M13TK3.5, which includes this mutated site, was reintroduced
into plasmid pETK to form pSHTK. The entire HSV-
tk
gene (1978 bp
Eco
RI-
Bam
HI fragment) was subcloned from pSHTK into the multiple cloning site of the
pGem3Zf(-) phagemid vector (Promega Corp., Madison, WI) along with the 958 bp
Bam
HI fragment from pND123 (
23
), which contains the chloramphenicol acetyltransferase gene (
cat
). The resulting vector, pGTK2 (Fig.
1
), can be used to produce both double-stranded and single-stranded HSV-
tk
-containing DNA forms. Plasmid pGTK3 was derived by oligonucleotide-directed mutagenesis of pGTK2 to inactivate the chloramphenicol
acetyltransferase enzyme (CAT) by a H195A change. A second oligonucleotide was
used to eliminate the
Eco
RI site within the
cat
gene by a silent base substitution in order to physically differentiate
plasmids pGTK2 and pGTK3. The reversion frequency of pGTK3 to confer
chloramphenicol resistance to plasmid-bearing
E.coli
was measured as <5 * 10
-8
.
DNA synthesis templates were created by hybridization of a 20mer oligonucleotide
complementary to the unique
Eco
RV restriction site to M13TK3.5 ssDNA at a 1:1 molar ratio. Using this template,
DNA synthesis is initiated at
tk
position 282, 34 nt downstream of the ATP binding site. The molar ratios of
template DNA to polymerase were optimized for each enzyme such that at least a
203 nt strand is synthesized to include the
Mlu
I restriction site. The
in vitro
reactions contained 2 pmol template DNA at 40 nM concentration. Reaction conditions for Exo
-
Klenow polymerase and polymerase [alpha]-primase were 20 mM Tris-HCl, pH 7.5, 10 mM MgCl
2
, 2 mM dithiothreitol, 1 mM dNTPs and 0.6 pmol Exo
-
Klenow polymerase per pmol template DNA or 1.2 U polymerase [alpha]-primase per pmol template DNA. Reaction conditions for polymerase [beta] were 50 mM Tris-HCl, pH 8.5, 50 mM NaCl, 1 mM dithiothreitol, 1 mM
dNTPs, 200 [mu]g/ml bovine serum albumin and 10 pmol enzyme per pmol template DNA. All
reactions were incubated at 37oC for 60 min and terminated with 15 mM EDTA. The extent of DNA synthesis
was determined by parallel reactions (0.2 pmol DNA, same molar ratios of enzyme
to substrate as above) supplemented with 5 [mu]Ci [[alpha]-
32
P]dCTP (3000 Ci/mmol). The DNA products of these reactions were analyzed on an
0.8% agarose gel followed by autoradiography. Complete synthesis by T7
polymerase was used as a full-length marker, while hybridization of a
32
P 5'-end- labeled
Mlu
I-
Eco
RV restriction fragment hybridized to the ssDNA template was used as a minimal
synthesis marker. When necessary, an 8-12% denaturing polyacrylamide gel was used to ensure that DNA synthesis
had proceeded past the
Mlu
I restriction site.
For mutant analyses, synthesis reactions were heated to 68oC for 3 min to inactivate the polymerase and washed with 9 vol. TE using a
Microcon-30 ultrafiltration unit (Amicon Inc., Beverly, MA). The DNA products were
then digested with
Eco
RV and
Mlu
I restriction enzymes at 37oC for 60 min. The 203 bp fragment (small fragment, SF) was separated from
larger DNA fragments by selective polyethylene glycol (PEG) precipitation using
0.54 M NaCl, 6% PEG8000 and 20 nM DNA (
26
). After overnight incubation on ice, the large DNA fragments were pelleted by
centrifugation and the SF precipitated from the supernatant with ethanol. The
SF was resuspended in 100 [mu]l purified H
2
O and concentrated to 10 [mu]l using a Microcon-30 device. The concentration and purity of the SF was estimated by
agarose gel electrophoresis using DNA mass ladder standards. Typically,
polymerase reactions that had completed synthesis of the 7.8 kb M13TK3.5 DNA
molecule yielded 50-100 ng purified SF (see Fig.
2
).
Alkylated DNA substrates were created by random DNA modification of
oligonucleotide-primed ssDNA to avoid potential loss of thermally labile DNA adducts
during hybridization. The DNA (final concentration 20 nM) was incubated in PEN
buffer (10 mM sodium phosphate, pH 7.4, 1 mM EDTA, 0.1 mM NaCl) containing
various doses of ENU (in dimethylsulfoxide, DMSO) at 37oC for 60 min. A 1 M ENU stock was freshly prepared for each experiment and
used immediately in the DNA modification reaction. Modification was quenched by
diluting the DNA in ice-cold PEN and washing with 20-40 vol. PEN using a Centricon-30 filtration device at 4oC, as previously described (
20
-
22
). Modified DNA was used immediately in the polymerase reactions.
A gapped duplex (GD) molecule was constructed by hybridizing a linear DNA
fragment from pGTK2 to ssDNA from pGTK3, thereby creating a single-stranded region complementary to the SF purified from the polymerase
reactions. The linear DNA fragment was prepared by digestion of pGTK2 with
Eco
RV and
Mlu
I, followed by selective PEG precipitation (
26
) at 0.55 M NaCl, 5% PEG8000, 0.4 [mu]g/[mu]l DNA in a 37oC water bath overnight. The 5.9 kb fragment was recovered by
centrifugation and purified by ethanol precipitation. Single-stranded pGTK3 DNA was prepared by R408 helper phage infection of plasmid-bearing
E.coli
strain DH5[alpha]-IQ. Uracil-containing ssDNA was isolated after helper phage infection of
plasmid-bearing
dut
,
ung
CJ236 bacteria (
25
). UV-irradiated ssDNA was obtained by exposure of 1 [mu]g/ml ssDNA in TE buffer to 254 nm light at a fluence of 800 J/m
2
. The irradiated DNA was protected from exposure to fluorescent light by
wrapping tubes in foil or working under yellow light. The full-length ssDNA was separated from other ssDNA forms by preparative agarose
gel electrophoresis and purified using silica (Pierce, Rockford, IL), according
to the manufacturer's instructions. GD DNA was formed by hybridization of
linear DNA to ssDNA at a 1:1 molar ratio. Linear duplex DNA was diluted in
water (final concentration in hybridization 250 ng/[mu]l) and heated to 85oC for 10 min. During the last minute, ssDNA was added to the tube. The
denatured DNA was placed on ice for 5 min before adding standard sodium citrate
(SSC) to a 2* final concentration (300 mM NaCl, 30 mM sodium citrate). The tubes were
incubated at 60oC for 30 min to promote hybridization and the DNA purified by ethanol
precipitation. Under these conditions, nearly all of the ssDNA was converted to
the GD form, with residual linear DNA as visualized by agarose gel
electrophoresis (see Fig.
2
).
To sample DNA synthesis fragments for mutations, SF prepared as described above
were hybridized to the GD, thus forming pGTK2 heteroduplex plasmid molecules.
The SF (35-150 ng, 0.25-1.2 pmol) was heated to 85oC for 5 min. After addition of SSC to 0.5* (75 mM NaCl, 7.5 mM sodium citrate), GD (200-225 ng, 0.05 pmol) was added to a final
concentration of 15 ng/[mu]l and hybridization allowed to proceed at 45oC for 60 min. An aliquot (30 ng) of the hybridization was removed for
mutational analysis. The remaining hybridization mixture was analyzed by
electrophoresis through a 0.8% agarose gel at 90 V for at least 10 h with GD
and double-stranded nicked pGTK2 standards. Successful hybridization of SF to GD
yielded DNA products that migrated coincident with the nicked standard, with no
evidence of products the size of the GD (Fig.
2
).
To select for SF containing HSV-
tk
mutations, the aliquoted DNA from the final hybridization was used to transform
rec
A13,
upp
,
tdk
E.coli
strain FT334 by electroporation and the bacteria were plated on VBA selective
medium as previously described (
20
), replacing ampicillin with chloramphenicol as the antibiotic. The presence of
50 [mu]g/ml chloramphenicol selects progeny of the `rescued' pGTK2 DNA strand. To
select for HSV-
tk
mutant plasmids, the bacteria were also plated in the presence of 40 [mu]M FUdR. The HSV-
tk
mutant frequency is defined as the number of FUdR-resistant + chloramphenicol-resistant colonies divided by the total number of chloramphenicol-resistant colonies. Mutation frequency curves were analyzed
statistically using one-way and two-way ANOVA tests (
27
).
The post-electroporation doubling time of FT334 in SOC medium at 37oC was measured as 45-50 min, while full FUdR resistance requires a 2 h recovery
period (data not shown). To ensure independence of the mutants selected for DNA
sequencing, bacteria were placed on ice immediately following electroporation
and aliquoted into multiple tubes containing 1 ml VBA medium. Following a 2 h
expression period at 37oC, the individual cultures were plated on selective medium and one FUdR-resistant mutant was isolated from each culture. The HSV-
tk
mutant phenotype was confirmed by replating plasmid-bearing bacteria in the presence of 2 [mu]g/ml trimethoprim, a folate analog used to select for TK
+
activity (
22
). The DNA sequence of the HSV-
tk
gene in the
Mlu
I-
Eco
RV region of each mutant was determined by dideoxy DNA sequence analysis of
plasmid DNA using Sequenase 2.0 according to the manufacturer's instructions
(Amersham Corp.). Differences in proportions of specific types of mutations
were analyzed statistically using Fisher's exact test (two-tailed) (
27
).
In the HSV-
tk
forward mutation assay (Fig.
2
), oligonucleotide-primed ssDNA is used as a template for DNA polymerase reactions. The
template may be treated with mutagen and purified prior to the addition of
polymerase. The DNA reaction products are digested with restriction enzymes and
a 203 bp DNA fragment (SF) is purified (Fig.
2
A). To recover and analyze these DNA fragments for the presence of mutations, a
GD molecule is used. The GD molecule is formed by hybridization of a linear
chloramphenicol-resistant (Cm
R
) DNA fragment to a chloramphenicol-sensitive (Cm
S
) ssDNA, thereby forming a molecule whose single-stranded region is complementary to the DNA fragments to be analyzed. DNA
synthesis fragments containing potential mutations within the
tk
gene are rescued by hybridization to the GD to form pGTK2 heteroduplex plasmid
molecules (Fig.
2
B). These plasmid DNA molecules are used to transform
tdk
E.coli
strain FT334. Incubation of the transformed bacteria in the presence of
chloramphenicol kills bacteria harboring plasmids that are progeny of the Cm
S
strand of the hybridized GD, therefore, the only bacteria which survive
selection are those that are progeny of the `rescued' DNA strand. HSV-
tk
mutant plasmids are selected by plating the bacteria in the presence of FUdR.
The resulting HSV-
tk
mutant frequency is a measure of the proportion of DNA fragments containing
mutations.
Quantitative rescue of DNA fragments.
We have quantitatively evaluated the effectiveness of our method for rescuing
HSV-
tk
mutant DNA fragments (Fig.
3
). Small fragments were isolated from wild-type and mutant (C -> T transition at position 256) HSV-
tk
plasmids. The two DNA populations were combined in different proportions
ranging from 0.1 to 100% mutant SF molecules. The fragments were hybridized to
GD molecules containing Cm
S
ssDNA and the HSV-
tk
mutant frequency determined. Full recovery of input mutant DNA molecules was
observed, where the FUdR
R
Cm
R
frequency was equal to the mutant SF fraction in the sampled DNA population
(Fig.
3
). We have compared our approach to two other methods which selectively cripple
replication of one strand of the GD molecule: uracil and UV lesions (
17
-
19
,
28
). The use of uracil lesions in the GD resulted in recovery of only ~10% of the mutant SF (Fig.
3
). Presumably, the uracil-containing DNA strand is not completely degraded by uracil DNA glycosylase
in FT334 prior to replication of the plasmids. This results in a majority (~90%) of bacteria containing both mutant and wild-type plasmids, which die in the presence of FUdR. We observed results
similar to the uracil GD when UV irradiated ssDNA was used in the GD (data not
shown). In contrast, in the Cm
S
/Cm
R
GD method, the observed HSV-
tk
mutant frequency is a direct quantitation of the proportion of SF DNA strands
containing mutations.
We have analyzed the frequency and specificity of errors produced in the HSV-
tk
assay by three 3' -> 5' exonuclease (proofreading)-deficient DNA polymerases: DNA polymerase [alpha]-primase, polymerase [beta] and Exo
-
Klenow polymerase (Table
2
). DNA synthesis by Exo
-
Klenow polymerase increased the HSV-
tk
mutant frequency to 1.4 +- 0.45 * 10
-3
, ~5-fold over the background mutant frequency. Similarly, the mutant
frequency was increased 8-fold after DNA synthesis by polymerase [alpha]-primase (2.2 +- 1.3 * 10
-3
) and polymerase [beta] (2.4 +- 0.79 * 10
-3
). Plasmid DNA was isolated from individual FUdR
R
colonies obtained from at least two independent synthesis reactions for each
polymerase and the DNA sequence changes within the target region were
determined. These DNA sequence analyses of HSV-
tk
mutants demonstrate the characteristic mutational properties of each polymerase
(Fig.
4
).
To demonstrate the applicability of the HSV-
tk
assay to the analysis of DNA damage, we have examined the responses of DNA
polymerases to lesions produced by ENU. As depicted in Figure
5
A, random ENU treatment of the DNA template resulted in a dose-dependent inhibition of DNA synthesis by Exo
-
Klenow polymerase (Fig.
5
A).
We have developed an
in vitro
forward mutation assay to analyze polymerase-mediated events during DNA synthesis using chemically modified DNA
templates. The experimental system is unique in that mutagenesis can be
analyzed in human cells (
22
-
24
), in bacterial cells (
20
-
21
) or in polymerase
in vitro
synthesis reactions at a single DNA target sequence: the HSV-
tk
gene. This allows us to make direct quantitative and qualitative comparisons
among the different levels of biological complexity. The HSV-
tk
system is general for any form of DNA damage, affording the opportunity to
study the mutagenic potential of previously uncharacterized carcinogens in a
human system.
Using the
in vitro
assay (Fig.
2
), damage-induced DNA polymerase errors can be detected within a variety of sequence
contexts along the 5'-region of the HSV-
tk
gene. Oligonucleotide-primed ssDNA is used as a template for DNA synthesis reactions. The
experimental design ensures exclusive analysis of mutations derived from the
DNA strand produced during
in vitro
synthesis. We have demonstrated that the resulting HSV-
tk
mutant frequency is a direct measure of the proportion of DNA fragments
containing mutations (Fig.
3
) and thus is a quantitative determination of the damage-induced polymerase error frequency. No selective loss of mutations in the
linear fragments due to mismatch repair of the heteroduplex was observed (Table
1
). Thus, our method quantitatively evaluates errors in the newly synthesized DNA
strand and the modified template strand is not introduced into
E.coli
for mutational analyses. The method can be adapted to adduct site-specific studies wherein single adducts are placed in an oligonucleotide
which is hybridized to the GD (
28
), thus allowing direct comparisons of site-specific lesions with random mutagenesis in the same system.
The background HSV-
tk
mutant frequency for the assay is estimated to be 2.9 +- 2.1 * 10
-4
, comparable with other
in vitro
polymerase forward mutation assays (
16
,
19
). We have not yet saturated the HSV-
tk
mutational target with regard to all possible mutations that will be detectable
by our selection scheme. To date, we have detected within the
Mlu
I-
Eco
RV region all 12 possible base substitution mutations at 58 sites and 18
repetitive sites at which we have detected one base frameshift mutations. Two
base frameshift mutations, complex mutations and large deletions have also been
observed in this region. The target sequence for this assay is operationally
defined by the restriction enzymes used to construct the GD molecule.
Therefore, the mutagenic target can be easily expanded to encompass a greater
number of sequence contexts within the 1150 bp HSV-
tk
coding sequence by using a different pair of restriction enzymes.
We have determined the accuracy of three 3' -> 5' exonuclease-deficient polymerases on undamaged DNA templates in
the HSV-
tk
assay: calf thymus DNA polymerase [alpha]-primase, polymerase [beta] and Exo
-
Klenow. DNA synthesis by all three polymerases increased the HSV-
tk
mutant frequency 5- to 8-fold over the background mutant frequency (Table
2
). DNA sequence analyses of HSV-
tk
mutants illustrate the characteristic mutational properties of each polymerase
(Fig.
4
). The observed HSV-
tk
base substitution mutation frequency was similar for polymerase [alpha]-primase and polymerase [beta] (~5 * 10
-4
), however, the enzymes differ in their preferences for misinserted bases and
exhibit significantly different transition to transversion ratios. While both
eukaryotic polymerase spectra displayed a high proportion (50%) of one base
frameshift mutations, the polymerase [beta] spectrum is characterized by a high frequency of two base deletion
mutations in a TATATA dinucleotide repeat motif (Table
3
and Fig.
5
). These general tendencies of the polymerases to produce base substitution
versus frameshift errors at the HSV-
tk
locus
are the same as those previously described for the
lac
Z gap-filling forward mutation assay (
31
,
32
).
Our analysis of the HSV-
tk
data suggests that the relatively low accuracy of polymerase [beta] is not general for all DNA sequences, but is restricted to the loss of
bases within mononucleotide and dinucleotide repeated sequences. The observed
base substitution accuracy of polymerase [beta] is merely 1.5-fold lower than that of polymerase [alpha]-primase, whereas the observed one base frameshift
accuracy of polymerase [beta] is only 2.5-fold lower than that of polymerase [alpha]-primase (Table
3
). However, the loss of two bases in a dinucleotide repeat is a high frequency
event for polymerase [beta] in the HSV-
tk
sequence (Fig.
4
), constituting 12% of the total mutants observed. Moreover, the frequency of
this type of mutation for polymerase [beta] is at least 10-fold greater than that observed for either polymerase [alpha]-primase or Exo
-
Klenow. These observations suggest that during base excision repair of non-repetitive DNA sequences, polymerase [beta] may be capable of accurate DNA synthesis in the proper sequence
context.
The HSV-
tk
in vitro
assay was developed to elucidate how the biochemical differences among
polymerases affect mutagenic processing of DNA lesions. A dose-dependent inhibition of DNA synthesis and concomitant increase in mutation
frequency was observed when DNA primer-templates were treated with ENU (Fig.
5
). Moreover, for the three polymerases examined, the ENU dose-response mutation frequency curves do not differ significantly from one
another (Fig.
6
). We conclude that the identity of the polymerase does not quantitatively
affect the overall mutagenic potency of the alkyl lesions. However, we have
observed significant qualitative differences in ENU-induced mutational spectra among the polymerases (
30
); thus, the types of mutations produced on each template are partially
dependent on the DNA polymerase that processes the lesions (Eckert,K.A.,
Vargo,P.L. and Hile,S.E., manuscript in preparation).
We are grateful to Drs Perrino and Wilson for their generous gifts of purified
polymerases. We thank Dr Monnat for providing the
E.coli
strain, Elaine Watters and Jeremy Edmonds for technical help with the mismatch
repair experiments and Chuck Hill and Jill Hite for critical reading of the
manuscript. Development of this assay was aided by grant IRG-196 from the American Cancer Society and the Four Diamonds Fund of The
Pennsylvania State University. This research was supported by the American
Cancer Society (CN-144) and the American Association for Cancer Research Gertrude Elion
Award. We gratefully acknowledge the generous contributions made to the Jake
Gittlen Cancer Research Institute.
REFERENCES
Return