[CANCER RESEARCH 52. 4183-4189, August 1.

1992]

Gene- and Strand-specific Damage and Repair in Chinese Hamster Ovary Cells
Treated with 4-Nitroquinoline 1-Oxide
Elizabeth G. Snyderwine1 and Vilhelm A. Bohr
Laboratory of Experimental Carcinogenesis, Division of Cancer Etiology- [E. G. S.J, and Laboratory of Molecular Pharmacology [V. A. B.], Division of Cancer
Treatment, National Cancer Institute, NIH, Bethesda, Maryland 20892

ABSTRACT much more efficient than repair in the flanking nontranscribed

regions or in the genome overall (1). The term "preferential
4-Nitroquinoline 1-oxide (4NQO) is a model chemical carcinogen DNA repair" describes the observation that repair in transcrip
that has often been referred to as a UV mimetic agent. Previous studies
have indicated that UV-induced pyrimidine dimers are repaired prefer
tionally active gene regions is faster than repair in noncoding
entially and strand-specifically in actively transcribing genes. In the regions. In addition to the DHFR gene, preferential repair of
current study we have examined the gene-specific and strand-specific pyrimidine dimers occurs in other active genes of CHO cells (3)
repair of 4NQO in Chinese hamster ovary B-l 1 cells treated with 2.5 /u\i and in active genes of human (4) and mouse cells (5). Subse
4NQO. The methodology used for detecting adducts involved the treat quently it was determined that the repair of pyrimidine dimers
ment of DNA from 4NQO-exposed cells with uvrABC excinuclease, occurs in a strand-specific manner in active genes (6). It was
which incises DNA at adduct sites, followed by denaturing gel electro- shown that dimer repair in the transcribed strand of the DHFR
phoresis of DNA, Southern hybridization, and probing for the sequence gene is more efficient than repair in the nontranscribed strand,
of interest. We examined the active and inactive coding regions of the which is repaired at a rate comparable to that of nontranscribed
DHFR gene, the active adenine phosphoribosyltransferase gene, rela genes (6). This finding suggests that the DNA repair mecha
tively inactive c-fos oncogene, and the mitochondria! genome for 4NQO
nism might be associated with the transcriptional complex.
adducts. Initial 4NQO adduct levels found in these genes varied from
1.10 to 1.52 adducts/10 kilobases. Little difference in repair was found
A number of compounds, including chemotherapeutic agents
between active coding and inactive regions of the DHFR gene, or be and chemical carcinogens, have recently been examined for
tween DHFR, adenine phosphoribosyltransferase, and c-fos genes, DNA damage and repair in specific genes (7-12). From these
which are transcribed at different levels. Approximately 71% of 4NQO studies it appears that the phenomenon of preferential repair
adducts were repaired within 24 h in all gene sequences examined. depends on the damaging agent. 4NQO is a model chemical
During this same time period, approximately 51% of adducts were re carcinogen which has been extensively studied with respect to
paired from the genome overall, as determined by comparing the re carcinogenicity, mutagenicity, DNA adduct formation, and
moval of bound radiolabeled 4NQO to total DNA. The results indicate DNA repair (13,14). Upon metabolic activation in cells, 4NQO
that 4NQO adducts, unlike UV light-induced cyclobutane pyrimidine
forms adducts in DNA primarily at guanine bases and to a
dimers (UV dimers), are not preferentially repaired in transcriptionally
lesser extent at adenine bases (14, 15). Three primary adducts
active genes. However, there may be regions of the genome that are not have been identified: 3-(/V2-guanyl)-4-aminoquinoline 1-oxide,
repaired with the same efficiency as the specific genes examined here. In A^-(guan-8-yl)-4-aminoquinoline 1-oxide, and 3-(/V6-adenyl)-4-
addition, little to no difference was observed in the repair of 4NQO
adducts in the transcribed and nontranscribed strands of the DHFR aminoquinoline 1-oxide, comprising 50%, 30%, and 10%, re
gene, a finding which is also in contrast to results with UV dimers. spectively, of total adducts. 4NQO behaves very similarly to UV
Interestingly, 4NQO adducts, unlike UV dimers, were removed from the light in several repair systems and has often been referred to as
mitochondrial genome, suggesting that repair of select lesions occurs in a UV mimetic agent (16, 17). Both bulky 4NQO adducts and
this organelle. Thus, there appear to be some differences in the repair UV light-induced pyrimidine dimers are repaired by nucleotide
pathways operating for 4NQO adducts and UV dimers, particularly excision repair pathways (18). The UV light mimicry of 4NQO
with respect to gene- and strand-specific DNA repair. has been seen in the sensitivity of Escherichia coli uvrA- and
uvrB- strains (16, 17) and in the sensitivity of various comple
INTRODUCTION mentation groups of xeroderma pigmentosum cells (19). In
addition, the DNA repair patterns of some xeroderma pigmen
DNA repair plays an important role in ensuring proper cell tosum cell lines have been shown to be similar for UV dimers
replication and preventing mutagenesis, carcinogenesis, or cell and 4NQO (20-22), and the excision repair of both UV light-
death after DNA damage. This damage can result from expo and 4NQO-damaged DNA shows a lack of involvement of poly-
sure to a myriad of physical and chemical agents. One recent (ADP-ribosyl)ation (23).
advance in the field of DNA repair is the ability to examine In E. coli, 4NQO adducts are repaired by uvrABC excinu
DNA damage and repair at the gene level. The first studies on clease, an enzyme complex which catalyzes the incision step
gene-specific repair were carried out measuring UV light-in
during nucleotide excision repair of U V dimers and many other
duced pyrimidine dimers (1, 2). It was shown that the repair of bulky DNA lesions (18). In vitro studies using purified uvrABC
pyrimidine dimers in the active DHFR2 gene of CHO cells was excinuclease have clearly established that the N2- and C8-gua-
nine base adducts of 4NQO are substrates for this enzyme (24).
Received 1/21/92; accepted 5/27/92. Recently, Thomas et al. (25) demonstrated that 4NQO adducts
The costs of publication of this article were defrayed in part by the payment of generated in vitro in naked DNA can be quantified at the gene
page charges. This article must therefore be hereby marked advertisement in accord
ance with 18 U.S.C. Section 1734 solely to indicate this fact.
level. Their methodology involves the treatment of 4NQO-
1 To whom requests for reprints should be addressed, at Laboratory of Experi modified DNA with purified uvrABC excinuclease followed by
mental Carcinogenesis, National Cancer Institute, Building 37, Room 3C28, NIH,
Bethesda, M D 20892.
denaturing gel electrophoresis, Southern blotting, and hybrid
2 The abbreviations used are: CHO-B11, Chinese hamster ovary B-l 1 cell line; ization with specific probes. In our study this methodology was
4NQO, 4-nitroquinoline 1-oxide; UV dimers, UV light-induced cyclobutane pyri used to examine 4NQO adduct formation and repair in genes in
midine dimers; DHFR, dihydrofolate reducÃ-ase;APRT, adenine phosphoribosyl
transferase; SDS, sodium dodecyl sulfate; TE, 10 mM Tris, pH 8.0, l min EOT A; vivo in cultured mammalian cells. The formation and repair of
DMS, dimethyl sulfate. 4NQO adducts were examined in several different gene regions
4183
 REPAIR OF 4-NITROQUINOLINE
of CHO-B11 cells that were of particular interest with respect
to the link between transcriptional activity and repair and with
respect to possible target genes in carcinogenesis. We were also
interested in studying 4NQO adduct repair in light of similar
ities to UV dimer repair. The CHO-B11 cell line was chosen
because gene-specific and strand-specific repair of pyrimidine
dimers has been well characterized in these cells (1,2, 6). The
genes examined in the current study are the active and inactive
regions of the DHFR gene locus, the relatively inactive c-fos
gene, the active APRT gene, and the mitochondria! genome.
The repair of 4NQO adducts in the transcribed and nontran-
scribed strands of the active coding region of the DHFR gene
was also examined. Another measure of preferential repair is
whether the active gene region is repaired more efficiently than
the overall genome, such as is the case for pyrimidine dimers in
these cells (1, 2). Thus in addition to gene-specific and strand-
specific repair, we also measured the overall genome repair of
4NQO adducts in CHO-B11 cells.
MATERIALS AND METHODS
Chemicals
4NQO, thymidine, and methotrexate [(+)-amcthopterin) were pur
chased from Sigma (St. Louis, MO). pH]4NQO (ring-labeled; 1.9 Ci/
minoli was purchased from the National Cancer Institute Chemical
Repository. [3H]Thymidine was obtained from NEN-Dupont (Wil
mington, DE). Ham's F-12 medium and dialyzed fetal bovine serum
were obtained from Gibco (Grand Island, NY). Purified uvrABC en
zyme subunits were provided by A. Sanear (University of North Caro
lina).
Cell Cultures
CHO-B11 cells carrying the amplified DHFR gene were grown in
Ham's F-12 medium without glycine, hypoxanthine, or thymidine and
supplemented with 10% dialyzed fetal bovine serum, penicillin-strep
tomycin, and 500 n\i methotrexate (26). For studies of repair in specific
DNA sequences, cells were prelabeled with [3H]thymidine and subcul-
tured into normal media without methotrexate (2.5 x 10* cells/10-cm
dish) 1 day prior to carcinogen treatment.
Carcinogen Treatment
4NQO (dissolved in dimethyl sulfoxide) was added to the medium,
and the cells were incubated for l h at 37°C.The medium was removed,
and the cells were washed twice with warm (37°C)media followed by
two washes with warm phosphate-buffered saline. Cells used for the
determination of initial adduct levels were lysed immediately with lysis
solution (10 imi Tris, 1 HIMEDTA, 0.5% SDS, and l mg/ml proteinase
K). Cells used for repair analysis were reincubated with medium sup
plemented with bromodeoxyuridine:fluorodeoxyuridine (10~5 ,\i:10~6
M). Following repair, cells were washed twice with phosphate-buffered
saline and lysed.
Purification of Parental DNA and Restriction
DNA purification and restriction have been described in detail pre
viously (26) and will be only briefly described here. Following incuba
tion with proteinase K, DNA was purified by phenohchloroform ex
traction and treated with RNase A. DNA was subsequently restricted
with Kpnl, and the parental DNA was purified by CsCl gradient frac-
tionation. Parental DNA was used for all repair measurements. This
was necessary because replicated DNA would be incorrectly assayed as
repaired under the conditions of our assay.
uvrABC Exinuclease Treatment
Conditions Used for Amplified Sequences. uvrA, uvrB, and uvrC
protein subunits (1.84 ^g each) were mixed on wet ice, added to 1 fig
4184
I-OXIDE ADDUCTS IN GENES
DNA in 22 n\ buffer (50 HIMTris-HCl, pH 7.5, 10 HIMKC1, 10 HIM
MgCl2, 5 imi dithiothreitol, 2 HIMATP), and incubated for 15 min at
37°C.Control samples were also run under identical conditions, except
that no uvrABC was added. The reaction was terminated by the addi
tion of 2 M!of a proteinase K-SDS solution (20 ng proteinase K, 5%
SDS) followed by incubation for l h at 37°C.The sample was then
dialyzed against TE buffer using a microdialyzer (Pierce, Rockford, IL).
The entire sample was then loaded onto a gel and electrophoresed as
described below.
Conditions Used for Nonamplified Sequences. The reaction condi
tions were identical to those described above, except that the entire
reaction was scaled up 10-fold such that 10 Mgof DNA were treated
with 18.4 »ig of each uvr subunit. Following termination of the reaction
by incubation with 20 p\ proteinase K-SDS solution, DNA was ex
tracted once with phenol, followed by phenohsevag (1:1) and then
sevag. DNA was precipitated with 2 volumes of ethanol after the addi
tion of ammonium acetate and washed with 70% ethanol. DNA was
dried to just dampness and redissolved in 55 n\ of TE. The concentra
tion of DNA was determined spectrophotometrically (in triplicate in
2.5 ii\ of sample), and 4-^g portions of DNA (adjusted to 45 n\ with TE)
were used for gel electrophoresis.
Gel Electrophoresis, Southern Transfer, and Hybridization Condi
tions. Alkaline gel electrophoresis, followed by Southern transfer to
Sure Blot nylon support membrane (Oncor, Gaithersburg), and hybrid
ization with labeled probes were carried out essentially as described
previously (9, 26).
DNA Probes and Washing Conditions. DNA probes for the 5'
DHFR gene, 3'DHFR gene, and downstream regions have been de
scribed previously (2). Fig. 1 shows the DHFR gene regions probed by
pMB5 (Fig. 1/1), pB13-7 (Fig. IB), and cs-14 (Fig. 1C) (subcloned to
ni ¡mini/orepetitive sequences). The cs-14 region is not transcribed, as
determined by Northern assay.' The c-fos probe (5-kilobase genomic
Hind\ll-BamHl mouse insert) was obtained from Lofstrand (Gaithers
burg, MD). In the Apnl-restricted hamster genome, c-fos probes for a
17-kilobase region which spans over the 3.5-kilobase reading frame of
the gene (9). APRT probe was provided by R. Nairn (University of
Texas). This is a 3.9-kilobase ßamHIfragment genomic probe which
spans over the 2.2-kilobase gene (9). This probe detects a 10-kilobase
Kpn\ fragment that encompasses the total reading frame of the gene.
The probe used in hybridizations with mitochrondrial DNA was pro
vided by S. Le Doux (University of South Alabama) and consists of the
entire 16-kilobase mouse mitochondria! genome inserted at the .Sari
site of plasmici pSP64. This probe recognizes a 16-kilobase Kpn\ frag
ment in the mitochondria! DNA of hamsters. The DHFR probes,
APRT, and c-fos were labeled by the random priming method, and the
mitochondria! probe was labeled by nick translation. Membranes were
hybridized with approximately IO7 counts/bag. After hybridization
membranes were washed as described (9, 26).
RNA Probes. The riboprobe constructs have been described in detail
elsewhere and will be just briefly described here.4 Plasmid pZ3d8 was
constructed by R. Nairn (University of Texas) from the pGem vector. It
contains a 2.5-kilobase fragment of the hamster DHFR genomic se
quences from the active coding region of the gene at the Aval site. The
plasmid pZ3d8 was digested with restriction enzymes //«ml11and Kpnl
to generate templates for strand-specific riboprobes. Reactions contain
ing either T7 or SP6 RNA polymerase produce the transcribed or
nontranscribed strand probes, respectively. ["P]CTP labeling of the
probes was carried out using Boehringer Mannheim (Indianapolis, IN)
SP6/T7 transcription kit reagents and protocols. The membranes were
prehybridized, hybridized, and washed as previously described (26).
After autoradiography, the resulting bands were analyzed as described
for double-stranded probing.
' K. Wassermann and V. A. Bohr, unpublished observations.
4 A. May, R. Nairn, D. Okumoto, K. Wassermann, T. Stevnsner, J. Jones, and
V. A. Bohr, Repair of individual DNA strands in the hamster DHFR gene after
treatment with UV, alkylating agents, and cisplatin. submitted for publication.
 REPAIR OF 4-NITROQUINOLINE
DHFR GENE
Kpnl
1 _L J L L
-60 -50-40-30-20-10 0 10 20 30 40 50 kb
Fig. 1. Map of the DHFR gene locus. Bold line, position of the gene; vertical
lines, Kpn\ restriction sites. Locations of the probes: A, p\iB5: B, pbl3-7; C
es-14.
Quantitation of Adduci Levels in Specific Genes. After washing the
membranes, band intensities were quantified using a Shimadzu dual-
wavelength thin-layer chromatoscanner model CS-930. Each mem
brane was deprobed and reprobed to examine the repair in a number of
different sequences. Calculations of adduci levels in specific genes were
carried out as described using ///mflll-restricted pBR322 as an internal
standard to control for DNA quantitation in individual samples (1, 26).
The average number of adducts per fragment (adduci frequency) was
calculated from the zero class measurements by the Poisson expression
(average number of adducts/fragment = —¿Inof the zero class fre
quency) and assuming a 50% incision efficiency of uvrABC (24).
Assays for Measuring 4NQO Adduct Level and Repair in the Overall
Genome
Binding to DNA. CHO-B1 1 cells (not prelabeled) were
plated at a density of 2.5 x IO6 cells/ 10-cm dish 1 day prior to treat
ment with pH]4NQO. Cells were treated as described above for unla-
beled 4NQO. Following treatment, cells were lysed immediately for the
determination of initial adduct levels or allowed to repair in the pres
ence of bromodeoxyuridine:fluorodeoxyuridine prior to lysis. The DNA
was isolated by phenol extraction and treated with RNase A. The DNA
was then sheared through a 25-gauge needle (twice), and parental DNA
was isolated by CsCl gradient centrifugation. The parental DNA was
dialyzed against TE and precipitated. The DNA concentration was
determined spectrophotometrically, and the [3H]4NQO adduct concen
tration was determined by liquid scintillation using Aquassure (NEN-
Dupont) on a Beckman scintillation counter programmed for quench
correction.
Alkaline Sucrose Sedimentation Analysis following uvrABC Excinu-
clease Treatment. DNA (1 n%)isolated from pH]thymidine-prelabeled
CHO-B11 cells exposed to 4NQO was treated with uvrABC excinu-
clease exactly as described above under conditions used for amplified
sequences. The reaction was terminated by the addition of 2 ¿il0.5 M
EDTA. The entire reaction was loaded onto a 5-ml, 5-20% alkaline
sucrose gradient containing 0.3 NNaOH, 2 MNaCI, and 0.01 MEDTA.
Centrifugation was carried out in a Beckman SW 50. 1 rotor at 30,000
rpm for 6 h. The gradients were fractionated into scintillation vials, and
the fractions were counted for radioactivity. The average molecular
weight in excinuclease-treated or untreated DNA was calculated
through the use of a computer program provided by A. Van Zeeland
(University of Leiden). The enzyme sensitive sites were determined
from the molecular weight averages as described previously (27). Cal
culation of the adduct levels per 10-kilobase fragment was extrapolated
from the adduct level per unit of DNA assuming a molecular weight
average of 330/nucleotide base and a 50% cutting efficiency of the
uvrABC enzyme.
Repair Replication. The repair replication method was carried out as
described by Smith et al. (28). [32P]Orthophosphate-prelabeled CHO-
Bl 1 cells were treated with 2.5 MM4NQO and then allowed to repair in
the presence of bromodeoxyuridine:fluorodeoxyuridine and pHjthymi-
dine for 4, 8, and 24 h. The amount of tritium incorporated into 4NQO
repair patches in parental DNA was determined following alkaline
rebanding (28).
4185
I-OXIDE ADDDCTS IN GENES
RESULTS
4NQO Adduct Levels and Repair in the Overall Genome. In
order to establish that uvrABC was recognizing and incising
4NQO adducts in the DNA of CHO cells, uvrABC excinuclease
incision of DNA from 4NQO-treated cells was first measured
in the overall genome. Using alkaline sucrose gradient analysis,
we observed that treatment of DNA isolated from 4NQO-ex-
posed cells with uvrABC caused incisions in DNA which were
detected as a decrease in the average molecular weight of single-
stranded DNA (Fig. 2). The molecular weight average of DNA
from unexposed cells treated with uvrABC or of DNA from
exposed cells not treated with uvrABC was 20.6 X IO6. In DNA
from 4NQO-exposed cells treated with uvrABC, the molecular
weight declined to 12.6 x IO6 and 0.54 x IO6 at 2.5 and 25 MM
4NQO, respectively. We calculated that at 2.5 MM4NQO there
were 3.1 enzyme-sensitive sites/108 daltons of DNA or 0.2
adducts/10-kilobase single-stranded DNA (assuming a 50%
cutting efficiency of uvrABC). At 25 MM4NQO, there were 180
enzyme-sensitive sites/IO8 daltons of DNA or 12 adducts/10-
kilobase single-stranded DNA.
Analysis of cesium chloride gradients of DNA from untreated
cells and cells treated with 2.5 MM4NQO indicated that a 15%
reduction in DNA replication occurred over a 24-h period in
cells exposed to 2.5 MM4NQO, suggesting that this concentra
tion of 4NQO was minimally toxic to the cells. The repair of
4NQO adducts in the overall genome was determined for com
parison with the gene-specific repair by measuring the removal
of radiolabeled 4NQO from bulk DNA at different time inter
vals after exposure to 2.5 MM[3H]4NQO. Approximately 51%
of 4NQO adducts found initially in DNA were removed 24 h
after treatment (Table 1). Repair synthesis, as measured by the
repair replication assay, also showed a linear increase over the
24-h period following exposure to 2.5 MM4NQO (data not
18
16
14
12
•¿Ã¨10
o
CO
=§ e
CO
OC
6
1.0 0.8 0.6 0.4 0.2 0
Relative Distance Sedimented
Fig. 2. Alkaline sucrose gradient analysis of DNA from untreated CHO cells
(•)and from cells treated with 2.5 MM(A) and 25 MM(O) 4NQO. All DNA samples
were treated with uvrABC excinuclease as described in "Materials and Methods"
prior to running the gradients. Arrow, sedimentation of a A DNA standard with a
molecular weight of 1.5 x IO7.
 REPAIR OF 4-NITROQUINOLINE 1-OXIDE ADDUCTS IN GENES

Table l 4NQO adduci levels and removal in hulk DNA of CHO cells" ing in and around the DHFR gene, APRT gene, c-fos gene, and
|'H)4NQO mitochrondrial genome, 4NQO adducts were clearly detectable.
Repair
(h)04824fmol
Time bound//ig
DNA*97 repaif17%23%51%
The adduct frequencies in different genes at different repair
±3180 times were quantitated from two or three biological experi
±2073
1743±
ments (Table 2). The initial adduct levels found in the genes
±6% were approximately 1.10-1.52 adducts/10 kilobases, and al
»CHO Bl1 cells were incubated with |-'H|4NQO (2.5 /IM)for l h at 3TC. Cells though there was some variation in initial adduct levels among
were lysed immediately (0 h, no repair) or lysed after being allowed to repair for the the genes examined, this difference did not appear to be signif
designated times in the presence of bromodeoxyuridine. DNA was isolated, and
parental DNA was purified by CsCI density centrifugation. The level of total ad- icant. In all regions examined a substantial repair of 4NQO
ducts was calculated from radioactivity bound per >igDNA. adducts was observed over a 24-h period (Fig. 5/1). While some
* Values are the mean ±SD of three experiments.
c The percentage repair was calculated for each experiment, and the value shown heterogeneity in the percentage of repair was observed among
is the mean of three experiments. the genes examined at 4 and 8 h, no correlation of repair with
transcriptional activity was apparent. By 24 h, little heteroge
neity in repair was observed among the different genes. In and
around the DHFR gene the percentage repair after 24 h was
approximately the same despite differences in the transcrip
HvrABC - +
" + - * ! Fragment Size tional activity of the different regions. The repair in the active
DHFR gene, 3'-DHFR region, and noncoding downstream re
gion was 70%, 77%, and 72%, respectively, after 24 h. In ad
A. 5' DHFR •¿Â»_••_ 14 kb
dition, repair levels in the active APRT gene and relatively
inactive c-fos gene were similar over the 24-h period. Approx
imately 65% and 71% of the initial 4NQO adducts were re
B. 3' DHPR 9 kb paired from the APRT and c-fos genes, respectively, during this
time. In the mitochondrial DNA, the rate of 4NQO adduct
removal during the first 8 h appeared to be slightly higher than
C. 3' Dwnstrm •¿Â» —¿ 14 kb that observed in nuclear genes. By 24 h, 70% of the initial
adduct levels were removed from the mitochondrial genome,
which was approximately the same as that observed in the other
10 kb
genes. This result suggests that the repair of 4NQO adducts
D. APRT
occurs in the mitochondria.
The same Southern blots that were hybridized with the ge
nomic DHFR probe pMB5 were also probed with the specific
£. c-fos 17 kb
probes for the transcribed and nontranscribed strand of this
same 5'-DHFR region. Representative autoradiograms are
shown in Fig. 4, and the results from two experiments were
1. Nitochnd 16 kb quantitated and pooled in Table 3. 4NQO adducts were clearly
detectable in both the transcribed and nontranscribed strands of
Fig. 3. Autoradiograms showing (he repair of 4NQO adducts in specific re the active DHFR gene at equal frequencies. After 4 h, repair in
gions of the CHO genome. CHO cells were exposed to 2.5 /IM4NQO and allowed
to repair for the indicated times. A'pnl-restricted parental DNA isolated from
the transcribed strand appeared to exceed that in the nontran
scribed strand (Fig. 5B). However, at 8 and 24 h approximately
CHO cells was treated with uvrABC excinuclease where indicated. After alkaline
agarose gel electrophoreses and Southern blotting the membrane was probed for the same level of repair was observed for both strands. The
the following regions: the DHFR gene (A); the 3' DHFR region (B): the DHFR
percentage repair for both strands was approximately 50% by 8
downstream noncoding region (O: the APRT gene (D); the c-fos gene (£f);and the h and increased to approximately 70-75% by 24 h. In addition,
mitochondrial genome (F). The level of adducts in each sequence was determined
as described in "Materials and Methods" using pBR.122 as an internal standard the level of repair after 24 h in the transcribed and nontran
(not shown). scribed strands of the DHFR gene was essentially the same as
that observed using the pMB5 probe for both strands of the
shown). This finding supported the conclusion that removal of DHFR gene.
radiolabeled 4NQO adducts was associated with the DNA re
pair processes.
Gene-specific Damage and Repair. For the analysis of DNA Table 2 Repair of4NQO adducts in different gene regions in CHO cells
adducts in specific genes, DNA from cells exposed to 2.5 MM Adduct levels" at various times after 4NQO treatment
4NQO was treated (or not) with uvrABC excinuclease, sepa Oh 4 II 8h 24 h
rated on gels under denaturing conditions, transferred to a sup DHFRActive
port membrane, and probed for the specific sequence of inter 3' adjacent
±0.27 ±0.20 ±0.07 + 0.21
est. Following autoradiography, membranes were deprobed and 1.51 ±0.22 1.34 ±0.08 0.68 ±0.05 0.34 ±0.08
DownstreamAPRT 1.52
±0.081.29 ±0.271.01
1.49 ±0.060.61
0.81 0.42
0.030.44
±
reprobed for other gene sequences. Fig. 3 shows representative
autoradiograms used to quantitate the formation and repair of genec-fos ±0.14i.m ±0.410.89 0.040.56
± 0.090.32
±
4NQO adducts in various genomic regions in CHO-B11 cells. n.is1.49
geneMitochondrial1.39 ' 0.380.96
± ±0.100.55 ±0.110.44
The decrease in hybridization to labeled probe seen in samples
treated with uvrABC exinuclease indicates the presence of ±0.341.27 ±0.090.84 ±0.070.42 ±0.07
" Adduct levels in the probed sequences were determined from the intensity of
4NQO adducts in that sequence. From the proportion of DNA Southern hybridization as described in "Materials and Methods" assuming a 50%
sample incised by the excinuclease, the number of adducts per incision efficiency of uvrABC. Each value was normalized to a 10-kilobase fragment
fragment was determined. In all gene regions examined, includ size and is the mean ±SD of two or three biological experiments.
4186
 REPAIR OF 4-NITROOlllNOLINE 1-OXIDE ADDUCTS IN GENES

Hours 24

UvrABC
Fig. 4. Repair of 4NQO adducts in the tran
scribed in and nomranscribed (B) strands of Fragment Size
the DHFR gene. Membranes probed with ge-
nomic probes (Fig. 3) were reprobcd with ri-
boprobes for the transcribed and nontran- A. Transcribed Strand 14 kb
scribed strands of the DHFR genes.

B. Non-Transcribed Strand 14 kb

Table 3 4NQO adducts and removal from the transcribed and nontranscribed
strands of the DHFR gene of (HO cells
treatmentTranscribed Adduci levels" at various times after 4NQO

h1.24 h0.71 h0.38

±0.34 ±0.49 ±0.16 ±0.20
Nontranscribed 1.59 ±0.17 1.53 + 0.17 0.74 ±0.22 0.42 ±0.05 DHFR 5'
Overall DHFROh1.431.39 ±0.274 1.27 + 0.208 0.84 ±0.0724 0.42 + 0.21 DHFR31
"Membranes were probed with pMBS for the DHFR gene region (overall + Downstr
DHFR) and then reprobed with riboprobes for the transcribed and nonlranscribed -O- APRT
strands of the DHFR gene. Adduci levels were determined form the intensity of A c-fos
Southern hybridization as described in "Materials and Methods." assuming a 50% —¿it- Mitoch
incision efficiency of uvrABC. Each value was normalised to a 10-kilobasc fragment
si/e and is the mean ±SD of two biological experiments. 24

DISCUSSION
A number of approaches have been developed to measure
carcinogen adducts and repair in specific genes since the earliest
studies conducted on pyrimidine dimers. These methods are
based on the ability of specific enzymes, such as T4 endonu-
clease V and uvrABC excinuclease, or alkali treatment to cause
single-strand nicks in DNA at adduci sites which are then quan-
titated following restriction, denaturing gel electrophoresis.
Southern blotting, and probing for the sequence of interest. In
the current study, purified E. coli uvrABC excinuclease was
used to measure 4NQO adducts and repair in a number of DNA
sequences in CHO cells. The decrease in the intensity of the
hybridization upon treatment of 4NQO-modified DNA with
uvrABC is a direct reflection of the level of adducts in the
sequence. Repair is measured by the increase in the hybridiza
tion signal over time in the uvrABC-treated DNA relative to
untreated DNA. This study is the first analysis of gene-specific
and strand-specific repair of 4NQO adducts.
4NQO forms two guanine adducts, the /V2- and CK-guanine
adducts, which comprise the majority of adducts, and one ad-
enine adduci which accounts for approximately 10% of total
adducts (14, 15). Previous studies have shown that uvrABC
excinuclease recognizes the A'2- and Cx-guanine adducts of
4NQO but not the adenine adduci of 4NQO (24). Thus our
assay provides a means for quantitating the ¿V2-
and C8-guanine
adducts of 4NQO in specific sequences. For 4NQO and many
olher lesions excised by uvrABC excinuclease ihe frequency of
culling is consislenlly 33-50% of the total lesions in DNA
fragmenls of Ihe kilobase size (8, 12, 24, 25). Il also appears
lhal Ihe excinuclease recognizes 4NQO guanine adducls in a
sequence-independent manner, since studies have shown that
the level of incision by uvrABC at a 4NQO-guanine adduci
closely follows the level of modificalion at thai sile (24). In our
calculai ions of 4NQO adduci levels in genes we have used Ihe
50% value for cutting efficiency (24), but further sludies are
needed lo delermine the precise frequency. We have delermined
4187
80 r
B.
Repair Time (Mrs)
Fig. 5. Percentage repair of 4NQO adducts in different genomic sequences (A)
and in the transcribed and nontranscribed strands of the DHFR gene (B). Exam
ined in A are the DHFR. .i'DHFR. and downstream (Downstr) DHFR gene regions:
the APRT and c-fos genes: and the niitochondrial (Mitoch) genome. Examined in
B are the transcribed (7) and nontranscribed (AT) strands of the active coding
region of the DHFR gene and in the overall DHFR gene (Both). The percentage
repair was calculated for each experiment, and the values shown are (he means of
2 or 3 experiments.
thai ihe level of 4NQO guanine adducts was approximately the
same in the gene regions examined and range from 1.10 to 1.52
lesions/10 kilobases. The level measured in the gene regions
was approximately 5-7 limes higher lhan ihe level expecled
from measuremenls of adduci levels in the overall genome us
ing uvrABC excinuclease and alkaline sucrose gradient analy
sis. The initial levels of adducts in gene regions were also about
4-5 times higher lhan lhal found in ihe overall genome by
[3H]4NQO binding (which was calculaled lo be 0.3 adducts/10-
kilobase single-slrand DNA from fmol adduci bound/ng DNA
assuming an average molecular weighl of 330). The discrepancy
in adduci level measuremenls belween Ihe overall genome and
Ihe genes may be due to the different melhodologies used lo
determine adduci levels. However, the discrepancy also raises
 REPAIR OF 4-NITROQUINOL1NE
the possibility that there are regions of the genome which con
tain considerably lower levels of 4NQO adducts than the spe
cific gene regions examined here.
The results of repair studies indicate that 4NQO-guanine
adducts are not preferentially repaired in active genes. The
extent of 4NQO adduct repair in the genes examined did not
appear to correlate with their known differences in transcrip-
tional activity. The percentage of 4NQO lesions repaired in the
active DHFR and /I/"/?/" genes (70% and 65%, respectively) was
similar to the repair seen in the relatively inactive c-fos gene
(71%) (Fig. 5/4). In addition, the extent of repair was similar in
the active coding region of the DHFR gene and in the down
stream noncoding region. Interestingly, the repair of 4NQO
adducts after 24 h from the overall genome was approximately
one-third less than that observed in the genes (51% for the
genome versus 65-77% for the genes). These data again suggest
that there are regions of the genome that are not being repaired
or are being repaired with much less efficiency than the specific
genes examined here. Whether the repair heterogeneity ob
served among the genes at the 4- and 8-h repair points and
between the genes and the overall genome over 24 h has impli
cations for 4NQO-induced mutagenesis and carcinogenesis re
quires further study.
Although there are common pathways for repairing UV and
4NQO adducts (16-23), the results of the studies shown here
indicate that there are important differences in the repair path
ways operating for UV dimers and 4NQO adducts with regard
to gene-specific repair. In contrast to the results shown here for
4NQO, UV dimers are repaired faster in actively transcribing
genes than in inactive regions (1, 2), as demonstrated when
either T4 endonuclease V (1, 2) or uvrABC excinuclease (25) is
used to cut at dimer sites. While a number of studies have
shown similarities in the repair of UV dimers and 4NQO ad
ducts in the mammalian genome (19-23), differences in their
repair pathways have also been reported (22, 29-32). The di
vergence in 4NQO and UV repair pathways has been shown by
differences in the sensitivities of repair-deficient mutants of
Chinese hamster cells to 4NQO adducts and UV dimers (29)
and in studies with xeroderma pigmentosum group A cell line,
which display some capacity to repair 4NQO adducts but not
UV lesions (22, 30, 31). In another study, HeLa cell extracts
which excised UV dimers were incapable of excising 4NQO
adducts, thus suggesting that there are differences in the initial
steps of the excision repair process (32). Interestingly, differ
ences between the repair of 4NQO adducts and UV dimers is
also seen in the mitochondria! genome. In the current study, we
observed that 4NQO adducts are removed from the mitochon
dria! DNA. UV dimers, however, are not repaired in this or-
ganelle (33). Another study has shown that adducts of strepto-
zotocin are removed from mitochondrial DNA (34). Thus it
appears that mitochondria may be capable of carrying out the
repair of select adducts. The mechanisms of this repair and the
reasons for its selectivity are not known.
The results of the studies shown here comparing the repair of
4NQO adducts in transcriptionally active regions, such as the
active coding region of the DHFR gene and APRT gene, with
transcriptionally inactive regions, such as those of the down
stream DHFR region and the relatively inactive c-fos gene, sug
gest that preferential DNA repair is not a generalized feature in
the repair of bulky carcinogen adducts. While the lack of pref
erential repair of 4NQO adducts is in contrast to results with
pyrimidine dimers, the results with 4NQO are similar to the
results with jV-acetoxy-2-acetylaminofluorene and DMS. In
4188
1-OXIDE ADDUCTS IN GENES
these studies, the rate of repair of DMS adducts in the active
coding region of the DHFR gene was found to be the same as in
that found in the downstream noncoding region (9, 35). In
addition, little difference in ./V-acetoxy-2-acetylaminofluorene
repair was observed between these two regions (7).
Among all the compounds examined to date in gene-specific
studies, the most dramatic difference between repair rates in
active and inactive regions is for pyrimidine dimers. A number
of lesions have been shown to be preferentially repaired (see
Refs. 36 and 37 for a review), although not to the extent of
pyrimidine dimers. Cisplatin intrastrand adducts are preferen
tially repaired in the active DHFR gene, when compared to
inactive genes, to noncoding genomic regions, or to the overall
genome (12). Cisplatin interstrand cross-links appear to be re
paired slightly faster than the intrastrand adducts, but the cross
links do not show preferential repair after 24 h (12). In the case
of alkylating agents, although DMS damage is repaired with the
same efficiency in an active gene and in a noncoding region,
nitrogen mustard (9) and methylnitrosourea (10, 11) adducts
are preferentially repaired in the active gene. Thus among the
alkylating agents it appears that even related compounds are
repaired in different ways.
In addition to gene-specific DNA repair measurements of
4NQO adducts, we also examined the strand bias of the repair
process. Strand bias of pyrimidine dimer repair with almost
exclusive repair of the transcribed strand has now been reported
in several mammalian genes, and there is increasing evidence
that this strand bias of repair correlates with the strand bias of
mutations found in these genes (37, 38). The extent of strand
bias in the repair of a particular lesion may be related to the
degree to which the lesion blocks transcription (37). While it
has been established that UV dimers block transcription (39), it
is not known if 4NQO adducts are similarly effective in vivo.
Recent studies have indicated that a transcription-repair cou
pling factor confers strand-specific repair in E. coli (40). Gene-
specific repair, however, may not be as closely associated with
transcription as is strand-specific repair. This is supported by
some recent studies showing that the efficiency of gene-specific
repair in a given genomic region does not appear to be related
to the level of transcription (41). In the current study, we ob
served no strand bias in the repair of 4NQO adducts, suggesting
that there is no coupling of 4NQO adduct repair to transcrip
tion. A lack of strand bias in repair has been observed for DMS
(35); however, this alkylating compound has never been con
sidered to be UV mimetic. Our finding with 4NQO is in sharp
contrast to the distinct strand bias seen with pyrimidine dimer
repair (6). Thus there appear to be distinct differences in the
repair pathways operating for UV dimers and 4NQO adducts
with respect to the coupling of repair to transcription.
It is still unclear what factors govern the gene-specific and
strand-specific preferential repair of select lesions. Neverthe
less, it appears plausible that variations in the repair of different
chemical adducts in select genes could play an important role in
the basic mechanisms of malignancy. Some factors that could
influence repair in genes are the specific alterations to DNA
structure, such as the size of the lesion and the degree of dis
tortion caused by the adduct. Factors involved in the transcrip
tion of genes also appear to play a role in preferential repair.
For example, it has been shown that the class of transcription to
which a gene belongs, specifically whether the gene is tran
scribed by polymerase I or II, can influence repair in that gene
 REPAIR OF 4-NITROQUINOLINE
(42). Further studies directed toward elucidating the mecha
nisms regulating preferential repair may help to determine why
certain adducts are preferentially repaired while others are not.
ACKNOWLEDGMENTS
We thank Dr. Rodney Nairn, University of Texas, Smithville, for the
construction of plasmid pZ3d8; Dr. Susan Le Doux, University of
South Alabama, Mobile, for the mitochondrial DNA probe; and Dr.
Albert Van Zeeland, University of Leiden, The Netherlands, for a com
puter program to calculate molecular weight values in the alkaline
sucrose analysis.
REFERENCES
1. Bohr. V. A., Smith. C. A., Okumoto. D. S.. and Hanawalt. P. C. DNA repair
in an active gene: removal of pyrimidine dimers from the DHFR gene of CHO
cells is much more efficient than in the genome overall. Cell. 40: 359-369,
1985.
2. Bohr. V. A., Okumoto, D. S., Ho. L., and Hanawalt. P. C. Characterization
of a DNA repair domain containing the dihydrofolate reducÃ-asegene in
Chinese hamster ovary cells. J. Biol. Chem.. 261: 16666-16672. 1986.
3. Okumoto, D. S., and Bohr, V. A. DNA repair in the mctallolhioncin gene
increases with transcriptional activation. Nucleic Acids Res., 15: 10021-
10030, 1987.
4. Mellon, I. M., Bohr, V. A., Smith, C. A., and Hanawalt. P. C. Preferential
DNA repair of an active gene in human cells. Proc. Nati. Acad. Sci. USA, S3:
8878-8882. 1986.
5. Madhani. H. D.. Bohr. V. A., and Hanawalt. P. C. Differential DNA repair
in transcriptionally active and inactive proto-oncogenes: c-ahl and c-mos.
Cell. «Ml7-423, 1987.
6. Mellon, I. M.. Spivak. G.. and Hanawalt. P. C. Selective removal of tran
scription-blocking DNA damage from the transcribed strand of the mamma
lian DHFR gene. Cell, 51: 241-249. 1987.
7. Tang, M-s., Bohr, V. A., Zhang, X-s.. Pierce, J., and Hanawalt. P. C. Quan
tification of aminofluorene adduci formation and repair in defined DNA
sequences in mammalian cells using the UvrABC nuclease. J. Biol. Chem.,
264: 14455-14462, 1989.
8. Thomas, D. C, Okumoto, D. S.. Sanear. A., and Bohr, V. A. Preferential
DNA repair of (6-4) photoproducts in the dihydrofolate reducÃ-asegene of
Chinese hamster ovary cells. J. Biol. Chem.. 264: 18005-18010. 1989.
9. Wassermann, K., Kohn, K. W., and Bohr, V. A. Heterogeneity of nitrogen
mustard-induced DNA damage and repair at the level of the gene in Chinese
hamster ovary cells. J. Biol. Chem.. 265: 13906-13913. 1990.
10. LeDoux. S. P.. Patton. N. J.. Nelson, J. W.. Bohr. V. A., and Wilson. G. L.
Preferential DNA repair of alkali-labile sites within the active insulin gene. J.
Biol. Chem., 265: 14875-14880. 1990.
11. LeDoux, S. P., Thangada, M., Bohr. V. A., and Wilson, G. L. Heterogeneous
repair of methylnitrosourea-induced alkali-labile sites in different DNA se
quences. Cancer Res.. 51: 775-779. 1991.
12. Jones, J. C., Zhen, W. P., Reed, E.. Parker, R. J., Sanear, A., and Bohr, V.
A. Gene-specific formation and repair of cisplatin intrastrand adducts and
interstrand cross-links in Chinese hamster ovary cells. J. Biol. Chem., 266:
7101-7107, 1991.
13. Nagao. M., and Sugimura. T. Molecular biology of the carcinogen 4-nitro-
quinoline 1-oxide. Adv. Cancer Res., 23: 131-169, 1976.
14. Bailleul, B., Daubersics. P., Galiegue-Zouitina. S., and Loucheux-Lcfebvre,
M-H. Molecular basis of 4-nitroquinoline 1-oxide carcinogenesis. Jpn. J.
Cancer Res., «0:691-697. 1989.
15. Galiegue-Zouitina. S.. Bailleul. B.. and Loucheux-Lefebvre, M-H. Adducts
from in viro action of the carcinogen 4-hydroxyaminoquinoline l-oxide in
rats and from in vitro reaction of 4-acetoxyaminoquinolinc l-oxide with
DNA and polynucleotides. Cancer Res.. 45: 520-525. 1985.
16. Kondo, S., and Kato, T. Photoreactivation of mutation and killing in Escher
ichia coli. Adv. Biol. Mcd. Phys., 12: 283-298. 1968.
17. Ikenega. M., Ichikawa-Ryo, H., and Kondo. S. The major cause of inactiva-
tion and mutation by 4-nitroquinoline I-oxide in Escherichia coli: excisable
4NQO-purine adducts. J. Mol. Biol.. 92: 341-356, 1975.
18. Sanear, A., and Sanear, G. B. DNA repair enzymes. Annu. Rev. Biochem..
57: 29-67. 1988.
19. Takebe, H., Furuyama. J-I., Miki, V'.. and Kondo. S. High sensitivity of
xeroderma pigmentosum cells to the carcinogen 4-nitroquinolinc-l-oxide.
4189
l-OXIDE ADDUCTS IN GENES
Mutai. Res., 15: 98-100. 1972.
20. Ikenaga. M., Takahe, H.. and Ishii, Y. Excision repair of DNA base damage
in human cells treated with the chemical carcinogen 4-nitroquinoline 1-oxide.
Mutât.Res., 43: 415-427, 1977.
21. Zelle, B., and Bootsma. D. Repair of DNA damage after exposure to 4-nit-
roquinolinc-l-oxidc in hetcrokaryons derived from xeroderma pigmentosum
cells. Mutât.Res., 70: 373-381. 1980.
22. Tanaka. K., Takebe, H.. and Okada. V. Unscheduled DNA synthesis induced
by 4-nitroquinolin-l-oxide in xeroderma pigmentosum cells and their com
plementing heterodikaryons. Somatic Cell Genet.. 6: 739-749. 1980.
23. Walker, I. G. Lack of effect of 4-nitroquinoline l-oxide on cellular NAD
levels. Mutât.Res., 139: 155-159. 1984.
24. Thomas, D. C.. Husain. I., Chancy, S. G., Panigrahi. G. B.. and Walker, I. G.
Sequence effect on incision by (A)BC excinuclease of 4NQO adducts and UV
photoproducts. Nucleic Acids Res.. 19: 365-370. 1991.
25. Thomas, D. C., Morton, A. G., Bohr, V. A., and Sanear, A. General method
for quantifying base adducts in specific mammalian genes. Proc. Nati. Acad.
Sci. USA, 85: 3723-3727, 1988.
26. Bohr, V. A., and Okumoto. D. S. Analysis of pyrimidine dimers in defined
genes. In: E. C. Friedberg and P. C. Hanawalt (eds.), DNA Repair. A Lab
oratory Manual of Research Procedures, Vol. 3, pp. 347-366. New York:
Marcel Dekker. Inc., 1988.
27. Van Houten. B.. Masker, W. E., Carrier, W. L., and Regan. J. D. Quantita-
tion of carcinogen-induced DNA damage and repair in human cells with
UVR ABC excision nuclease from Escherichia coli. Carcinogenesis (Lond.).
7:83-87, 1986.
28. Smith, C. A., Cooper, P. D.. and Hanawalt, P. C. Measure of repair repli
cation by equilibrium sedimentation. In: E. C. Friedberg and P. C. Hanawalt
(eds.), DNA Repair. A Laboratory Manual of Research Procedures, Vol. IB.
pp. 289-305. New York: Marcel Dekker, Inc., 1981
29. Zdzienicka, M. Z., and Simons, J. W. I. M. Mutagen-scnsitive cell lines are
obtained with a high frequency in V79 Chinese hamster cells. Mutât.Res..
/ 78: 235-244, 1987.
30. Mirzayans, R.. Paterson, M. C., and Waters, R. Defective repair of a class of
4NQO-induced alkali-labile DNA lesions in xeroderma pigmentosum com
plementation group A fibroblasts. Carcinogenesis (Lond.), 6: 555-559, 1985.
31. Jones, C. J.. Edwards, S. M.. and Waters, R. The repair of identified large
DNA adducts induced by 4-nitroquinoline-1-oxide in normal or xeroderma
pigmentosum group A human fibroblasts. and the role of DNA polymerases
«or ó.Carcinogenesis (Lond.). 10: 1197-1201. 1989.
32. Panigrahi. G. B.. and Walker, 1.G. Lack of excision of 4HAQO adducts from
DNA by cell extracts that excise pyrimidine dimers. Biochem. Biophys. Res.
Commun., 140: 775-781, 1986.
33. LeDoux, S. P., Wilson, G. L., Beecham, E. J., Stevnsner, T.. Wassermann,
K.. Bohr, V. A. Repair of mitochondrial DNA after various types of DNA
damage in Chinese hamster ovary cells. Carcinogenesis. in press.
34. Pettepher, C. C., LeDoux, S. P., Bohr, V. A., and Wilson, G. L. Repair of
alkali-labile sites within the mitochondrial DNA of RINr 38 cells after ex
posure to the nitrosourea streptozotocin. J. Biol. Chem., 266: 3113-3117.
1991.
35. Scicchitano, D. A., and Hanawall. P. C. Repair of A'-methylpurines in spe
cific DNA sequences in Chinese hamster ovary cells: absence of strand spec
ificity in the dihydrofolale reducÃ-asegene. Proc. Nati. Acad. Sci. USA, 86:
3050-3054. 1989.
36. Snyderwine. E. G.. and Bohr, V. A. General and gene-specific DNA repair.
Agrie. Biotechnol. News Inform., 4.-45N-52N, 1992.
37. Bohr. V. A. Gene specific DNA repair. Carcinogenesis (Lond.), 12: 1983-
1992. 1991.
38. Chen. R.-IL, Mäher.V. M.. Brouwer. J., van de Putte, P., McCormick, J. J.
Preferential repair and strand-specific repair of benzo[a]pyrene diol epoxide
adducts in the HPRT gene of diploid human fibroblasts. Proc. Nail. Acad.
Sci. USA, 89: 5413-5417. 1992.
39. Selby, C. P.. and Sanear. A. Transcription preferentially inhibits nucleotide
excision repair of the template DNA strand in vitro. J. Biol. Chem., 265:
21330-21336. 1990.
40. Selby, C. P.. and Sanear. A. Gene- and strand-specific repair in vitro: partial
purification of a transcription-repair coupling factor. Proc. Nati. Acad. Sci.
USA, 88: 8232-8236, 1991.
41. Beecham. E. J., Mushinski. J. F.. Shacter. E.. Potter, M., and Bohr, V. A.
DNA repair in the c-m>r proto-oncogene locus: possible involvement in
susceptibility or resistance to plasmacytoma induction in BALB/c mice. Mol.
Cell. Biol., //: 3095-3104. 1991.
42. Vos, J-M. H.. and Wauthicr, E. L. Differential introduction of DNA damage
and repair in mammalian genes transcribed by RNA polymerases I and II.
Mol. Cell. Biol., //: 2245-2252. 1991.