THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269, No. 17, Issue of April 29, pp. 12749-12754, 1994 Printed in U.S.A.

Repair of Benzo(a)pyrene Diol Epoxide- and W-induced DNA Damage in Dihydrofolate Reductase and Adenine Phosphoribosyltransferase Genes of CHO Cells*

(Received for publication, December 20, 1993, and in revised form, February 17, 1994)

Moon-shong Tan&, Annie Pao, and Xu-shiao Zhang From the University of Texas M. D. Anderson Cancer Center, Science Park-Research Division, Smithville, Texas 78957

Using Uvr proteins we have quantified benzo- (a)pyrene diol epoxide (BPDE)-DNA adduct formation and repair at the dihydrofolate reductase (DHFR) and adenine phosphoribosyltransferase (APRT) genes in two Chinese hamster ovary cell lines: B-11 cells, which are SO-fold amplified for DHFR, and AT3-2 cells, which are diploid for DHFR. We have found that: 1) BPDE-DNA adduct formation in different regions of the DHFR gene is proportional to the concentration of BPDE. 2) There is no significant difference in the repair of BPDE-DNA ad- ducts between the coding and noncoding regions in ei- ther amplified or nonamplified DHFR gene domains. 3) Repair in the nonamplified DHFR gene is more efficient (3&40%) than in the amplified DHFR genes. 4) There are no significant differences of repair in the transcribed or nontranscribed strands of the DHFR gene. 5 ) BPDE-DNA adduct formation and repair in the APRT gene in B-11 and AT3-2 cells are the same. These results contrast those for the repair of cyclobutane pyrimidine dimers, which occurs preferentially in the transcribed strand of the DHFR gene and in which gene amplification appears to play no role.

Benzo(a)pyrene (BPI’ is a strong mutagen and carcinogen. The mutagenic and carcinogenic effects of BP exposure are presumably a consequence of the covalently bonding of meta- bolically activated BP derivatives such as benzo(a)pyrene diol epoxide (BPDE) with cellular DNA. Although more than a dozen BPDE isomers have been identified in vivo, the (+)- BPDE-I isomers are by far the most abundant and carcino- genic. The bonding of BPDE-I with DNA is primarily through C10 of the BPDE with the exocyclic amine of guanine residues, or to a lesser extent, with adenine residues (for review, see: Weinstein (1981), Harvey (1981), Pelkonen and Nebert (1982), and Beland and Porier (1989)). The covalently bound BPDE-I adducts tend to be positioned externally to the DNA helix and induce unwinding of the helix (Geacintov et al., 1978; MacLeod et al., 1982; Harvey and Geacintov, 1988; Undemann et al., 1983).

* This research was supported by Grant ES03124 from United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

t To whom correspondence should be addressed.

t-8-dihydroxy-9,t-10-epoxy-7,8,9,10-tetrahydro-BP; CHO, Chinese ham- ’ The abbreviations used are: BP, benzo(a)pyrene; BPDE-I, k)-r-7,

ster ovary; DHFR, dihydrofolate reductase; APRT, adenine phosphori- bosyltransferase; HPRT, hypoxanthine-guanine phosphoribosyltrans- ferase; CPD, cyclobutane pyrimidine dimers; NAAAF, N-acetoxy-2- acetylaminofluorene; endo V, endonuclease V; kb, kilobase; UNSS, UvrABC nuclease-sensitive site(s).

Although the BPDE-I adduct conformation and its impact on DNA helix structure are very different from ultraviolet ( U V ) light-induced DNA damage-cyclobutane pyrimidine dimers (CPD), these two kinds of DNA damage appear to be repaired by the same mechanisms (Seeberg et al., 1983; Tang et al., 1992). It has been found that both mammalian and prokaryotic mutant cells which are sensitive to UV-induced cytotoxicity are also sensitive to BPDE-I treatment. Moreover, those mutant cells, which are deficient in removal of CPD, are also deficient in removal of BPDE-I-DNA adducts (Yang et al., 1982; MacLeod et al, 1988).

Using T4 endonuclease (endo) V as a probe, Bohr et al. (1985) have demonstrated that the cultured Chinese hamster ovary (CHO) cells repair CPD significantly more efficiently in the coding region of the DHFR gene domain than in noncoding regions. Mellon et al. (1987) further demonstrated that the repair of CPD in DHFR gene preferentially occurs in the tran- scribed strand, These findings suggest that cells have the ca- pacity to remove DNA damage in transcriptionally active genes and that the transcription process may facilitate repair. It has been suggested that biological end points such as mutation and viability may be better related to the ability of cells to conduct gene- and strand-specific repair than the overall levels of DNA repair (Bohr et al., 1986; Bohr, 1991). Indeed, results from Chen et al. (19901, Carothers et al. (1991), Menichini et al. (1991), and Vrieling et al. (1989, 1992) suggest that mutations occur pref- erentially in the nontranscribed DNA strand in UV- or bulky chemical-treated mammalian cells.

We have recently demonstrated that the uvr system in Escherichia coli cells is the major system for repairing BPDE- I-DNA adducts and that the E. coli UvrA, UvrB, and UvrC proteins, working together (we term the collective function of these Uvr proteins the UvrABC nuclease), incise BPDE-I-DNA adducts specifically and quantitatively (Tang et al., 1992). Us- ing these enzymes and Southern DNA transfer hybridization techniques, we have developed a method to quantify BPDE-I- DNA adduct formation and repair in defined sequences in mammalian cells. We have investigated the effect of gene am- plification on the BPDE-DNA adduct formation and repair by examining these processes in the coding and noncoding regions of the DHFR gene domain in cultured B-11 CHO cells, which are amplified 50-fold for the DHFR gene, and in AT3-2 CHO cells, which are diploid for the DHFR locus. As an internal standard, we have also examined adduct formation and repair in another nonamplified APRT gene in these two cell lines. We have found that both gene amplification and transcription have significant different effects on the repair of BPDE-DNA ad- ducts and UV-induced photoproducts.

MATERIALS AND METHODS Cell Culture and Carcinogen Zkeatment-The Chinese hamster ovary

cell line, B-11 (Kaufman and Schimke, 1981) was grown in Ham’s F-12

12749

12750 Repair of BPDE-DNA Adduct in the DHFR and APRT Genes medium without glycine, hypoxanthine, and thymidine and supple- mented with 10% fetal calf serum and 500 m methotrexate to maintain the selection pressure for DHFR gene amplification. CHO AT3-2 cells were grown in a-minimum Eagle's medium with 10% fetal calf serum. Fresh confluent cells were split 1 to 16 with fresh medium. When the cells were grown to about 50-70% confluence, cells were washed and replenished with DPBS buffer (4.7 m~ MgCI,, 8.5 m~ CaCI,, 68 m~ NaCl, 1.94 m~ KCI, 1.07 m~ KH,PO,, 6.16 m~ Na,HPO,, pH 7.4). The cells were then either irradiated with UV or treated with BPDE. For BPDE treatment different amounts of BPDE (with or without 3H label) in dimethyl sulfoxide were added, and the cultures were incubated at 37 "C for 30 min. At the end of incubation, the buffer with BPDE was removed, and the cells were incubated in fresh growth medium with 5-bromo-2'-deoxyuridine (10 p ~ ) and 5-fluorodeoxyuridine (1 p ~ ) . M e r further incubation at 37 "C for 0,2,6, and 24 h, the cells were harvested and DNA was isolated.

For W irradiation, GE1518 germicidal lamp (major emission, 254 nm) was used as the W source. The cells were irradiated with UV fluence rate of 0.6-0.8 J/m2/s.

DNA Isolation-For DNA isolation cells were washed with DPBS buffer three times and lysed by incubation with lysing solution (0.5% SDS, 10 m~ Tris, pH 7.8,lO m~ EDTA, and 10 m~ NaCI) for 2 min. The cell lysates were treated with proteinase K (100 pg/ml) at room tem- perature overnight. The nucleic acids were isolated after precipitation by the addition of sodium acetate (0.33 M), 2.5 volumes of 95% ethanol, and spooled on a glass rod. The precipitated DNAs were dissolved in TE (10 m~ Tris, pH 7.5, 1 m~ EDTA) and precipitated by ethanol twice, followed by washing with 95% ethanol three times. The final pellet was dissolved in TE. RNA contaminations were reduced by treatment with RNase A (0.4 pg/ml) and RNase T1 (0.7 pg/ml) at 37 "C for 2 h. The RNase and residual proteins in the solution were removed by phenol extractions (two times) followed by diethyl ether extractions (three times), and the DNAs were ethanol-precipitated. The purified DNAs were then digested with the restriction enzyme Asp728 (1 unit/3 pg of DNA) at 37 "C overnight. The digestions were checked for completion by electrophoresis of samples on agarose gel. Replicated and nonreplicated DNAs were separated by CsCl gradient centrifugation in Ti 50 rotor, 37,000 rpm for 64 h at 21 "C. For determining the relationship between BPDE concentration and the DNA adduct formation, total DNA isolated from treated cells were used. For determining the kinetics of adduct removal, only the nonreplicated DNAs were used. This precautional step is important, since a round of semiconservative DNA replication will dilute the adduct per unit length of DNA to one-half. Moreover, it has been found that treatment of cells with DNA damaging agent in- duces significant amount of semiconservative DNA replication (Cohn et al., 1984).

The UurABC Nuclease Beatment-The Uvr proteins were prepared as described previously (Tang et al., 1992). A typical 10 x (proteidl4-kb DNA molar ratio) UvrABC reaction contains 330 nmol of UvrA, UwB, and UvrC proteins and 3 pg of DNA with 3 ng of internal standard of linearized pBR322 DNA in a final volume of 100 p1. The final reaction mixture contains 50 II~M Tris, pH 7.5, 100 m~ KCl, 1 m~ ATP, 10 m~ MgCl,, and 1 m~ dithiothreitol. The reaction mixtures were incubated at 37 "C for 90 min. At the end of incubation the protein-DNA were ethanol-precipitated with 15 pg of tRNA as carrier and washed by ethanol and resuspended in 10 pl of TE solution. T4 endo V Deatment-T4 endo V (8 x lo3 unitdml) was isolated as

described by Friedberg et al. (1980). In order to obtain optimal incision, the enzyme/DNAratio was emperically determined, and the level which renders the maximum incision in W irradiated DNA and minimum incision in control DNA was chosen. Under our conditions the optimal enzymell4-kb DNA molar ratio was 6. The DNAs were reacted with T4 endo V in a solution of 100 m~ NaC1, 10 m~ Tris, pH 7.5,0.5 m~ EDTA at 37 "C for 60 min. At the end of incubation, the protein-DNA mixture was ethanol-precipitated with 15 pg of tRNA as carrier, washed with ethanol, and resuspended in 10 pl of TE solution.

DNA Denaturation and Gel Electrophoresis-DNAs were denatured by neutral formamide treatment (Ross and Tang, 1985). A 90-pl aliquot of fresh deionized fonnamide solution was added to 10 pl of DNA s o h - tion, and the mixtures were incubated at 37 "C for 60 min. This treat- ment dissociated proteins from the protein-DNA complex. Immediately after incubation the mixtures were electrophoresed a t 5 V/cm for 3 h in a preformed 0.5% agarose horizontal gel in 0.5 x TBE buffer (25 m~ Tris, pH 7.9, 25 m~ m~ borate, and 2.5 m~ EDTA) with 0.5 pg/ml ethidium bromide. After electrophoresis, the DNA in the gels was de- purinated according to the procedure of Maniatis et al. (1982) and denatured and transferred to a Zetabind or Oncor membrane in 0.5 M NaOH and 0.6 M NaCl solution. The DNA in the membrane was subse-

A B 1-14kb-1

1

t3 I I

pMB5 cs-I4

APRT 1 9.5 kb Asp7 18

I ' 1 3.Ykb BamHl

pCAL

FIG. 1. Restriction maps of DHFR and AF'RT gene domains of CHO cells. The 5'-coding region and the 3'-downstream noncoding region of DHFR were identified by probing with pMB5 and cs-14 plas- mid DNA, respectively, and the AF'RT was identified by pGAL plasmid DNA.

quently hybridized with 32P-labeled DNA or strand-specific riboprobes as described by Bohr et al. (1985) and Mellon et al. (1987). The restriction sites of DHFR and APRT gene domains and the probing plasmid DNAs are shown in Fig. 1.

Quantitation-The autoradiographs were scanned in a Bio-Image Analyzer with a 100 Visage whole band analysis software program. The intensity of full-length fragment was normalized with internal standard pBR322 band. The number of UvrABC nuclease-sensitive sites (UNSS) and T4 endo V-sensitive sites was calculated by the Poisson distribution equation: P(0) = e-", where n is the number of UNSS or T4 endo V- sensitive sites per a full-length DNA fragment. The ratio of the normal- ized intensity of full-length fragment from UvrABC nuclease-treated sample to that of sample without enzyme treatment is equal to P(0).

RESULTS

Detection of BPDE-DNA Adducts in the DHFR Gene Domain in CHO Cells by UurABC Nuclease Incision Method-Using BPDE-modified DNA fragments, we have previously demon- strated that purified UvrA, UvrB, and UvrC proteins working together incise 6-7 bases 5' and 4 bases 3' to a BPDE- modified guanine. Moreover, these enzymes incise BPDE-DNA adducts quantitatively regardless of whether the adducts are in linear or supercoiled DNA (Tang et al., 1992). These results suggest that our UvrABC nuclease incision method may be used to quantify BPDE-DNA adduct formation and repair at defined genomic sequences.

The approaches used to detect BPDE adduct formation are similar to those we have previously used to detect N-(deox- yguanosin-C8-yl)-2-aminofluorene adducts at defined se- quences in mammalian cells (Tang et al., 1989). DNA isolated from BPDE-treated cells were digested with restriction enzyme and then reacted with optimal levels of UvrABC nuclease. The resultant DNAs were denatured in neutral formamide solution and then separated by agarose gel electrophoresis in buffer solution with moderate pH (7.9); this method eliminates the possibility of introducing nonenzymatic DNA degradation at BPDE adduct sites and abasic sites. The separated DNAs were then transferred to a nylon membrane and hybridized with 32P-labeled probes. Fig. 2, a and b, show the results of BPDE- DNA adduct formation in DHFR gene coding and 3' down- stream noncoding sequences of B-11 cells treated with concen- trations of BPDE ranging from 0 to 8 JIM. DNA isolated from cells treated with higher BPDE concentrations were more sus- ceptible to UvrABC nuclease incision, for both coding and non- coding regions. Densitometric analyses indicated that the num- ber of UNSS per unit length was the same in both coding and noncoding regions of the DHFR gene (Fig. 2). Furthermore, the final level of UNSS formation displays a linear relation to the amount of BPDE up to 4 p ~ , used for treating the cells. This relation is the same as the result obtained from the determi- nation of tritium-labeled BPDE adducts and 14C-labeled DNA. These results demonstrate that optimal levels of UvrABC nuclease can incise BPDE-DNA adducts quantitatively in DNAs isolated from mammalian cells. Treatment with 4 pM BPDE produced an average of 2.5 UNSS/14-kb length of B-11

Repair of BPDE-DNA Adduct in the DHFR and APRT Genes 12751

A BPDE(pJWt 0 , -, 2 4 , -, 8 UVRABC - 6x 8x - 6x 8x - 6~ 8x - 6x 8~

14kb

(a) pBR322

14kb -

pRR322 -

pBR322 -

B

3

m m

2

z 9

1-

m coding ~ - 1 1 o noncoding

[A by 3H specific activity ,T3-2[. coding

o noncoding P 0 1 2 3 4

BPDE (pM) FIG. 2. Formation of UNSS in the coding region and 3'-down-

stream noncoding region of the DHFR gene domain as a func- tion of BPDE concentration. DNA were prepared from B-11 cells (a and b ) or AT3-2 cells (c and d ) treated with different concentrations of BPDE (0, 2, 4 and 8 p ~ ) , restricted with Asp"', reacted with 6 and 8 x UvrABC nuclease (1 x of UvrABC nuclease renders a molar ratio of

B-ll[:

AT3-2[',

coding noncoding coding noncoding

0 12 24 40

Time (h)

noncoding (NC) regions of DHFR gene. DNA isolated from 4 PM FIG. 3. Removal of UNSS from coding (C) and 3"downstream

BPDE-treated B-11 cells or AT3-2 cells with different post-treatment incubation periods were analyzed as in Fig. 2. Signs are the same as in Fig. 2. A, typical autoradiographs; B, quantitations. The number of BPDE-DNA adducts in the coding and noncoding regions of DHFR gene is the UNSS calculated from the densitometric scanning results (aver- age of two to three experiments, except the 40-h points, which are from a single experiment). The percent of UNSS remaining in DNA was calculated based on the assignment of the number of DNA adducts a t time 0 as 1006. Solid lines represent results from B-11 cells and the broken line results from AT3-2 cells. For the purpose of clarity, some points are not precisely aligned.

cellular DNA. We observed no significant difference of BPDE- DNA adduct formation between coding and noncoding regions (Fig. 2) or between transcribed and nontranscribed DNA strands (data not shown).

A slightly lower level of BPDE-DNA adduct formation in the DHFR gene domain was observed in AT3-2 cells; 4 PM BPDE treatment produces an average of 1.6 UNSS/14-kb length of DNA. However, as in B-11 cells, UNSS formation in the DHFR gene domain of AT3-2 cells is proportional to the BPDE con- centration (Fig. 2).

The Removal of BPDE-DNA Adducts in the Coding and Non- coding Regions of DHFR Gene Domain-Having established that the UvrABC nuclease incision method can quantify the BPDE adduct a t defined sequences, we then used this method to examine the repair of BPDE-DNA adducts in the coding and noncoding regions of the DHFR gene. Nonreplicated DNAs were isolated from cells a t different incubation times after be- ing exposed to 4 p~ BPDE. Fig. 3A shows the typical autora- diographs for measurements of UNSS for coding and noncoding regions of DHFR gene domain in both B-11 and AT3-2 cells. Quantification by densitometry of these autoradiographs is shown in Fig. 3B. These results demonstrate that the repair of BPDE-DNA adducts in both coding and noncoding regions of

proteidl4-kb DNA equals one), electrophoresed, transferred to nylon membrane, and hybridized with "P-labeled probe, pMB5 (a and c). After removal of the hybridized probes, the blots were hybridized with 32P-labeled probe, cs-14 ( b and d) . Signs (-1 represent DNA without UvrABC nuclease treatment. A, typical autoradiographs; B , quantita- tions. The number of UNSS in the coding (0, W) and noncoding (0 ,O) sequences were calculated by Poisson distribution based on densitomet- ric scanning of the autoradiographs. Symbols (A) represent the number of BPDE114-kb fragment calculated from 'H specific activity. The solid line represents results from B-11 cells and the broken line results from AT3-2 cells. The results are the average of two to three experiments. For the purpose of clarity, some points are not precisely aligned.

12752 Repair of BPDE-DNA Adduct in the DHFR and APRT Genes

// nontranscribed

o nontranscribed

I. .

0 12 24 40

Time (h)

scribed (NT) strands of DHFR gene. DNAisolated from 4 y BPDE- FIG. 4. Removal of UNSS from transcribed (T) and nontran-

treated B-11 cells or AT3-2 cells with different post-treatment incuba- tion periods (2, 6, and 24 h) were probed with "P-labeled riboprobes specific for either the transcribed or nontranscribed DNA strand, as described under "Materials and Methods." Signs are the same as in Fig. 2. A, typical autoradiographs; B, quantitations. The BPDE-DNA ad- ducts in the DNA sequences are the UNSS calculated from the densi- tometric scanning of autoradiographs (average of two to three experi- ments, except the 40-h points, which are from a single experiment). The percent of UNSS remaining in DNA was calculated based on the as- signment of the number of DNA adducts a t time 0 as 100%. Solid lines represent results from B-11 cells and the broken line results from AT3-2 cells. For the purpose of clarity some points are not precisely aligned.

the DHFR gene domain in AT3-2 cells (which are diploid for the DHFR locus) is more efficient than in B-11 cells (which are amplified 50-fold for the DHFR gene). Furthermore, there ap- pears to be no significant difference in repair of BPDE-DNA adducts in coding and noncoding regions of the DHFR gene domain in both types of cells. These results contrast sharply with the repair of cyclobutane pyrimidine dimers (CDP) in CHO cells in two aspects: one is that the repair of CPD is faster and more efficient in the coding region of the DHFR gene than in noncoding regions and the other is that the rate of repair of CPD in the DHFR gene appears to be the same in cells which have DHFR gene amplified and cells which are diploid for the DHFR locus (Bohr et al., 1985).

The Removal of BPDE-DNA Adducts in the Dunscribed and Nontranscribed Strand of DHFR Gene-It has been shown that CHO cells preferentially repair CPD in the transcribed strand of DHFR gene rather than those in the nontranscribed strand (Mellon et al., 1987). These workers have concluded that the process of transcription itself may facilitate the removal of CPD. In order to determine whether there is a similar strand bias of repair of BPDE-DNA adducts, we have hybridized the same membranes used for assaying repair of coding and non- coding regions of DHFR gene domain with the strand-specific riboprobes as described by Mellon et al. (1987). The autoradio- graphic results are shown in Fig. 4A, and densitometric scan- ning results for these autoradiographs are shown in Fig. 4B. Although each individual experiment consistently shows that in B-11 cells, repair of BPDE-DNA adducts is slightly more eficient (10-15%) for the transcribed strand than the nontran- scribed strand, this difference is much smaller than that ob- served for CPD repair (Mellon et al, 1987; Table I). No signifi- cant strand bias of repair is observed in AT3-2 cells, and repair in these cells appears to be significantly more efficient than

Removal of W (20Jlm2)-DNA damage after 24-h incubation in CHO TABLE I

B-11 and AT3-2 cells in coding uersus noncoding regions and transcribed uersus nontranscribed strands of the DHFR gene and

APRT gene, as detected by T4 endo V and UvrABC nuclease C, coding; NC, noncoding; T, transcribed; NT, nontranscribed.

Gene I repair"

T4 endo V UVTABC

DHFR (B-11) C 82 2 3 (5)' 81 2 5 (7, NC 30 f 3 (6) T

83 f 3 (5)

NT 100 f 0 (4) 98 t 6 (5) 43 f 6 (4) 77 f 13 (6)

APRT (B-11) 67 2 12 (4) 93 f 12 (3) DHFR (AT3-2)

C 64 2 2 (5) NC

73 f 10 (4) 11 2 7 (4)

T 64 2 2 (2)

85 4 (3) NT

66 2 1 (2) 21 2 3 (2) 50 f 9 (2)

APRT (AT3-2) 65 f 5 (2) 82 2 4 (3)

Values are the average -c S.D. * The number of independent experiments is in parentheses.

repair in B-11 cells. Detection of the Formation and Removal of BPDE-DNA Ad-

ducts in the APRT Gene-The more efficient BPDE-DNA ad- duct removal observed for the unamplified DHFR gene ofAT3-2 cells in comparison with the amplified DHFR genes in B-11 cells could either be due to a difference in repair efficiency between two CHO cell lines or the effect of gene amplification. If the former is the case, then we would expect that the differ- ences observed in the case of the DHFR gene would also be observed for other genes and DNA regions between these two cell lines. In order to distinguish these two possibilities, we examined the repair of BPDE-DNA adducts at another nonam- plified gene locus in these two cell lines, the APRT gene. The same membranes used for detection of BPDE-DNA adducts in the DHFR gene were deprobed and later reprobed with 32P- labeled APRT gene DNA. Typical autoradiograms are shown in Fig. 5A, and densitometric scanning results are shown in Fig. 5B. These figures demonstrate that the repair of BPDE-DNA adducts in the APRT gene is the same in both B-11 and AT3-2 cells, even though the kinetics of repair for this gene are dif- ferent from those for the DHFR gene. Therefore, the lower extent of removal of BPDE-DNA adducts in DHFR genes of B-11 cells in comparison with AT3-2 cells appears to be a con- sequence of DHFR gene amplification in these cells.

Detection of UV-induced DNA Damage in DHFR Gene Do- main by UvrABC Nuclease and T4 endo V Incision Methods- Since Bohr et al. (1985) and Mellon et al. (1987) have unam- biguously demonstrated that CHO cells preferentially repair T4 endo V-sensitive CPD in the transcribed strand of active genes, how does one explain the fact that we detected a much smaller extent of strand-specific preferential repair for BPDE- DNA adducts in B-11 cell and that we detected no preferential repair at all in AT3-2 cells? One possible explanation could be that the transcription process has less effect on the removal of BPDE-DNA adducts than for CPD. A second possibility is that the UvrABC nuclease incision method is relatively insensitive in detecting gene- or strand-specific preferential repair. A third possibility is that, for unknown reasons, cells grown under our conditions do not have the ability to repair DNA damage in transcriptionally active genes. To date, the most distinct strand- and gene-specific repair is observed when T4 endo V is used for CPD detection. It is possible that transcription may have a profound effect on the removal of CPD but much less of an effect on the removal of other kinds of DNA modifications. On the other hand, since UvrABC nuclease is capable of recog-

Repair of BPDE-DNA Adduct in the DHFR and APRT Genes 12753

B 100

I B-11 0 AT3-2

1-50

I I I I I

0 12 24 I I '0

Time (h) FIG. 5. Detection of the remuval of UNSS in APRT gene in B-11 (.) and AT3-2 (0) cell lines treated with BPDE (4 p"). The same

membranes used for detection of BPDE removal in DHFR gene in Figs. 3 and 4 were deprobed and reprobed with "P-labeled pGAL DNA, and the details of the experiments and the quantifications are as described in the text. A, a typical autoradiograph; B, the quantitations.

nizing a wide range of DNA damage while T4 endo V recognizes only cyclobutane pyrimidine dimers, it is possible that less preferential repair is detected by the UvrABC nuclease incision method, because this enzyme system incises several different types of BPDE-induced DNA damage, including some whose removal is not affected by the process of transcription. In order to distinguish these possibilities, we have examined the repair of UV-induced DNA damage in coding versus noncoding and transcribed uersus nontranscribed DNA strands in B-11 cells and AT3-2 cells. Exponentially growing cells (50% confluence) were W-irradiated, and their DNAs were purified and pro- cessed as for BPDE-treated cells. UV damage was detected either as T4 endo V or UvrABC nuclease-sensitive sites. The autoradiograph results are shown in Fig. 6, and the densito- metric scanning results are shown in Table I. These results clearly demonstrate that there is a significant amount of gene- and strand-specific repair of UV-induced DNA damage when T4 endo V is used for detection. In contrast, the extent of gene- and strand-specific preferential repair is markedly reduced when UvrABC nuclease is used for detection. Moreover, there is no significant difference for the repair of photoproducts at the DHFR gene domain detected by T4 endo V or by UvrABC nucle- ases between CHO-B-11 cells in which the DHFR gene is 50- fold amplified and CHO-AT3-2 cells which are diploid for the DHFR locus. These results rule out the trivial explanation that cells grown under our conditions are incapable of preferential repair of CPD in the transcribed strand of an active gene. More importantly, these results suggest that the effects of gene am- plification and transcription on the repair of CPD and BPDE- DNA adducts may be fundamentally different.

DISCUSSION

We have previously demonstrated that UvrABC nuclease can incise BPDE-DNA adduct specifically and quantitatively (Tang et al., 1992). In this report, we have used UvrABC nuclease as

DHFR

I T- T

NT 6 NT

APRT

~ -. . .. .. Y

FIG. 6. Detection of UV-DNA damage at DHFR and APRT gene domain in UV-irradiated (20 J/m2) CHO cells with or without 24 h of post-irradiation incubation. The DNA isolated from these cells were treated with T4 endo V or with UvrABC nuclease and then dena- tured, separated by electrophoresis, and transferred to a Zetabind mem- brane. The membranes were then probed for coding ( C ) versus noncod- ing regions ( N C ) and transcribed (2') versus nontranscribed (NT) strands of the DHFR gene and the APRT gene using the same method as described in Figs. 3 and 4. The signs and symbols are the same as in Figs. 3 and 4. M i c a 1 autoradiographic results from B-11 cells are shown.

a reagent to quantify the BPDE-DNA adduct formation in de- fined genomic sequences in mammalian cells. This approach has been successfully used for measuring DNA damage induced by N-acetoxy-2-acetylaminofluorene (NAAAF) (Tang et al., 1989) and (6-4) photoproducts (Bohr et al., 1991) in defined sequences in mammalian cells. However, Thomas et al. (1988) reported that under their conditions, UvrABC nuclease de- tected only one-third of CPD and psoralen-DNA adducts. I t is possible that some of these discrepant results may reflect the differential efficiency of UvrABC nuclease in incising different kinds of DNA damage. In order to obtain quantitative results, it is imperative to optimize reaction conditions so that reactions can go to completion. It has been shown that in the absence of DNA repair synthesis and ligation, UvrABC nuclease remains bound to damaged DNA after the incision reaction; conse- quently, under in vitro reaction conditions, incision by UvrABC nuclease is stoichiometric rather than catalytic (for review, see Sancar and Tang (1993)). Therefore, to insure complete incision of all DNA adducts, the nuclease must be present in significant excess over the number of DNA adducts. Empirically, we have found that reactions containing molar ratio 6 to 8 of UvrABC nuclease over 14-kb DNA fragment yield quantitative results.

Our finding that BPDE-DNA adducts are repaired with the same efficiency in coding and noncoding regions of the DHFR gene domain, in both B-11 and AT3-2 cells, contrasts sharply with the results for repair of CPD, but is consistent with results for the repair of NAAAF (Tang et al., 1989)-, dimethyl sulfate (Scicchitano and Hanawalt, 1989, 1990; Wassermann et al., 1990)-, and 4-nitroquinoline-1-oxide (Snyderwine and Bohr, 1992)-induced damage in the same genomic regions. These findings suggest that "gene-specific preferential repair" may be highly dependent upon the structures of the DNA damage.

Our results showed that BPDE-DNA adducts are repaired much more efficiently in the amplified DHFR gene domain than in bulk DNA (MacLeod et al., 1988), and the degree of repair is even higher in the DHFR gene of AT3-2 cells, which are diploid for the DHFR gene locus. If gene expression induces openness of a chromosomal domain, including the regions flanking the expressed gene region, and if this openness of chromosome increases the accessibility of repair complexes, then perhaps this is the reason that higher repair is observed in active genes such as DHFR, as well as their nearby noncoding regions. Since the amplified DHFR genes are not necessarily all located in the same chromosomal domain, and may not be uniformly transcribed, the overall repair level observed for the

12754 Repair of BPDE-DNA Adduct in the DHFR and APRT Genes amplified DHFR genes may depend on the proportion of the genes which are actively transcribed. Our observation that the efficiency of repair of BPDE-DNA adducts in amplified DHFR genes is higher than that of bulk DNA, but lower than that of a nonamplified DHFR locus, suggests that many of the DHFR genes in an amplified array may not be transcriptionally active. Our finding that B-11 and AT3-2 CHO cell lines repair BPDE- DNA adducts in APRT gene with the same kinetics and effi- ciency rule out the trivial explanation that the observed differ- ences in BPDE-DNA adduct removal in amplified and nonamplified DHFR genes is simply due to variations in repair efficiency between cell lines.

Previously, we also observed neither gene- nor strand-spe- cific repair for N M - i n d u c e d DNA damage (Tang et al., 1989).' However, using the same UvrABC method, we were able to detect strand-specific repair for W-induced DNA dam- age. We also confirmed that the repair of CPD, detected by T4 endo V, in amplified and nonamplified DHFR genes is the same (Fig. 6 and Table I). These results suggest that the effects of the process of transcription and the structure of chromosomes on the repair is highly dependent on the structures of the DNA damage per se .

Using the UvrABC nuclease and the same neutral formam- ide denaturing method as we described previously (Tang et al., 1989), Chen et al. (1992) have reported a greater extent of transcribed strand biased repair (11-29%) for BPDE-DNA ad- ducts in hypoxanthine-guanine phosphoribosyltransferase (HPRT) gene in human fibroblasts. This could reflect intrinsic difference in repair between human fibroblasts and CHO cells and/or from differences in repair between HPRT and DHFR genes. It is also possible that BPDE treatment has less effect on the transcription of HPRT gene than on that of DHFR gene in these two cell systems. However, consistent with their findings of preferential repair of BPDE-DNA adducts in the transcribed strand of HPRT gene, these workers observed a higher fre- quency of mutations in the nontranscribed strand than in the transcribed strand (Chen et al., 1990). They concluded that strand-biased repair is the cause of this strand-biased muta- tion. Intriguingly, Vrieling et al. (1989, 1992) and Menichini et al. (1991) found that the UV-induced mutations in the HPRT gene of UV-sensitive CHO mutant cells preferentially occurs in the replication leading strand which is the transcription tem- plate strand. Recently, Melchior et al. (1992) reported that in the prokaryotic system the aromatic amine-DNA adducts in- duce the same mutation frequency in the transcribed and non- transcribed strands, and they also found that the major muta- tion (G + T transversions) induced by 1-aminopyrene and acetylbenzidine occur preferentially at the transcribed strand. These results indicate that factors such as the structure of the DNA damage, its impact on the helix structure, and the loca- tion of the damage relative to the direction of DNA replication may all contribute greatly to mutagenesis. Furthermore, strand-biased effects on mutagenesis may reflect either a strand bias in fidelity of DNA replication or a strand bias in repair. These factors together may determine the lethality of the DNA damage.

Our results that UvrABC nuclease detects significantly less U V damage than T4 endo V at 3' downstream region and non- transcribed strand of DHFR gene in W-irradiated cells with 24-h post-irradiation incubation, whereas both enzyme systems detect the same amounts of damage within the DHFR gene domain in cells without post-irradiation incubation, are in-

M.-S. Tang and I. Mellon, unpublished results.

triguing. I t is possible that T4 endo V and UvrABC nuclease incise different photoproducts, and cells repair the T4 endo V-sensitive photoproducts and UvrABC nuclease-sensitive pho- toproducts at different rates. The number of T4 endo V-sensi- tive photoproducts and the number of UvrAEC nuclease-sensi- tive photoproducts in the DNA isolated from cells immediately after W irradiation may be fortuitously the same. Alterna- tively, both of these two enzymes may primarily detect CPD, but modification of CPD during the post-irradiation period may result in modified CPDs which are still sensitive to T4 endo V incision but resistant to UvrABC nuclease. The ability of the rodent cells to resume DNA replication before a significant fraction of the T4 endo V-sensitive sites have been removed has long been a puzzle. Perhaps such a CPD modification mecha- nism could serve as a means of increasing the probability of cell survival by allowing the DNA replication to proceed at the expense of fidelity.

Acknowledgements-We thank Drs. G. Adair, I. Mellon, and R. Hewitt for critical reviewing this manuscript and Drs. J. Hamlin, L. Chasin, I. Mellon, R. Nairn, and G. Spivak for providing plasmid probes.

REFERENCES

Beland, F. A.. and Poirier, M. C. (1989) in The Pathobiology ofNeoplasia (Sirica, A.

Bohr, V. A. (1991) Carcinogenesis 12, 1983-1992 E., ed) pp. 57-76, Plenum Press, New York

Bohr, V. A,, Smith, C. A,, Okumoto, D. S., and Hanawalt, P. C. (1985) Cell 40,

Bohr, V. A., Okumoto, D. S., and Hanawalt, P. C. (1986) Proc. Natl. Acad. Sci.

Carothers, A. M., Mucha, J. , and Grunberger, D. (1991) Pmc. Natl. Acad. Sci.

Chen, R.-H., Maher, V. M., and McCormick, J. J. (1990) Proc. Natl. Acad. Sci.

Chen, R.-H., Maher, V. M., Brower, J., van de Putte, P., and McComick, J. J. (1992)

Cohn, S. M., Krawisz, B. R., Dresler, S. L., and Lieherman, M. W. ( 1984) Proc. Natl.

Friedberg, E. C., Ganesan, A. K., and Seawell, P. C. (1980) Methods Enrymol. 65,

Geacintov, N. E., Cagliano,A., Ivanovic, V, and Weinstein, I. B.(1978)Biochemistry

Harvey, R. (1981) Acc. Chem. Res. 14, 218-226 Harvey, R. G., and Geacintov, N. E. (1988) Acc. Chem. Res. 2 1 , 6 6 7 3 Kaufman, R. J., and Schimke, R. T. (1981) Mol. Cell. Biol. 1, 1069-1076 MacLeod, M. C., Mansfield, B. K., and Selkirk, J. K. (1982) Carcinogenesis (Land.)

MacLeod, M. C., Adair, G., and Humphrey, R. M. (1988) Mutat. Res. 199,243-254 Maniatis, T., Fritsch, E. F., and Sambmok, J. (1982) Molecular Cloning: A Labo-

Melchior, W. B., Jr., Lindsey, L. A,, and Beland, E A. (1992) Proc. Am. Assoc. Cancer

Mellon, I., Spivak, G., and Hanawalt, P. C. (1987) Cell 51, 241-249 Menichini, P., Vrieling, H., and van Zeeland, A. A. (1991) Mutat. Res. 251,143-155 Pelkonen, O., and Nebert, D. W. (1982) Pharmacol. Rev. 38, 189-221 Ross, J., and Tang, M. (1985) Anal. Biochem. 144,212-217 Sancar, A,, and Tang, M. (1993) Photochem. Photobiol. 57,905-921 Scicchitano, D. A,, and Hanawalt, P. C. (1989) Proc. Natl. Acad Sci. U. S. A. 86,

Scicchitano, D. A,, and Hanawalt, P. C. (1990) Mutat. Res. 233 ,3147 Seeberg, E., Steinurn, A,-L., Nordenskjokd, M., Soderhall, S.. and Jernstrom, B.

Snydenvine, E. G., and Bohr, V. A. (1992) Cancer Res. 52,41834189 Tang, M., Bohr, V. A., Zhang, X., Pierce, J., and Hanawalt, P. C. (1989) J. Biol.

Tang, M., Pierce, J. R., Doisy, R. P., Nazimiec, M. E., and MacLeod, M. C. (1992) Chem. 264, 14455-14462

Thomas, D. C., Morton, A. G., Bohr, V. A,, and Sancar, A. (1988) Pmc. Natl. Acad. Biochemistry 31,8429-8436

Undemann, 0.. Lycksell, P., Graslund,A..Astlind, T., Ehrenberg,A.. Jernstrom, B., Sci. U. S. A. 85,3723-3727

Vrieling, H., van Rooijen, M. L., Groen, N. A,, Zdzienicka, M. Z. , Simons, J. W. I. M., Tjerneld, F., and Norden, B. (1983) Cancer Res. 43, 1851-1860

Vrieling, H., Zhand, L.-H., van Zeeland, A. A,, and Zdzienicka, M. Z. (1992) Mutat. Lohman, P. H. M., and van Zeeland, A. A. (1989) Mol. Cell. Biol. 9, 1277-1283

Wassermann, K., Kohn, K. W., and Bohr, V. A. (1990) J. Biol. Chem. 265, 13906 Res. 274,147-155

13913 Weinstein, I. B. (1981) J. Supmmol. Struct. Cell. Biochem. 17, 99-120 Yang, L. L., Maher, V. M., and McCormick, J. J. (1982) Mutat. Res. 94,435447

359-369

U. S. A. 83,3830-3833

U. S. A. 88,5749-5753

U. S. A. 87,868G3684

Proc. Natl. Acad. Sci. U. S. A. 89, 5413-5417

Acad. Sci. U. S. A. 81,4828-4832

191-201

17,525C5262

3, 1031-1037

ratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Res. 33, 175

3050-3054

(1983) Mutat. Res. 112, 139-145