A sensitive, enzymatic assay for the detection of closely opposed cyclobutyl pyrimidine dimers induced in human diploid fibroblasts

187

Mutation Research, 166 (1986) 187-198 DNA Repair Reports Elsevier MTR 06166

A sensitive, enzymatic assay for the detection of closely opposed cyclobutyl pyrimidine dimers induced in human diploid fibroblasts L u n H . L a m * a n d R i c h a r d J. R e y n o l d s ** Laboratory of Radiobiology, Department of Cancer Biology, Harvard University School of Public Health, 665 Huntington Ave., Boston, MA 02115 (U.S.A.) (Received 12 October 1985) (Revision received 3 January 1986) (Accepted 3 March 1986)

Summary

A sensitive, enzymatic assay has been developed for the detection of closely opposed cyclobutyl pyrimidine dimers induced in UV-irradiated human diploid fibroblasts. In this assay closely opposed dimers are detected as bifilar enzyme-sensitive sites. Single-strand incisions are made at the positions of individual pyrimidine dimers through the action of M . luteus pyrimidine dimer-DNA glycosylase. Incisions at closely opposed dimers, effectively expressed as double-strand breaks, are quantified from the resulting reduction in DNA double-strand molecular weight as determined by velocity sedimentation through neutral sucrose density gradients. The stability of the bacteriophage X cos site under our reaction conditions indicates that opposed incisions must be relatively close to be expressed as a double-strand break. The dose response for the induction of bifilar enzyme-sensitive sites in mammalian cells was found to be complex but can be approximated by a function that increases as the 1.2-1.4 power of UV dose. The frequency of bifilar enzyme-sensitive sites observed decreased during postirradiation incubation of excision-repair-proficient human diploid fibriblasts with less than 20% still detectable at 24 h after irradiation with 5 J / m 2 (254 nm). By comparison, over 80% of the bifilar enzyme-sensitive sites induced in fibroblasts assigned to xeroderma pigmentosum complementation group A remained detectable 24 h after irradiation. The implications of these results for models addressing the induction and repair of closely opposed pyrimidine dimers are discussed.

In its simplest form incision repair operating at pyrimidine dimers is thought to consist of a series of enzymatic steps initiated by the recognition of a dimer as an abnormal structure and incision at or near this DNA lesion. Incision of the damaged strand is followed by excision of the altered region * Present address: Department of Genetics, Stanford University School of Medicine, Stanford, CA 94305, U.S.A. ** Present address: Mail Stop M886, Life Sciences Division, Los Alamos National Laboratory, Los Alamos, NM 87545, U.S.A. Please address all correspondence to Dr. Reynolds.

and gap closure through the actions of a DNA polymerase and polynucleotide ligase. An important aspect of this model is the conservation of genetic information in the complementary strand (Haynes, 1966). Due to the redundancy built into the nucleotide sequence of DNA double helices, genetic information is not lost provided that only one of the two DNA strands is damaged within a limited region. The induction of closely opposed dimers is expected to present a more complex situation for excision repair. Dimers in the template strand may

0167-8817/86/$03.50 © 1986 Elsevier Science Publishers B.V. (Biomedical Division)

188 disrupt normal DNA replication in vitro (Villani et al., 1978; Doubleday et al., 1981; Moore et al., 1981; Chan and Haseltine, 1983), and in vivo (Rupp and Howard-Flanders, 1968; CailletFauquet et al., 1977; Lehmann, 1979; Park and Cleaver, 1979; Clark and Hanawalt, 1983). Excision extending past a dimer in the complementary strand may also be disruptive for normal repair synthesis (Harm, 1968; Rupp and HowardFlanders, 1968; Cleaver, 1977; Moore et al., 1981). Simultaneous excision at closely opposed dimers may result in overlapping gaps (Harm, 1968; Setlow, 1968). The resulting double-strand breaks may cause cell death (Harm, 1968; Moss and Davies, 1974; Bonura and Smith, 1975a,b), or require additional repair functions similar or identical to those thought to be necessary for the repair of X-ray-induced double-strand breaks (Krasin and Hutchinson, 1977; Resnick and Moore, 1979) or interstrand crosslinks (Cole et al., 1978). Information pertaining to the induction, repair and biological consequences of closely opposed pyrimidine dimers is limited. Bonura and Smith (1975a) observed the appearance of double-strand breaks in excision-repair-proficient, UV-irradiated E. coll. Double-strand breaks were not observed with excision-deficient E. coli under the same conditions. Similar observations with human fibroblasts have been reported by Bradley and Taylor (1981). In both studies excision repair operating at closely opposed dimers was proposed to explain the appearance of double-strand breaks. The induction and/or repair of closely opposed dimers has been invoked to explain dose rate and dose fractionation effects (Harm, 1968), loss of photoreactivability (Moss and Davies, 1974), induction of SOS repair (Sedgwick, 1976), and UVinduced cellular lethality through the exponential region of survival curves (Harm, 1968; Moss and Davies, 1974; Chadwick and Leenhouts, 1983). Despite the apparent interest in closely opposed dimers, however, quantitative assays for their detection have been unavailable. It has been observed previously that doublestrand breaks are formed during the incubation of UV-irradiated DNA with the pyrimidine dimer (PD)-DNA glycosylase from M. luteus (Strauss et al., 1966; Meneghini et al., 1981) or bacteriophage

T4 (Minton and Friedberg, 1974; Lloyd et al., 1980). At least 80% of such double-strand breaks are apparently due to PD-DNA glycosylase cleavage at closely opposed dimers (Lam and Reynolds, 1986). Based upon these findings we have developed a sensitive and quantitative assay that detects closely opposed dimers as bifilar (doublestranded) enzyme-sensitive sites (ESS). This assay has been utilized to monitor the disappearance of closely opposed dimers in excision-repair-proficient or -deficient human fibroblasts. Materials and methods

Cell strains

Two strains of human diploid fibroblasts were used in this study. AG1522 cells were derived from the foreskin of a 3-day-old male and have normal sensitivity to ultraviolet radiation (254 nm) (Little et al., 1980). GM2990 (XPICA) cells were derived from an 8-year-old female afflicted with xeroderma pigmentosum (XP) (Hashem et al., 1980). GM2990 cells exhibit abnormal sensitivity to UV radiation and have been assigned to XP complementation group A (Hashem et al., 1980; Grosovsky and Little, 1983). These strains were obtained from the NIA Aging Cell Repository (AG1522) or the NIGMS Human Genetic Mutant Cell Repository (GM2990) (Institute for Medical Research, Camden, N J). Growth and labeling conditions

Cells were routinely cultured in Eagle's minimal essential medium (Gibco Laboratories, Grand Island, NY) supplemented with 10% or 15% heat-inactivated fetal calf serum (M.A. Bioproducts, Walkersville, MD), 100 units/ml penicillin (Gibco Laboratories), 100 /~g/ml streptomycin (Gibco Laboratories), and 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) (pH 7.2 at 37°C). Ceils were grown at 37°C under a humidified atmosphere of 5% CO 2 and 95% air. To label cellular DNA individual 60-mm-diameter polystyrene tissue culture dishes (Falcon Plastics, Oxnard, CA) were seeded with approximately 2 × 105 cells. Cells were allowed to attach and grow for 18 h prior to being labeled for 24 h in media containing either 0.25 /~Ci/ml [Me-3H]thymidine (20 Ci/mmole; New England Nuclear,

189 Boston, MA) or 0.125/xCi/ml [Me-14C]thymidine (51 mCi/mmole; New England Nuclear).

UV irradiation and postirradiation treatments Prior to irradiation radioactive medium was removed, cells were rinsed twice with 2.5 ml/dish of calcium- and magnesium-free Earle's balanced salt solution without phenol red (EBSS: 116 mM NaC1, 5.4 mM KC1, 1 mM NaHzPO 4, 2.9 mM glucose, 26.2 mM NaHCO 3, pH 7.3) and then covered with an additional 2.5 ml/dish of EBSS. UV radiation was provided by an array of five G8T5 germicidal lamps (General Electric Company, Cleveland, OH) at an incident dose rate of 0.35 W / m 2 as determined with an IL254 germicidal photometer (International Light, Inc., Newburyport, MA). After irradiation the EBSS was removed and cells, still attached to the plastic culture dishes, were either frozen immediately at 90°C or were covered with 4 ml of medium and returned to 37°C for repair periods of various durations. Upon the completion of each repair period the medium was removed, cells were rinsed with EBSS and frozen at - 9 0 ° C as indicated above. Dishes were stored at - 9 0 ° C until just prior to cell lysis and DNA purification. -

Cell lysis and DNA purification Cell were lysed by the addition of 1.5 ml of a solution containing 10 # g / m l proteinase K (Boehringer Mannheim Biochemicals, Indianapolis, IN), 0.5% (w/v) sodium N-lauroyl sarcosine, 100 mM NaC1, 10 mM sodium ethylenediaminetetraacetate (EDTA), 10 mM tris (hydroxymethyl)aminomethane-(Tris-) HC1, pH 7.6, to each dish. In double-label experiments, cells in the two dishes to be pooled were lysed in succession with the same 1.5 ml of lysing solution. Cell lysates were gently transferred to 15 ml Corex tubes and then incubated for 1 h at 37°C to permit digestion of cellular proteins by proteinase K. Lysates were further deproteinized by extraction with 1.5 ml of neutralized, redistilled phenol. Although extractions were routinely accomplished with gentle agitation to permit the recovery of large molecular weight DNA, samples were vortexed to reduce DNA molecular weights where indicated. Following the separation of the aqueous and organic phases by low speed centrifugation at 4°C, a

400-/~1 sample of the aqueous phase was dialyzed against 3 changes (23 ml each) of PD-DNA glycosylase reaction buffer (50 mM NaC1, 1 mM EDTA and 10 mM Tris-HC1, pH 7.6) for durations of approximately 2, 6 and 14 h.

Detection and quantification of unifilar and bifilar ESS Unifilar and bifilar ESS were detected and quantified as described previously (Lam and Reynolds, 1986). Briefly, 0.2-0.4/xg of dialyzed DNA in 90 /~1 of PD-DNA glycosylase reaction buffer was incubated with and without PD-DNA glycosylase prepared from M. luteus by a method modified from that of Carrier and Setlow (1970) in which nucleic acids were removed by precipitation with streptomycin sulfate. The preparation used in this study corresponded to fraction II of Carrier and Setlow (1970). Incubations were for 60 min at 37°C unless otherwise indicated. Samples were analyzed for the presence of unifilar or bifilar cleavage by velocity sedimentation through precalibrated alkaline or neutral sucrose density gradients. Gradients were calibrated against linearized SV40 (Bethesda Research Laboratories, Gaithersburg, MD), bacteriophage T2, and bacteriophage T7 DNAs. Bacteriophage T2 and T7 were kindly provided by Dr. Thomas Bonura of Stanford University and Dr. William Studier of Brookhaven National Laboratory, respectively. Velocity sedimentation was at 20°C in Sorvall model OTD-2 preparative ultracentrifuges (DuPont Instruments, Inc., Newtown, CT) equipped with SW 50.1 rotors (Beckman Instruments, Inc., Palo Alto, CA). Rotor speeds and sedimentation times were adjusted to place final DNA distributions relatively near to the center of each gradient. To avoid speed-dependent artifacts (Zimm, 1974), sedimentation through neutral gradients was at rotor speeds of less than 25 000 or 10000 rpm for vortexed or unvortexed samples, respectively. Gradients were fractionated and analyzed as described previously (Lam and Reynolds, 1986). For each sample the frequency of ESS was calculated from a comparison of DNA molecular weights before and after t~eatment with PD-DNA glycosylase (Reynolds, 1978). Stability of a 12-base-pair overlap Bacteriophage k DNA (Bethesda Research

190 Laboratories) in 10 mM MgCI 2, 50 mM NaC1 and 100 mM Tris-HC1, p H 7.5 was restricted with EcoR I (Bethesda Research Laboratories). Digestion of A D N A with EcoR I results in 5 restriction fragments with the cos site (Wu and Taylor, 1971) situated asymmetrically in the largest fragment. After purification by phenol extraction and ethanol precipitation the restricted DNA was redissolved at a concentration of 200 /~g/ml in annealing buffer (1 M NaC1, 10 mM MgC12 and 100 mM Tris-HC1, pH 8.0). To ensure maximum annealing at the cos site the restriction fragments were incubated at 42°C for 90 h and then allowed to cool slowly to 35°C. After adjustment of the buffer to 100 mM NaC1, 10 mM MgC12 and 100 mM Tris-HC1, p H 7.5, the 3' termini of the restriction fragments were labeled at the EcoR I cleavage sites by incubation with the Klenow fragment "of E. coli D N A polymerase I (Bethesda Research Laboratories) in the presence of [a32p]TTP (684 Ci/mmole; New England Nuclear) and [a-32p]dATP (684 Ci/mmole; New England Nuclear). Unincorporated label was eliminated by gel exclusion chromatography through Sephadex G-50 M (Pharmacia, Inc., Piscataway, N J) in a spin column followed by ethanol precipitation of the DNA. Prior to testing the effects of various reaction conditions on the stability of the cos site, the restricted, end-labeled D N A was redissolved in annealing buffer, incubated at 42°C for 65 h and allowed to cool slowly to 35°C as described above. The annealing buffer was replaced with P D - D N A glycosylase reaction buffer by repeated dialysis against the reaction buffer. DNA samples (0.1 /~g of D N A in 18 #1 of reaction buffer) were then incubated at 37°C for 1 or 2 h in the presence of either 2/xl of the M. luteus P D - D N A glycosylase prepared as described above, or 2 /~1 of the P D - D N A glycosylase storage buffer (10% (v/v) ethylene glycol, 0.1 mM 2-mercaptoethanol, 10 m M Tris-HC1, p H 8.0). Reactions were terminated by the addition of 1 #1 of 10% (w/v) N-lauroyl sarcosine and 1 /~1 of 1 m g / m l proteinase K followed by incubation for 60 rain at 37°C. Where indicated individual samples were also subjected to mild denaturing conditions (68°C, 15 min) to disrupt the cohesive ends of the ~ cos site. Disruption of the cos site was monitored by

electrophoresis through 20 cm × 25 cm × 0.4 cm (W x L × D) 0.4% agarose gels (2 mM EDTA, 89 mM boric acid, 89 mM Tris base, pH 8.3). 5/~1 of gel loading solution containing 15% (w/v) Ficoll 400, 0.25% (w/v) bromophenol blue, 0.25% (w/v) xylene cyanol were added to each 22-/~1 sample and 6 ~1 of each sample was loaded onto the agarose gel. The remainder of each sample was heated to 68°C for 15 min to disrupt the cos site cohesive ends. 6/~1 of each denatured sample were loaded onto the agarose gel in lanes adjacent to their nonheated counterparts. After electrophoresis at 50 V for 24 h, DNA was fixed in the gel by soaking it in 10% (v/v) glacial acetic acid for 1 h. The gel was then rinsed for 1 h with several changes of deionized water and blotted until nearly dry with Whatman 3MM paper. One side of the gel was covered with Saran wrap and a sheet of Kodak XAR-5 X-ray film was pressed against the Saran wrap. The film was exposed for approximately 6 h at room temperature. Results Reaction kinetics for the expression of bifilar E S S as double-strand breaks

To determine the rate at which bifilar ESS are expressed as double-strand breaks, DNA samples from cells irradiated with 600 J / m 2 were incubated at 37°C with either 10 or 20 ~1 of M. luteus extract for various durations from 0 to 150 min. A time-dependent increase in the frequency of bifilar ESS expressed as double-strand breaks was observed for approximately 120 min (Fig. 1). Thereafter the frequency of ESS detected remained relatively constant. Longer incubations, second additions of P D - D N A glycosylase followed by further incubation and additional P D - D N A glycosylase throughout the entire reaction period were all without effect on the final frequency of bifilar ESS detected. This final frequency was also unaffected by preincubation of the P D - D N A glycosylase for up to 30 rain under conditions similar to those used for the expression of bifilar ESS except for the absence of DNA (Reynolds, unpublished results). It would therefore appear that the plateau at long reaction times is due to the expression of all available bifilar ESS as double-strand breaks and not the result of insufficient P D - D N A glycosylase activity.

191

Dose response

Unifilar ESS, indicative of total pyrimidine dimers, were induced as a linear function of UV dose (Fig. 2). Approximately 0.04 unifilar ESS per 106 single-strand molecular weight were induced per J / m 2 from 0 to 50 J / m 2. The linear dose dependence of unifilar ESS is consistent with the linear dose dependence for the induction of pyrimidine dimers at low UV doses. In contrast to the dose response for unifilar ESS, bifilar ESS exhibited a nonlinear dose response exhibiting a slight upward curvature (Fig. 3). Unvortexed samples irradiated with 0 - 5 0 J / m 2 were the same samples used to determine the unifilar ESS dose response (Fig. 2). Similar results were obtained with large molecular weight D N A ( - 350 × 106 weight-average molecular weight) extracted by gentle procedures and D N A subjected to vortexing to reduce the initial molecular weight ( - 120 × 10 ° weight-average molecular weight) prior to enzyme treatment. It would therefore appear that shear degradation does not occur

•

40

20

16

o_

~o ',,9 j08 <~ J ~04

/•

z-

e/

O0 / 0

i

i

i

I0

20

30

UV

DOSE

i

L

40

50

( d / m ~)

Fig. 2. Dose response for the induction of unifilar ESS in UV-irradiated human fibroblasts.

preferentially at positions detected as bifilar ESS and that there is no selective loss of bifilar enzyme-sensitive sites during the D N A extraction procedure. To determine if the induction of bifilar ESS is a simple exponential function of dose, the data in Fig. 3 have been replotted as a log-log plot (Fig.

=E

3O

uJ

4O

1

Q.

o

3o ~E

% 28

u.

m

o/

3z

20

n-

24

lO

o,_ 2 0 03 u3 16

°o

3'0

6'0

o'o

t~o

t~o

LtJ 12

Minutes

Fig. 1. Bifilar ESS reaction kinetics. ]4C-Labeled D N A was purified from human diploid fibroblasts previously irradiated with U V light (600 J/m2; 254 nm). D N A samples were incubated at 37°C for up to 150 rain with either 10 ( e , O ) or 20 #! ( i , o ) of P D - D N A glycosylase. In some reactions (O,ES) a second 10-/~1 addition of P D - D N A glycosylase was made 30 min prior to termination of the reaction. Samples were then assayed for bifilar ESS by velocity sedimentation through neutral sucrose gradients and comparison with minus-enzyme controls.

--J

8

/l /

E ~n

4

,.II Ii ~

0

t

I00

200

UV

i

1

i

i

300

400

500

600

DOSE

(d/m ~)

Fig. 3. Dose response for the induction of bifilar ESS in UV-irradiated human fibroblasts. Individual DNA samples were isolated by the normal, gentle procedure (e) or were vortexed ( O ) to reduce to average D N A molecular weight.

192

4). Theoretical relationships expected for doublestrand breaks arising from randomly induced, closely opposed, single-strand breaks calculated from the equation of Freifelder and Trumbo (1969) have been plotted for comparison. It is apparent from this figure that bifilar ESS induction is not predicted by the theoretical curves. Too many bifilar ESS are observed at low doses and too few at high doses. Overall the slope is between 1.2 and 1.4 instead of a slope of 2 predicted by the theoretical curves. Slopes between 1.2 and 1.4 have been obtained consistently in less extensive experiments with Chinese hamster ovary and AG1522 cells (Lam and Reynolds, unpublished results).

cubation periods. Approximately 80% of the unifilar ESS were removed within 24 h after irradiation (Fig. 5A). Bifilar ESS disappeared during postirradiation incubation with kinetics similar to those observed for unifilar ESS (Fig. 5B). In comparison with the time-dependent loss of unifilar and bifilar ESS in normal excision-proficient cells, little change was obserevd for either unifilar or bifilar ESS during postirradiation incubation of XP1CA cells (Fig. 1A and B). More than 85% of the unifilar ESS and 80% of the bifilar ESS originally induced in XP1CA cells were still present 24 h after irradiation.

Distance between closely opposed dimers Fate of bifilar ESS in human fibroblasts To determine if closely opposed dimers are subject to cellular D N A repair processes, UVirradiated, excision-proficient fibroblasts (AG 1522) were examined for the presence of unifilar and bifilar ESS after various postirradiation in-

g

To obtain a preliminary estimate of the distance between closely opposed unifilar cleavages resulting in the expression of a bifilar ESS as a double-strand break, we have investigated the stability of the nicked c o s site of bacteriophage X under our reaction conditions. In the nicked configuration this site contains 2 single-strand breaks,

/

/g

I00

7

/•

;

,o

~.i ,

=o_

Z

\,

80

O/

'o uJ n,"

,,'/,, /////

-J

60

u.l

2O

~ o.

.S

B

I00

(.9 m

ca;

Z

i

OOl

i IOO

IO

UV

80

Z

.,;;; .4'

DOSE

~ Iooo

( J / m z)

Fig. 4. Log-log plot of the dose response for the induction of bifilar ESS in UV-irradiated human fibroblasts. Data points have been replotted from Fig. 3 together with additional lowdose points not included in Fig. 3. Dashed lines indicate the dose response for double-strand breaks predicted to arise from randomly induced single-strand breaks by the equation of Freifelder and Trumbo (1969). The parameter 'h' in this model represents the maximum number of base pairs unable to prevent strand separation between two closely opposed singlestrand breaks. O, unvortexed D N A ; O , v o r t e x e d D N A .

hi

6o ~

a: ~_ 4o kd

2O 0 0

i

i

i

i

i

i

4

8

12

16

20

24

REPAIR

INCUBATION (HRS)

Fig. 5. Repair of unifilar (A) or bifular (B) ESS in UV-irradiated human fibroblasts (5 j/m2; 254 nm). O, normal (AG1522) fibroblasts; e, UV-sensitive (GM2990) fibroblasts assigned to xeroderma pigmentosum complementation group A. Each point is the mean of 2 independent determinations.

193

one in each of the two complementary strands, separated by 12 base pairs (bp). Digestion of ?~ D N A with EcoR I produces 5 restriction fragments with the cos site situated 3.5 and 21.2 kilobase pairs (kbp) from the termini of the largest (24.7-kbp) fragment. The 12-bp overlap is stable under mild conditions and is sufficient to prevent strand separation during electrophoresis through agarose gels at p H 8.3 (Fig. 6, lane 1). Under mild denaturing conditions (68°C for 15 rain), however, the cohesive ends melt. Separation of the cohesive ends results in the disappearance of the 24.7-kbp fragment and the appearance of 2 smaller fragments of 3.5 and 21.2 kbp (Fig. 6, lane 2). Incubation under P D - D N A glycosylase reaction conditions was found to have no discernible effect on the stability of the nicked cos site. Neither loss of 24.7-kbp fragments nor increase in the

proportion of 3.5- and 21.2-kbp fragments was evident after incubation with enzyme storage buffer for 1 h (Fig. 6, lane 3) or with M . luteus extract for 1 or 2 h at 37°C (Fig. 6, lanes 5 and 7, respectively). Incubation under mildly denaturing conditions after each of these treatments results in a marked reduction in the 24.7-kbp species and a corresponding increase in the 3.5- and 21.2-kbp species (Fig. 6, lanes 4, 6 and 8). Thus the c o s - s i t e - c o h e s i v e ends remain sensitive to mild denaturing conditions during incubation under the standard reaction conditions. To determine if disrupted cos sites may be subject to reannealing during incubation under our reaction conditions, one sample of EcoR I restricted X DNA was incubated at 68°C for 15 min prior to incubation with the M . luteus extract for I h at 37°C. No reannealing at the cos site was evident (Fig. 6, lane 9). We have concluded from these results that hydrogen bonding at 12 bp between closely opposed single-strand breaks is sufficient to prevent strand separation and the appearance of a double-strand break under our reaction conditions. It should be noted that nicks in the X cos site many differ from nicks induced at closely opposed dimers and that these results can therefore give only a rough estimate of the distance between nicked dimers necessary to prevent strand separation and expression as bifilar ESS. Discussion

Fig. 6. Stability of bacteriophage A cos site during incubation with P D - D N A glycosylase. Samples of Eco-R-I-restricted A DNA were end labeled with 32p and then incubated without or with M . luteus P D - D N A glycosylase (ENZ). Half of each sample was heat treated (68°C for 15 min; + A) and both fractions were analyzed by electrophoresis through a 0.4% agarose gel. One sample was also heat treated (68°C for 15 min) prior to incubation with P D - D N A glycosylase.

We have developed an assay for the detection of closely opposed pyrimidine dimers induced in human diploid fibroblasts. In this assay closely opposed dimers are detected as bifilar ESS that give rise to D N A double-strand breaks upon incubation with M . luteus pyrimidine d i m e r - D N A glycosylase. In a previous study we demonstrated that these double-strand breaks did not result from bifilar nicking at isolated D N A photoproducts or from shear-induced breakage of nicked D N A (Lam ane Reynolds, 1986). In this study the high molecular weight D N A used to increase the sensitivity of the assay may be more prone to shear-induced breakage at single-strand nicks introduced during the enzyme treatment (Haward, 1974); however, dose-response curves observed with DNAs of high and low molecular weights

194 were indistinguishable. Thus, within reasonable limits, the number of bifilar enzyme-sensitive sites detected is independent of the initial D N A size distribution and few, if any, of the double-strand breaks detected in enzyme-treated, UV-irradiated high molecular weight D N A are due to shear-induced breakage at nicked dimers. As part of the characterization of our assay for closely opposed dimers we examined the reaction kinetics for the conversion of bifilar ESS to double-strand breaks. Expression of bifilar ESS was found to be much slower than expression of unifilar ESS in the same D N A samples. Under conditions where unifilar ESS were fully expressed as single-strand breaks in less than 5 min, full expression of bifilar ESS as double-strand breaks required up to 2 h (Reynolds, unpublished results). That further bifilar ESS are not expressed during incubation periods longer than 120 min is apparently not due to loss of P D - D N A glycosylase activity. Longer reaction periods, second additions of enzyme at long times and increased amounts of enzyme throughout the reaction period were without effect on the final frequency of bifilar ESS detected. Preincubation of P D - D N A glycosylase for up to 30 min at 37°C was also without effect on the final frequency of bifilar ESS expressed (Reynolds, unpublished results). Hence the final plateau does not appear to be the result of insufficient activity or P D - D N A glycosylase instability. It would therefore appear that under the appropriate conditions a finite population of closely opposed dimers can be detected as bifilar ESS and that a quantitative assay for their detection is possible. We have also investigated the stability of the lambda cos site under our reaction conditions to determine the limits within which 'closely opposed' dimers might be expressed as a bifilar ESS. In its nicked configuration the cos site contains 2 single-strand breaks, one in each of the complementary strands, separated by 12 base pairs. The stability of the annealed cos site under our reaction conditions indicates that 'closely opposed' dimers must be relatively near to each other to be expressed as a bifilar ESS. These results demonstrate that bifilar ESS do not arise by extensive unwinding of the D N A helix between nicks at relatively distant dimers.

One of the more interesting findings in this study is that the dose response for the induction of bifilar ESS does not conform to models that predict a dose-squared relationship for the induction of double-strand breaks from closely opposed single-strand nicks (Schumaker et al., 1956; Thomas, 1956; Hagen, 1967; Freifelder and Trumbo, 1969; Schweitz, 1969). Relative to the numbers of bifilar events predicted to arise from independently induced, random events, too many bifilar ESS were observed at low doses and too few at high doses. Similar results have been obtained in studies with S a c c h a r o m y c e s c e r e v i s i a e (Reynolds, unpublished results) and Chinese hamster ovary cells (Lam, unpublished results). Minton and Friedberg (1974), using bacteriophage T4 P D - D N A glycosylase, have also observed that more bifilar ESS were induced at low UV doses than are expected from random, independent events. They attributed the larger than expected number of sites at low doses to preferential induction of additional dimers at positions closely opposed to previously induced dimers. We have determined in previous studies, however, that dimer induction in one strand is independent of dimer induction in the complementary strand; the same number of bifilar ESS were formed irrespective of whether complementary strands were irradiated together or independently (Lam and Reynolds, 1986). Since the individual dimers that comprise ESS appear to be induced independently of each other, we have concluded that certain D N A sequences must be particularly susceptible to the formation of closely opposed dimers, and that this accounts for the greater than expected number of bifilar ESS at low UV doses. Specifically, we expect bifilar ESS to be induced preferentially at closely opposed pyrimidine runs, where more closely opposed interactions are possible. Furthermore, there may be an increased probability of dimer formation at certain positions within pyrimidine runs (Gordon and Haseltine, 1982; Lam and Reynolds, unpublished results). Thus, the induction of closely opposed dimers should be favored where two pyrimidine runs occur in close opposition. Recent results from experiments in which DNA sequence analysis was used to investigate the nature of unifilar ESS contributing to the formation of a

195 bifilar ESS support this model (Lam and Reynolds, unpublished results). We therefore propose that the dose response observed in these studies for bifilar ESS induction is actually a composite of the individual responses at many different sites. We further propose that site-specific induction probabilities depend upon the local nucleotide sequence as well as upon other factors that are not yet understood. In this study we also observed a time-dependent disappearance of bifilar ESS in excision-proficient normal cells but not in excision-deficient xeroderma pigmentosum cells during postirradiation incubations (Fig. 5). The simplest explanation for these results is that bifilar ESS were being repaired in excision-proficient cells by a process that depends on some function also required for normal excision repair; however, several other processes that do not constitute repair may also provide explanations for the disappearance of bifilar ESS. Closely opposed dimers would not be detected as bifilar ESS if a double-strand break (or two closely opposed single-strand breaks) occurred at or near the position of the closely opposed dimers prior to DNA purification or if one of the damaged strands became paired with an intact, undamaged strand. Double-strand breaks may arise in excision-proficient cells from simultaneous incision a n d / o r incision and gap extension at both dimers of a closely opposed pair. They might also arise from shear-induced degradation occurring preferentially at a gap resulting from excision repair acting at one of two closely opposed dimers. Pairing with an undamaged strand could result either from the successful repair of one of the two closely opposed dimers or through the successful completion of semiconservative DNA synthesis past the closely opposed dimers. Double-strand breaks have been observed in UV-irradiated, exision-proficient human fibroblasts after exposure to relatively high UV doses or after exposure to low doses and incubation in the presence of repair replication inhibitors (Bradley and Taylor, 1981, 1983). In theory doublestrand breaks present at the time of cell lysis should be detected in our assay as a reduction in the double-strand molecular weight of cellular DNA prior to treatment with PD-DNA glycosylase. In this study controls incubated in the

absence of enzyme were run for each PD-DNAglycosylase-treated sample, but the sensitivity of these controls to the presence of pre-existing double-strand breaks is limited by the presence of shear-induced breaks introduced during DNA isolation. Although the frequencies of shear-induced breaks observed in this study were small, they exceeded the frequency of bifilar ESS induced by 5 J / m 2. Therefore formation of double-strand breaks during DNA repair and prior to DNA extraction would not result in sufficient numbers of double-strand breaks to be detected in our assay under the conditions employed. Gapped structures may be generated in vivo as intermediates during excision repair (Haynes, 1966) and may result in the presence of gaps opposite dimers during the operation of excision repair at one of two closely opposed dimers. The appearance of S1 nuclease-sensitive sites in the DNA of UV-irradiated, excision-proficient cells indicates that such structures may be formed in human fibroblasts (Bradley and Taylor, 1981). Gapped structures present at the time of cell lysis may provide preferential sites for shear-induced breakage (Haward, 1974). Any such site-specific breakage at a pair of closely opposed dimers would prevent its detection as a bifilar ESS in our assay. Closely opposed dimers would also not be detectable as bifilar ESS if one of the two damaged strands became paired with an undamaged strand either by successful semiconservative DNA synthesis past a pair of closely opposed dimers or by successful repair of one of the two dimers. Either process acting at closely opposed dimers forming a bifilar ESS would result in the disappearance of that site. In both cases repair of the remaining dimer could be accomplished by normal excision repair and either process could be viewed as part of the normal repair process for closely opposed dimers. Despite the relative insensitivity of our assay to the presence of double-strand breaks, the time-dependent loss of bifilar ESS in excision-proficient ceils is best explained by the repair of closely opposed dimers with restoration of intact DNA strands. Although our assay lacks the sensitivity to determine if bifilar ESS are simply being converted to double-strand breaks, results with neutral elution indicate that few if any double-strand

196

breaks are present in excision-proficient cells irradiated with low doses of UV (Bradley and Taylor, 1983). Less information is available for 'gapped' structures, but results supporting the presence of such structures has been obtained only with cells exposed to high UV doses (Bradley and Taylor, 1981). As indicated above, successful completion of repair for one of the two dimers or successful semiconservative D N A synthesis past a pair of closely opposed dimers would result in the loss of a bifilar ESS but may also be considered as part of the normal repair process for closely opposed dimers. Nevertheless it is unlikely that semiconservative synthesis past closely opposed dimers accounts for the loss of bifilar ESS in this study. Postirradiation incubation in the presence of hydroxyurea was found to have no effect on the rate of bifilar ESS disappearance (Reynolds, unpublished results). This mode of 'repair' would have been inhibited in XP1CA cells since few if any XP cells of complementation group A are expected to recover semiconservative DNA synthetic capacity after irradiation with 5 J / m 2 (Rud6 and Friedberg, 1977). Although our data seem to indicate that closely opposed dimers are subject to cellular repair processes, we still know relatively little about the molecular mechanisms by which closely opposed dimers are repaired, the fidelity of any such repair, and any related biologic effects. Our results together with those of Bradley and Taylor (1981, 1983) indicate that repair at closely opposed dimers requires some of the same gene products necessary for the operation of excision repair at isolated dimers. Excision repair operating at some sites may remove one of the two closely opposed dimers with a gap being introduced opposite the second dimer. (It should be noted that the bifilar ESS assay detects closely opposed dimers that are both upstream and downstream from each other. Due to the polarity of excision repair, gaps opposite dimers would be expected to be possible only for a subclass of the closely opposed dimers detected as bifilar ESS.) In other cases simultaneous incision a n d / o r incision followed by excision may result in the formation of double-strand breaks (Bradley and Taylor, 1981). Completion of repair could be accomplished either by by-pass synthesis with

the possibility of misincorporation or through double-strand break repair, a process that may involve recombination (Resnick and Martin, 1976; Resnick, 1978). These possibilities are currently under investigation.

Acknowledgements We thank Dr. John Cairns for his critical reading of the manuscript and many helpful suggestions. This work was supported by research grant CA-33949, training grant CA-09078, and center grant ES-00002 from the U.S. National Institutes of Health.

References Bonura, T., and K.C. Smith (1975a) Enzymatic production of deoxyribonucleic acid double-strand breaks after ultraviolet irradiation of Escherichia coli K-12, J. Bacteriol., 121, 511-517. Bonura, T., and K.C. Smith (1975b) Quantitative evidence for enzymatically induced DNA double-strand breaks as lethal lesions in UV irradiated pol ÷ and polA1 strains of E. coli K-12, Photochem. Photobiol., 22, 243-248. Bradley, M.O., and V.I. Taylor (1981) DNA double-strand breaks induced in normal human ceils during the repair of ultraviolet light damage, Proc. Natl. Acad. Sci. (U.S.A.), 78, 3619-3623. Bradley, M.O., and V.I. Taylor (1983) Repair-induced DNA double strand breaks after ultraviolet-light and either aphidocolin or 1-fl-o-arabinofuranosylcytosine/hydroxyurea, Carcinogenesis, 4, 1513-1517. Caillet-Fauquet, P., M. Defais and M. Radman (1977) Molecular mechanisms of induced mutagenesis, Replication in vivo of bacteriophage q~X174 single-stranded, ultraviolet lightirradiated D N A in intact and irradiated host cells, J. Mol. Biol., 117, 95-110. Carrier, W.L. and R.B. Setlow (1970) Endonuclease from M i crococcus luteus which has activity toward ultraviolet-irradiated deoxyribonucleic acid: Purification and properties, J. Bacteriol. 102, 178-186. Chadwick, K.H., and H.P. Leenhouts (1983) A quantitative analysis of UV-induced cell killing, Phys. Med. Biol., 28, 1369-1383. Chan, G.L., and W.A. Haseltine (1983) Consequences of UVinduced cyclobutane dimer and pyrimidine (6-4) adduct to DNA replication - - in vitro studies with D N A templates of defined sequence, in: J.J. Broerse, G.W. Barendsen, H.B. Kal and A.J. van der Kogel (Eds.), Proceedings of the Seventh International Congress of Radiation Research, sessions A, Martinus Nijhoff, Amsterdam, A3-13. Clark, J.M., and P.C. Hanawalt (1983) Arrest sites for DNA replication in vivo in UV-irradiated SV40, in: J.J. Broerse,

197 G.W. Barendsen, H.B. Kal and A.J. van der Kogel (Eds.), Proceedings of the Seventh International Congress of Radiation Research, sessions A, Martinus Nijhoff, Amsterdam, A3-114. Cleaver, J.E. (1977) DNA repair processes and their impairment in some human diseases, in: D. Scott, B.A. Bridges and F.H. Sobels (Eds.), Progress in Genetic Toxicology, Elsevier/North-Holland, Amsterdam, pp. 29-42. Cole, R.S., R.R. Sinden, G.H. Yoakum and S. Broyles (1978) On the mechanism for repair of cross-linked DNA in E. coli treated with psoralen and fight, in: P.C. Hanawalt, E.C. Friedberg and E.C. Fox (Eds.), DNA Repair Mechanisms, Academic Press, New York, pp. 287-290. Doubleday, O.P., G. Michel-Maenhaut, A. Brandenburger, P. Lecomte and M. Radman (1981) Inhibition or absence of DNA proofreading exonuclease is not sufficient to allow copying of pyrimidine dimers, in: E. Seeberg and K. Kleppe (Eds.), Chromosome Damage and Repair, Plenum, New York, pp. 447-459. Freifelder, D., and B. Trumbo (1969) Matching of single-strand breaks to form double-strand breaks in DNA, Biopolymers, 7, 681-693. Gordon, L.K., and W.A. Haseltine (1982) Quantitation of cyclobutane pyrimidine dimer formation in double- and single-stranded DNA fragments of defined sequence, Radiat. Res., 89, 99-112. Grosovsky, A.J., and J.B. Little (1983) Mutagenesis and lethality following S-phase irradiation of xeroderma pigmentosum and normal human diploid fibroblasts with ultraviolet light, Carcinogenesis, 4, 1389-1393. Hagen, V. (1967) Bestimmung von Einzel- und Doppelbriichen in bestrahlter Desoxyribonukleins~ture durch die Molekulargewichtsverteilung, Biochim. Biophys. Acta, 134, 45-58. Harm, W. (1968) Dark repair of photorepairable UV lesions in Escherichia coli, Mutation Res., 6, 25-35. Hashem, N., D. Bootsma, W. Keijzer, A. Greene, L. Coriell, G. Thomas and J.E. Cleaver (1980) Clinical characteristics, DNA repair, and complementation groups in xeroderma pigmentosum patients from Egypt, Cancer Res., 40, 13-18. Haward, G.S. (1974) Unique double-stranded fragments of bacteriophage T5 DNA resulting from preferential shearinduced breakage at nicks, Proc. Natl. Acad. Sci. (U.S.A.), 71, 2108-2112. Haynes, R.H. (1966) The interpretation of microbial inactivation and recovery phenomena, Radiat. Res., Suppl. 6, 1-29. Krasin, F., and F. Hutchinson (1977) Repair of DNA doublestrand breaks in Escherichia coli, which requires recA function and the presence of a duplicate genome, J. Mol. Biol. 116, 81-98. Lam, L.H., and R.J. Reynolds (1986) UV-induced, bifilar enzyme-sensitive sites are indicative of closely opposed cyclobutyl pyrimidine dimers, Biophys. J., in press. Lehmann, A.R. (1979) The relationship between pyrimidine dimers and replicating DNA in UV-irradiated human fibroblasts, Nucleic Acids Res., 7, 1901-1912. Little, J.B., J. Nove and R.R. Weichselbaum (1980) Abnormal sensitivity of diploid skin fibroblasts from a family with Gardner's syndrome to the lethal effects of X-irradiation, ultraviolet light and mitomycin-C, Mutation Res., 70, 241-250.

Lloyd, R.S., P.C. Hanawalt and M.L. Dodson (1980) Processive action of T4 endonuclease V on ultraviolet-irradiated DNA, Nucleic Acids Res., 8, 5113-5127. Meneghini, R., C.F.M. Menck and R.I. Schumacher (1981) Mechanisms of tolerance to DNA lesions in mammalian cells, Quart. Rev. Biophys., 14, 381-432. Minton, K., and E.C. Friedberg (1974) Evidence for clustering of pyrimidine dimers on opposite strands of U.V.-irradiated bacteriophage DNA, Int. J. Radiat. Biol. 26, 81-85. Moore, P.D., S.D. Rabkin and B.S. Strauss (1981) In vitro replication of mutagen-damage DNA: sites of termination, in: J.F. Lemontt and W.M. Generoso (Eds.), Molecular and Cellular Mechanisms of Mutagenesis, Plenum, New York, pp. 179-197. Moss, S.H., and D.J.G. Davies (1974) Interrelationship of repair mechanisms in ultraviolet-irradiated Escherichia coli, J. Bacteriol., 120, 15-23. Park, S.D., J.E. Cleaver (1979) Postreplication repair: questions of its definition and possible alterations in xeroderma pigmentosum cell strains, Proc. Natl. Acad. Sci. (U.S.A.), 76, 3927-3931. Resnick, M.A. (1978) The repair of double-strand breaks in DNA: A model involving recombination, J. Theor. Biol., 59, 97-106. Resnick, M.A., and P. Martin (1976) The repair of doublestrand breaks in the nuclear DNA of Saccharomyces cerevisiae and its genetic control, Mol. Gen. Genet., 143, 119-129. Resnick, M.A., and P.D. Moore (1979) Molecular recombination and the repair of DNA double-strand breaks in CHO cells, Nucleic Acids Res., 6, 3145-3160. Reynolds, R.J. (1978) Removal of pyrimidine dimers from Saccharomyces cerevisiae nuclear DNA under nongrowth conditions as detected by a sensitive, enzymatic assay, Mutation Res., 50, 43-56. Rudg, J.M., and E.C. Friedberg (1977) Semi-conservative deoxyribonucleic acid synthesis in unirradiated and ultraviolet-irradiated xeroderma pigmentosum and normal human skin fibroblasts, Mutation Res., 42, 433-442. Rupp, W.D., and P. Howard-Flanders (1968) Discontinuities in the DNA synthesized in an excision-defective strain of Escherichia coli following ultraviolet irradiation, J. Mol. Biol., 31, 291-304. Schumaker, V.N., E.G. Richards and H.K. Schachman (1956) A study of the kinetics of the enzymatic digestion of deoxyribonucleic acid, J. Am. Chem. Soc., 78, 4230-4236. Schweitz, H. (1969) Dggradation du DNA par H202 en prgsence d'ions Cu ÷÷, Fe ÷+, et Fe +++, Biopolymers, 8, 101-119. Sedgwick, S.G. (1976) Misrepair of overlapping daughter strand gaps as a possible mechanism for UV induced mutagenesis in uvr strains of Escherichia coli: a general model for induced mutagenesis by misrepair (SOS repair) of closely spaced DNA lesions, Mutation Res., 41, 185-200. Setlow, R.B. (1968) Photoproducts in DNA irradiated in vivo, Photochem. Photobiol., 7, 643-649. Strauss, B., T. Searashi and M. Robbins (1966) Repair of DNA studied with a nuclease specific for UV induced lesions, Proc. Natl. Acad. Sci. (U.S.A.), 56, 932-939.

198 Thomas Jr., C.A. (1956) The enzymatic degradation of desoxyribose nucleic acid, J. Am. Chem. Soc., 78, 1861-1868. Villani, G., S. Boiteaux and M. Radman (1978) Mechanisms of ultraviolet-induced mutagenesis: Extent and fidelity of in vitro DNA synthesis on irradiated templates, Proc. Natl. Acad. Sci. (U.S.A.), 75, 3037-3041.

Wu, R., and E. Taylor (1971) The nucleotide sequence analysis of DNA, II. Complete nucleotide sequence of the cohesive ends of bacteriophage ~ DNA, J. Mol. Biol., 57, 491-511. Zimm, B.H. (1974) Anomalies in sedimentation, IV. Decrease in sedimentation coefficients of chains at high fields, Biophys. Chem., 1,279-291.