Periodic Changes of Chromatin Organization Associated with ...

Val. 261,No. 13, Issue of May 5, pp. 5758-5765,1986

THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1986 by The American Society of Biological Chemists, Inc.

Printed in U.S.A.

Periodic Changesof Chromatin Organization Associated with Rearrangement of Repair Patches Accompany DNA Excision Repair of Mammalian Cells* (Received for publication, October 4,1985)

Georg Mathis and Felix R. Althausl From the Department of Pharmacology and Biochemistry, University of Zurich (Tierspital), CH-8057Ziirich, Switzerland

We have used 8-methoxypsoralen to probe the chrochromatin structure. It has been speculated that controlled matin structure of mammalian cells in situ while they accessibility to regions of chromatin and specific sequences repair pyrimidine dimers or bulky lesions in DNA. We of DNA may be one of the primary regulatory mechznisms by observed that excision repair of these DNA lesions is which higher eukaryotes establish active chromatin domains accompanied by periodic alterations of chromatin or- (for reviews see Mathis et al., 1980; Reeves, 1984). A similar ganization. In parallel, fluctuationsof the rates of re- mechanism involving altered higher order structure might pairpatch synthesis accompaniedthesestructural operate in DNA repair. changes. Taking advantageof the accessibility offree Our understanding of the distribution of carcinogen-inDNA domains for psoralen intercalation,we have de- duced DNA modifications relative to the nucleosomal array veloped a technique to quantitatively isolate the microof chromatin hasimproved considerably in recentyears (Hancoccal nuclease-sensitive,free DNA fraction of native awalt et aL, 1979; Kaneko and Cerutti, 1980; Beard et al., bulk chromatin. We have determined the location of newly synthesized repair patches relative to free DNA 1981; Lang et al., 1982; Seidman et al., 1983). Also, evidence domains as a function of repair time.Extensive rear- derived from nuclease probing of native chromatin suggests that in the course of DNA excision repair, the micrococcal rangements of repair patches from these domains into micrococcalnuclease-resistant DNA were observed. nuclease sensitivity of newly synthesized repair patches varies Our results indicate that periodic changes of chromatin as a function of repair time. This phenomenon has been organization associatedwith rearrangement of repair interpreted as rearrangement of chromatin structure accompatches accompany the process of excision repair in panying excision repair of DNA (Smerdon and Lieberman, 1980; Bodell and Cleaver, 1981; Zoianet al., 1982a). mammalian cells. We have developed a technique which allows for in situ probing of chromatin organization of intact cells while they undergo DNA excision repair. The procedure takes advantage The accurate maintenance of the genetic record of eukar- of the known property of psoralen compounds to preferenyotic cells is continually challenged by environmental factors, tially intercalateinto free DNA (Song and Tapley, 1979; such as physical and chemical carcinogens. Therefore, the Hearst, 1981). For the present investigation, we have used 8genetic integrity and function of living cells depends on effi- methoxypsoralen, which upon activation with UV light of 365 cient DNA repair mechanisms which coordinately act to nm, forms stable, mostly bifunctional adductsthat cross-link eliminate DNA lesions. Carcinogenic DNA damage normally the DNA strands. That the number of cross-links formed involves a large number of sites in chromatin, and efficient under these conditions is diagnostic for the proportion of free repair of such lesions apparently requires that the components DNA in various states of chromatin activity has been demof eukaryotic DNA repair systems can simultaneously handle onstrated by a number of laboratories (Cech and Pardue, multiple lesion sites. Therefore, by contrast to other active 1977; Cech et al., 1979; Sogo et al., 1984). Based on these processes on chromatin such as transcription,a large portion observations we have developed a novel procedure to study of total cellular DNA becomes involved in repair reactions chromatin domains that become 8-methoxypsoralen-accessible in the course of DNA repair. A major advantage of this (Lindahl, 1982; Friedberg, 1985). The process of DNA repair requires the coordinate inter- technique is the possibility of quantitatively isolating these action of the proteins of the repair machinery with DNA at domains from native bulk chromatin and determining the the lesion sites. Relatively little attention has been directed exact proportion of repair patches associated with these chroto the question of whether or not such DNA-protein interac- matin domains in relation to the progression of the repair tions are associated with alterations of chromatin organiza- process. In addition, this approach preserves a chromatin tion (for review see Hanawalt et aL, 1979). However, a large fraction for analysis which necessarily becomes destroyed body of data suggests that thecommitment of genomic regions when micrococcal nuclease probing of chromatin structure is for active transcription isaccompanied by distinct changes of performed (Wiesehahn et dl., 1977). Complementary to these investigations, we have determined whether or not repair *This workwas supported by Swiss NationalFoundation for patches which abandon 8-methoxypsoralen-accessiblechroScientific Research (Grant 3.375.083), the Sandoz Stiftung, the Stif- matin domains can be recovered in the nuclease-resistant, tung fir wissenschaftliche Forshung der Universitat Zurich, and the psoralen-inaccessible core DNA fraction prepared from native Krebsliga des Kantons Zurich. The costs of publication of this article chromatin, as would be predicted from our considerations. were defrayed in part by the payment of page charges. This article Our results suggest that transient structuralalterations of must thereforebe hereby marked “uduertisement” in accordance with chromatin with distinct periodicity accompany early steps of 18 U.S.C. Section 1734 solely to indicate this fact. DNA excision repair. These waves of structural changes are $ To whom correspondence should be addressed.

5758

Structural Changes of Chromatin Accompany Excision Repair

5759

soaked in 20% trichloroacetic acid. The filters were washed four times with 5% trichloroacetic acid and once with ethanol and the 3H radioactivity was determined using a scintillation counter. DNA was quantified as described by Labarca and Paigen (1980). Alternatively, the nuclease S1-resistant DNA was precipitated with ethanol, lyophilized, and processed for analysis by either polyacrylamide (Laemmh, 1970) or agarose gel (Butler and Thomas, 1980) electrophoresis. EXPERIMENTAL PROCEDURES Random DNA was isolated according to the same protocol, except Chemicals-8-Methoxypsoralen and nuclease S1 (EC 3.1.30.1) were for the nuclease S1 digestion step, which was omitted. purchased from Sigma. Nuclease S1 was diluted to 10 units/pl in Preparation of Oligosomal DNAFractions from Bulk Chromatinglycerokbuffer (l:l, 180 mM NaC1, 60 mM sodium acetate, 3 mM Nuclei were prepared as described above and incubated with microZnClz, pH 4.6) and stored a t -20 "C prior to use. 8-[methoxyL3H] coccal nuclease (EC 3.1.31.1, Worthington, 6,000-15,000 units/mg of Methoxypsoralen and [methyl-3H]thymidine were obtained from protein) for 20 min as described by Butler and Thomas (1980). Amersham. Proteinase K (EC 3.4.21.14) and RNase A (EC 3.1.27.5) Oligosomal DNA fragments were isolated as described above and were from Boehringer Mannheim. Stock solutions were adjusted to a separated on a 1.5% horizontal agarose gel. DNA and radioactivity concentration of 1 mg/ml of Hz0 (proteinase K) or 10 mM Tris, 15 contents of individual fractions were determined as described above. mM NaCl, pH 7.4 (RNase A) and stored at -20 "C prior to use. NNucleoid Sedimentation Analysis of DNA Strand Breaks-The Acetoxy-2-acetylaminofluorene was obtained from the National Can- method described by Cook et al. (1976) with the modifications recer Institute Repository, IIT Institute, Chicago, IL and dissolved in ported by Weniger (1979) wasused to determine DNA strand breaks. MezSO immediately prior to treatmentof cell cultures. Briefly, 3-5 X lo6 cells in 200 p1 of phosphate-buffered saline were Hepatocyte Cultures-Primary cultures of adult rat hepatocytes layered on top of 1 ml of lysis buffer (2 M NaC1, 100 mM EDTA, 2 were prepared from the livers of male SIVZ rats (180-240 g, fed ad mM Tris, 0.5% (v/v) Triton X-100, pH 8.0) and lysed at room libitum) by the collagenase perfusion method of Berry and Friend temperature for 15 min. The lysates were then centrifuged (24,000 (1969) as modified by Bonney et al. (1974). The hepatocytes were rpm, 30 min, Beckman SW 27 rotor, 20 "C) through 15-30% linear then maintained as previously described (Althaus et al., 1982a) on sucrose (in 2 M NaCl, 100 mM EDTA, 2 mM Tris, 16 pg/ml of 100-mm Corning tissue culture dishes at a seeding density of 1.2 X ethidium bromide, pH 8.0) gradients (gradient volume: 14 ml). At the 107/plate. Experiments were started after anadaptation period of 24 end of the centrifuge run, the tubes were kept inthe dark for another h, with a medium change at 4 h of culture in order to remove 60 min. The position of fluorescent bands was determined using a unattached cells. 360-nm light source. Labeling of DNA Repair Patches-DNA repair synthesis was initiated by UV irradiation (Philips TUV 40 W bulb, 254 nm) at 45 J/ RESULTS m2. This UV dose produces approximately 80% of maximal stimulation of DNA repair synthesis in hepatocytes over an 18-h incubation In SituProbing of Chromatin Organization of Intact Cellsperiod (Althaus et al., 1982~). Repair patchessynthesized in response to this treatment were pulse-labeled with [methyl-3H]thymidine (49 Fig. 1demonstrates that thecellular uptake and intercalation in DNA is a very rapid Ci/mmol, 10-20 pCi/3 ml of culture medium) for 20 min at 37 "C. of 8-[meth03cyl-~H]methoxypsoralen Incorporation of radioactivity was stopped by two changes of medium process which is completed within less than 30 s. The quantity containing unlabeled thymidine (1.2 pM), followed by a chase period of intercalated 8-methoxypsoralen which is then available for in the same medium for various time periods. Alternatively, stimu- photochemical reaction with pyrimidines remains constant lation of DNA repair synthesis was achieved by treating hepatocyte monolayers with the ultimate hepatocarcinogen N-acetoxy-2-acetyl- for a prolonged period of incubation (Fig. 1). In separate experiments using alkaline unwinding analysis of DNA (Birnaminofluorene for 20 min in the presence of [rnethyL3H]thymidine. DNA Interstrand Cross-linking-Stock solutions of 8-methoxy- boim and Jevcak, 1981), we determined that thephotochempsoralen were prepared in ethanol and added to the culture medium ical conversion of intercalated 8-methoxypsoralen into interto a final concentration of 46 p ~ After . a 5-min incubation at 37 "C, strand cross-links reached a plateau after3 min of irradiation the formation of DNA interstrand cross-links from intercalated 8- with UV 365 nm (data notshown). Thus, 8-methoxypsoralen methoxypsoralen was effected by irradiating the monolayers for 5 min with UV light of 365 nm (Sylvania black light bulbs type can be stably integrated into the DNA of intact cells by a F20T12BLB, 7.5 cm irradiation distance) a t room temperature. At simple photochemical reaction which takes less than 4 min the end of this treatment, hepatocyte monolayers were washed with and yields highly reproducible results. ice-cold phosphate-buffered saline (136.9 mM NaCl, 2.7 mM KCl, 4.3 Electron microscopic analysis (Cech and Pardue, 1977; mM Na2HPO4,1.5 mM KH2PO4,pH 7.4), and thecells were harvested Cech et al., 1979) as well as nuclease probing (Wiesehahn et using a rubber policeman. al., 1977) of the location of psoralen-induced interstrand Nuclease 5'1-resistant Isolation of 8-Methoxypsoralen-Cross-linlzed, DNA-Nuclei were isolated from hepatocytes as described earlier cross-links have firmly established that psoralen compounds (Althaus et al., 1982a). 2 X lo7 nuclei were suspended in 900 pl of exhibit a distinct preference for intercalation in micrococcal buffer (50 mM Tris, 9 mM NaCl, 10 mM EDTA, pH 8.0) in a shaking nuclease-sensitive chromatin domains. Sogo et al. (1984) have water bath and incubated for 30 min a t 37 "C with 10 pg of RNase A. convincingly demonstrated that the extent of psoralen-crossProteinase K (20 pg)and sodium dodecyl sulfate (final concentration linking is diagnostic for the structure of active nucleolar 0.5%) were added to the reaction and the incubation was continued for another 45 min. At this point, the sodium dodecyl sulfate concentration was raised to 1%, and the incubation was continued a t 25 "C for 10 min. The reactions were stopped by addition of sodium perchlorate (1.3 M final concentration) and further incubation a t 25 "C. Following three to four extractions with ch1oroform:isoamyl alcohol (241), theDNA was precipitated with 3 volumes of 100% ethanol a t -20 "C overnight. Precipitated DNA was pelleted, lyophilized, and dissolved in10 mM Tris, 1 mM EDTA, pH 8.0. Following heat INCUBATION TIME (MIN) denaturation for 2 min in a boiling water bath, limited renaturation FIG. 1. Kinetics of intercalation of 8-methoxypsoralen (8was allowed by immersion of the test tubes in liquid nitrogen. NonCultured hepatocytes (approximately renatured, single-stranded DNA domains were then enzymatically MOP) in intact hepatocytes. removed by a 30-min treatment with nuclease S1 (10-15 units/pg of 2 pg of DNA equivalents) were incubated in the presence of 2.1 p~ (74 pCi/mmol). At the times indiDNA) in a buffer containing 180 mM NaCl, 60 mM sodium acetate, 3 8-[metho~yl-~H]rnethoxypsoralen mM ZnClz, pH 4.6, a t 37 "C. The reaction was stopped by adding 1 cated, the incubation was stopped by addition of ice-cold trichloroavolume of 50 mM EDTA. The denaturation-renaturation-nuclease S1 cetic acid (20%). The trichloroacetic acid precipitates were then treatment was repeated for the times indicated (cf. Table 111). For transferred onto Whatman GF/C filters and processed for determianalysis of the nuclease S1-resistant radioactivity contained in this nation of radioactivity as described under "Experimental Procepreparation, aliquots were transferred onto Whatman GF/C filters dures.''

accompanied by extensive redistribution of repair patches relative to the nucleosomal organization of chromatin. Fluctuations in the rates of repair patch synthesis parallel these structural alterations.

Structural Changes of Chromatin Accompany Excision

5760

chromatin of Dictyostelium discoideum cells as well as of rat liver bulk chromatin. With the recent availability of radioactively labeled 8-methoxypsoralen of high specific activity, it became possible to directly test the psoralen accessibility of bulk chromatin in intact cells (Fig. 2). The saturation curve of the reaction of 8-methoxypsoralen with hepatocellular DNA exhibited two linearity ranges with distinctly different slopes. The micrococcal nuclease sensitivity of DNA-bound 8-[methoxyl-3H]methoxypsoralen following incubation of cells in the presence of a total concentration of 33 or 100 p M 8-methoxypsoralen was almost identical (TableI). Thus, even at psoralen concentrations selected from the upper linearity range of saturation shown in Fig. 2, the reaction was preferentially with micrococcal nuclease-sensitive DNA. The lower linearity range of the curve in Fig. 2 could be a consequence of preferential psoralen photoaddition to sequences rich in specific nucleotides, as hasbeen reported by others (for review see Song and Tapley, 1979), while the upper range could reflect sites less favorable for photoaddition. Nevertheless, in order to ensure maximal selectivity of probing in subsequent experiments, we selected a maximally saturating concentration derived from the lower linearity range of the saturation I

Repair

curve (i.e. 46 p M ) comprising the most readily accessible reaction sites of the micrococcal nuclease sensitive fraction of bulk chromatin. I n Situ Probing of Chromatin of Intact Cells Undergoing DNA Excision Repair-Irradiation of cells with UV light of 254 nm wavelength introduces a specific type of chemical modification in DNA, i.e. pyrimidine dimers, which are randomly distributed in chromatin (Zolan et al., 1982b). These lesions are efficiently removed by DNA excision repair. Using 8-methoxypsoralen as a probe, we analyzed the chromatin organization of cells which had been induced to repair pyrimidine dimers (Fig. 3). Within 20 min following the formation of pyrimidine dimers, we observed a sharp increase in the nuclear DNA fraction that became accessible for intercalation with 8-methoxypsoralen. As cells wereallowed to proceed with DNA excision repair, the extent of psoralen interaction with DNA became similar to the levels of untreated control cells. A second, also transient, increase in psoralen accessibility of chromatin was observed during the following repair period. If cells were kept at 0 “C following UV treatment in order to suppress repair activity, no significant change in psoralen binding was observed over the entire time interval (Fig. 3), suggesting that the presence of pyrimidine dimers per se does not influence this process. We therefore conclude that the fluctuations of psoralen accessibility of chromatin are a direct consequence of DNA excision repair processes. Thus, it appears that UV excision repair is accompanied by at least two wavesof transient alterations of chromatin structure. In separate experiments we determined the relationship

I 0

50

MEDIUM

100

CONCENTRATION

150

(pM)

FIG. 2. Dosedependence of the8-methoxypsoralen (8MOP) reaction with hepatocellular DNA.Cultured hepatocytes were incubated for 5 min at 37 “Cin the presence of various concentrations of 8-methoxypsoralen containing 8-[rneth0xyl-~H]methoxypsoralen (range of specific activities: 0.03-9 mCi/mmol) and then irradiated with UV 365 nm light as described under “Experimental Procedures.” The cells were then harvested and DNA was extracted following nuclei isolation. The radioactivity bound to nuclear DNA was then quantified as described under “Experimental Procedures.” 0 (IO 120 The plotted values are corrected for the range of specific activities used in the incubation mixture. rl and r2, correlation coefficients REPAIR TIME (mln) derived from linear regression analysis; s1 and s2, slope factors for FIG. 3. In situ probing of 8-methoxypsoralen (8-MOP) accurves 1 and 2, respectively. cessibility of hepatocellular chromatin. Cultured hepatocytes were exposed to UV 254 nm (45 J/m2) andallowed to repair lesions TABLE I a t 37 “C. Under these conditions, this treatment produced approximately 80% of maximal stimulation of repair synthesis in cultured Micrococcal nucleasesensitivity of DNA-bound 8-[metho~yl-~H] hepatocytes (Althaus et al., 1982~).At the times indicated in the methoxypsoralen (8-MOP) (46 Experimental conditions: Hepatocytes were exposed to 33 or 100 abscissa, cells were treated with 8-[rnetho~yl-~H]methoxypsoralen p ~ specific , activity adjusted to 50-150 mCi/mmol) followedby p~ 8-methoxypsoralen as described in the legend to Fig. 2. Micrococcal nuclease digestion was performed according to Berkowitz and photoactivation at 365 nm as described under “Experimental ProceRiggs (1981), using 8 A260 units of chromatin/incubation. Averaged dures.” Following this DNA cross-linking treatment, nuclear DNA was extracted and processed for the quantification of radioactivity as values from two analvses. described under “Experimental Procedures.” The specific activity of % DNA-bound %MOP nuclear DNA is expressed relative to control values. Control values Micrococcal nuclease hydrolyzed were obtained under identical treatment conditions except for the incubation time UV 254 nm exposure which was omitted. The values represent the 33 pM 100 p M means f SE of four independent experimentsinvolving four separate min % cell preparations (except for the time points indexed a and b, which 0 0 0 are derived from two separate cell preparations (averaged values) and 15 63 f 2 72 + 1 one cell preparation, respectively. Significant differences from 0-min 30 82 f 7 78+ 1 values were obtained for the 20- and 80-min ( p < 0.01) and for the 45 82 +: 1 83 f 1 130-min ( p < 0.05) values. ~~

~

Structural Chunges of Chromatin Accompany Excision Repair

5761

We next determined whether the repair of a different type of DNA lesion, namely bulky adducts formed with the ultimate carcinogen N-acetoxy-2-acetylaminofluorene, would be associated with similar rearrangements in chromatin organization of intact hepatocytes (Fig. 4). It should be noted that this chemical carcinogen interacts in a distinctly different, nonrandom manner with chromatin, i.e. the vast majority of adducts with N-acetoxy-2-acetylaminofluorene (mainly C-8 adducts of guanine) are preferentially located in linker regions of chromatin (Langet aL, 1982).The data inFig. 4 show that the interaction of 8-methoxypsoralen with chromatin changes dramatically, as hepatocytes proceed to remove these nuclear DNA adducts. When cells were kept on ice for identical time periods following the adduct formation in order to prevent repair activity, no change in psoralen binding was observed, suggesting that thepresence of the lesions per se did not affect this reaction. Thus, repair of these bulky lesions was accompanied by transient changes of chromatin organization, with amplitudes inverse to those obtained during pyrimidine dimer repair, but otherwise with strikingly similar time characterTABLE I1 istics. The amplitude of these alterations apparently is related t o the number of lesions which are initially present (Fig. 4). 8-Methoxypsoralen (8-MOP) binding and micrococcal nuclease (MNase) sensitivity of hepatocellular chromatin following 20-min UV For reference, a concentration of 5 X M N-acetoxy-2excision repair acetylaminofluorene in the culture medium led to theformaExperimental conditions T~~~~~~~ tion of 250 & 50 fmol (mean f SD, n = 4) of dG-8 adducts/ over Probing technique pg of DNA. Cultured hepatocytes are capable of completely control UV repair, control 20 min“ eliminating these adducts through excision repair during subsequent incubation in the absence of the carcinogen’ as de1.01 f 0.04 38.3 8-MOP binding 0.73 f 0.03 termined by radioimmunoassay (Poirier, 1981). At this level (pmollpg DNA)* 54 f 1 28 MNase sensitivity 42 f 1 of modification and during the first 10 min after onset of (% DNA diexcision repair, the 8-methoxypsoralen accessibility of chrogested)’ matin decreases by 20% compared to untreated control cells a UV 254 nm; 45 J/mZ. or cells that had not been allowed to repair DNA damage. Experimental procedures were as described under “Experimental This drop in psoralen accessibility became even more signifProcedures;” the values represent mean f SE derived from four icant at higher concentrations of N-acetoxy-2-acetylaminoseparate experiments. MNase digestions were performed as described in the legend to fluorene with maxima after 20 min repair time. A second, also Table I; the datawere derived from 45-min incubations and represent transient, decrease of psoralen accessibility was observed at the averages from two analyses. 80 min after onset of repair (data not plotted). We conclude from these observations that therepair of two different types of DNA lesions, namely bulky lesions primarily associated with linker DNA, and randomly distributed pyrimidine dimers, is accompanied by distinct waves of transient 150 structural changes of chromatin. The first wave of fluctuations overlaps with a time period where a maximum steady state concentration of DNA strand breaks is attained (Fig. 5). Also, we have determined that the rates of repair patch synthesis at different time intervals after UV 254 nm treat100 ment exhibit similar fluctuations (Fig. 6) which roughly parallel those observed at thestructural level. This thensuggests that thealterations of chromatin organization as detected by psoralen probing might interrelate with the kinetics of indi50 vidual steps of DNA excision repair. Analysis of the Distribution of Repair Patches Relative to 0 15 30 80 Chromatin Organization in Intact Cells-The following series of experiments was designed to determine whether repair REPAIR TIME (rnin) patches become located in 8-methoxypsoralen-accessible FIG. 4. Fluctuations of 8-methoxypsoralen(8-MOP) accessibility of hepatocellular chromatinas a function of carcino- chromatin domains while hepatocytes recover from UV 254 gen concentrations.Cultured hepatocytes were exposed to various nm DNA damage. This would allow us to analyze the mode doses of UV 254 nm or N-acetoxy-2-acetylaminofluorene (37 “C, 10 of appearance and disappearance of repair patches ina defined min) followedby treatment with 8-[metho~yl-~Hjmethoxypsoralenfraction of chromatinas these cells progressively move and UV 365 nm as described in the legend to Fig. 3. The amount of through subsequent steps of excision repair. Repair patches radioactive psoralen associated with nuclear DNA after different which disappear from 8-methoxypsoralen-accessiblechromarepair intervalsis expressed as a percentage of control cultureswhich had not been treated with the carcinogens. Treatment conditions: tin domains should then become detectable in areas which o ” 0 , UV 254 nm 15 J/m2;a”-o, 30 J/m2; A-A, 45 J/m2. N- are not accessible for this intercalating chemical. Acetoxy-2-acetylaminofluorene(NAcAAF): “ .-,. 5 X M; F. R. Althaus and M. C. Poirier, unpublished observations. U , 5 x 10-5 M; A-A, 5 x 10-4M.

between psoralen binding and micrococcal nuclease sensitivity of bulk chromatin derived from cells undergoing UV excision repair (Table 11).The results show that theobserved increase in psoralen binding following 20-min UV excision repair (Fig. 3) is paralleled by an increase in the micrococcal nuclease sensitivity of chromatin. A 38% rise in psoralen binding corresponded to a 28% higher micrococcal nuclease sensitivity as compared to control cells that were not induced to repair UV damage. Similar fluctuations of psoralen accessibility were also observed at lower fluences of UV 254 nm (i.e. 15 and 30 J/m2, Fig. 4). However, after 20 min of repair, the time course of changes in psoralen binding became distinctly different a t lower fluences. Following a UV 254 nm dose of 15 J/m2, the establishment of the original psoralenbinding status was significantly delayed, a dose of 30 J/m2 yielded intermediate results. It is possible that further differences wouldbecome apparent at shorter time intervals of psoralen probing; however, such experiments are limited by the availability of cell material and expensive radioisotope.

E

Structural Changes of Chromatin Accompany Excision Repair

5762 1.0

0

2 0.: U

C

1

0

1

1

It I

1 ” 12 REPAIR TIME (HOURS)

0.5

I

24

FIG. 5. Effect of UV 254 nm (45 J/m2) or N-acetoxy-2acetylaminofluorene (5 x lom6M) treatment on the relative level of DNA strand breaks in hepatocytes as a function of repair time. DNA strand breaks were analyzed by nucleoid sedimentation analysis as described under “Experimental Procedures.” The values are expressed as ratios of sedimentation of treated to untreated control cells. The datarepresent the averages derived from two determinations with separate cell preparations. o ” 0 , N-acetoxy-2-acetylaminofluorene;M, UV 254 nm.

150

l”---l

(Fig. 7b). The most critical step in this procedure involved carefully controlled renaturation of heat-denatured DNA in liquid nitrogen in order to avoid extensive renaturation beyond the cross-linked domains. Under these conditions, a constant proportion of approximately 20% of total nuclear DNA became resistant to S1 nuclease digestion (Fig. 8). This DNA fraction, operationally defined as 8-methoxypsoralenaccessible, nuclease S1-resistant DNA (MOPS-DNAz),was analyzed on 0.9% agarose gels (Table 111).The results showed that a single round of denaturation-renaturation-nuclease S1 treatment produced MOPS-DNA with a size distribution between 200 and 6000 nucleotides. A second treatment cycle reduced the total fraction as well as thesize heterogeneity of MOPS-DNA. A third treatment decreased the size heterogeneity but did not furtherreduce the DNA recovered as MOPS-

DNA. Repair patches formed in response to irradiation of hepatocytes with UV light of 254 nm were labeled using a pulsechase protocol (Fig. 9). At different times afterUV irradiation, MOPS-DNA was extracted from these cells and analyzed for its content of radioactively labeled repair patches. In order to correct for interexperimental variations in the overall incorporation of radioactivity during repair, changes in the radioactivity content of MOPS-DNA measured in different repair intervals were expressed relative to changes in random nuclear DNA using the formula F(t0 + t,) =

Z ( t 0 + tl)

I ’ ( t 0 + tl)

where F expresses the quotient of the relative change of specific radioactivity in MOPS-DNA (I)over random DNA (1’)in the repair interval to+ tl. Following a 20-min pulse-labeling of repair patches synthesized in response to UV 254 nm treatment of cultured hepatocytes, a minor proportion of repair-incorporated radioactivity was found associated with MOPS-DNA (F value of 0.5). However, during a subsequent chase period of 100 min, a large o 30 ro 150 180 1020 proportion of radioactively labeled patches became transloREPAIRINTERVAL ( m i d cated into the MOPS-DNA fraction ( F value = 1.56). This FIG. 6. Rates of DNA repair synthesis of intact hepatocytes shift of repair patches relative to chromatin organization was at different time intervalsfollowing irradiation withUV 254 transient, since during an additional chase period of 60 min, nm. Cultured hepatocytes were irradiated with UV 254 nm (45 J/m2) a relocation of patches from MOPS-DNA into psoralen-inand allowed to repair at 37 “ C for the times indicated on the abscissa. At different intervals during this repair period, [n~ethyl-~HJthymidineaccessible chromatin domains was observed (Fig. 9). Theoretically, a transient increase of repairpatchesin (PHITdR,3 pCi/ml of medium, 42 Ci/mmol) was added to theculture medium prior to theharvest of cells. Nuclei were prepared from these MOPS-DNA should inversely lead to a deficit of repair cells and theamount of radioactivity incorporated into nuclear DNA patches innuclease-resistant chromatin domains. The results was determined as described under “Experimental Procedures.” No of Fig. 10 are compatible with this prediction. A large proporreplicative DNA synthesis was measured in thesenonreplicating cells tion of repair radioactivity, ie. 53% (average from two sepa(Althaus et al., 1982a). Fluctuations in endogenous thymidine pools are unlikely to contribute to altered rates of [n~ethyl-~HIthymidine rate experiments) is recovered in micrococcal nuclease-resistincorporation in the course of repair (Althaus et al., 1982~).Repair ant core DNA following a pulse-labeling of 20 min, and during intervals are: A , 0-10 min; B, 10-20 min; c, 20-30 min; D,30-70 min; a subsequent chase period of 100 min, a significant proportion E , 70-150 min; F, 150-180 min; G, 180-1020 min. The plotted values of this radioactivity ( i e . 28%) is lost from the monosomal are derived from two independent experiments with separate cell DNA fraction derived from bulk chromatin. Thus, we conpreparations which yielded very similar results (deviations from avclude that repair patches shift from micrococcal nucleaseeraged values: S % ) . resistant core DNA stretches intoMOPS-DNA domains (and Fig. 7a summarizes the principal features of our approach vice uersa) as pyrimidine dimers are progressively repaired in to the isolation of 8-methoxypsoralen-accessible DNA do- these cells. DISCUSSION mains from bulk chromatin. The procedure is based on the observation by Ben-Hur et al. (1979) who demonstrated that The datapresented above suggestthat theprocess of DNA psoralen-cross-linked DNA sequences exhibit different rena- excision repair is accompanied by extensive transient changes turation kinetics compared to un-cross-linked DNA. By virtue of chromatin architecture associated with rearrangement of of their faster renaturation following denaturation under al- newly synthesized repair patches relative to nucleosomal arkaline or high temperature conditions, cross-linked DNA ray. In the time frame examined, at least two distinct waves fragments become differentially resistant to S1 nuclease attack. We have modified this procedure to purify 8-methoxThe abbreviation used is: MOPS-DNA, 8-methoxypsoralen crossypsoralen-cross-linked DNA domains on a preparative scale linked, nuclease SI-resistant DNA.

5763

Structural Changesof Chromatin Accompany Excision Repair l s o l a t l o nP r o t o c o l

w + Nuclei

l!!~llt)(iii!!,ll

UV365

I’’ L

1

COVALENT BONDS

RNAse A Protelnase K

%

c

ChlorofomVEthanol

I

QM

HEATDENATURATION

c

Heat Denaturat Ion

1

Liml ted Renaturat lon

(Liuuld Nltrogen)

LIMITED RENATURATION

&Nuclease

ANUCLEASE DIGESTION

-L

S1

3

A

S1

Liiuld electrophoresis sclntillatlon SpectroDhotometrY countlng

FIG. 7 . Scheme summarizing the critical stepsfor the isolation of free DNA from bulk chromatin. A, principIes of the experimental approach; B, isolation protocol for the 8-methoxypsoralen-cross-linked, S1 nucleaseresistant DNA (MOPS-DNA)fraction.

TABLEI11 Characterization of the proportionand size heterogeneity of MOPS-DNA following repeated denaturation-renaturation-nuclease SI treatment cycles Treatment MP-DNA Size distributionb cycles

0 200-6000 1 0

3 0

2 3

CONTROL

30 60 DIGESTION TIME (MIN)

FIG.8. Nuclease S1 digestion kinetics of DNA extracted from hepatocellular chromatin following in situ cross-linking with 8-methoxypsoralen. Hepatocytes were subjected to theDNA cross-linking treatment asoutlined in Fig. 7A and DNA wasextracted and processed for nuclease S1 treatment as summarized in Fig. 7 B (for details see “Experimental Procedures”). Control preparations were handled identically except for the cross-linking treatment which was omitted.

recovered” %

no. of nucleotides

100 23.9 9.3 9.0

>>10,000 550-4000 550-2000

Determined as described under “Experimental Procedures.” Analyzed on 0.9% agarose gels as described under “Experimental Procedures.” The values slightly overestimate the sizes of fragments because the presence of covalently bound 8-methoxypsoralen reduces the electrophoretic mobility of DNA (cf.Sogo et al., 1984). a

these results suggest that lesion-type specific changes occur at the level of chromatin structure concomitant with DNA excision repair. It should be noted that these structural alterations were not induced by the presence of the lesions per se. of structural changes with similar amplitudes and peaks at 20 Combined evidence from electron microscopic and bioand 80 min after UV, respectively, were observed (Fig. 3). chemical analyses has convincingly demonstrated that the These fluctuations could be demonstrated for the repair of number of DNA interstrand cross-links formed following UV-induced pyrimidine dimers and, in an inverse sense, also photoactivation of intercalated psoralen is diagnostic for the for the repair of bulky substituents introduced by N-acetoxy- presence of non-nucleosomally organized stretches of DNA in 2-acetylaminofluorene. A striking difference inthese two highly active nucleolar DNA (Sogoet al., 1984). By inference, types of lesions is their patternof distribution relative to the our data could be interpreted to indicate variations in the nucleosomal organization of chromatin, i.e. random distribu- proportion of free DNA as DNA excision repair proceeds. tion for pyrimidine dimers uers‘susthe preferential localization Alternatively, Sinden et at. (1980) have shown that theextent inlinker DNA of the adducts formed with N-acetoxy-2- of psoralen-binding increases in supercoiled DNA. This could, acetylaminofluorene. While we do not know whether this however, hardly account for the observed transient increases initial distribution per se induces a different mode of repair in psoralen-binding in pyrimidine dimer excision, since enand consequently different patterns of structural fluctuations, donucleolytic incision next to the sites of pyrimidine dimers

5764

Structural Changes of Chromati:nAccompany Excision Repair DNA REPAIR DAMAGE HOURS

I

PERIOD CL

24 bNNNNj120’

0’

.c

b 4 1 1 2 0 ’

CL

.c 1180’

PULSE

A

CHASE

F-VALUE INTERVALTIME

1

I 0 - 2 0 min

20-120min 120-180 min

I 1.0

8

2.o

FIG. 9. Rearrangement of repair patches relative to 8methoxypsoralen-accessible,S1 nuclease-resistant DNA. A, cultured hepatocytes were treated with UV 254 nm (45 J/m2) in order to introduce pyrimidine dimers into DNA. Cells wereallowed to repair for the times indicated, and newly synthesized repair patches were labeled according to the pulse-chase protocol shown. CL, crosslinking treatment (for details, see “Experimental Procedures”). B, rearrangement of repairpatches relative to MOPS-DNA. The F values were calculated using the formula given in the text. Each bar 1 SE of five independentexperiments with represents the mean separate cell preparations (p < 0.01).

tri di

mono

FIG. 10. Association of newly synthesized repair patches with mononucleosomal DNA fragments. Cultured hepatocytes were irradiated with UV 254 nm and allowed to repair DNA damage for 20 min in the presence of [met/~yl-~H]thyrnidine. Alternatively, cells were allowed to repair for an additional time period of 100 min in the absence of radioactive label. Hepatocytes were harvested and nuclei were prepared. Oligosomal DNA fragments were prepared and separated on 1.5% agarose gels as described by Butler and Thomas (1980). The percentage of repair-incorporated radioactivity in different oligosomal fractions relative to the radioactivity contained in bulk DNA was determined using a scintillation counter. Agarose gel, lane A: 20 min repair; lane B: 120 min total repair time. Mono, di, and tri designate the positions of mononucleosomal, disomal, and trisomal DNA fragments. The marker positions are (from top to bottom): 1358, 1079,872, 603, 310, 281, 234,and 194 base pairs.

would supposedly induce a relaxed state of DNA and consequently reduce psoralen binding. The findings reported by Sinden et a1. corroborate this view. They found that the rate of psoralen-DNA photobinding is unchanged in cellular DNA which contains a frequency ofDNA strand breaks 10-200fold greater than thatsufficient to relax DNA in vitro.A third possibility is that prior to incision, supercoiling increases in DNA domains as they become committed to repair. This increase could result from transient partial dissociation of

these DNA domains from chromatin proteins which would synergistically enhance psoralen binding. Such local disruptions of nucleosomal packing could serve to facilitate access of excision repair proteins to the sites of lesions. An intriguing feature of the structural alterationsobserved in the present study is the fact that they follow a periodic pattern. A conceivable interpretation of this finding is that components of the repair machinery progressively interact in a step by step manner with subsets of identical DNA lesions which, however, differ in their localization relative to higher order chromatin structure. This mode of repair would involve structural changes of chromatin whereby topologicalrestrictions in the accessibilityof subsets of lesions become gradually eliminated. Thus, each round of repair involving a new subset of lesions could be initiated by rearrangements of chromatin structure. In thepresent report we have described two distinct waves of such gross rearrangements. Considering these fluctuations at the structural level as part of a processive mode of repair, one might expect that the kinetics of individual repair reactions in intact cells might be sensitive to these rearrangements. Our observation regarding fluctuations in the rates of patch synthesis (Fig. 6) is compatible with this view. To our knowledge, this report presents the first attemptto quantitatively isolate micrococcal nuclease-sensitiveDNA domains from native bulk chromatin. These areas necessarily become destroyed when nuclease probing techniques are applied. The use of psoralen cross-linking as a startingpoint for the isolation of free DNA from intact cells undergoing DNA repair offers the principal advantage that newly synthesized repair patches become stably associated with a distinct chromatin domain. Thus, in the present probing procedure, the original localization of repair patches in native chromatin is preserved in situ and becomes resistant to subsequent fractionation procedures. This eliminates the uncertainties associated with nuclease probing techniques which might per se affect the original localization of repair patches (for review see Reeves, 1984). Using our technique in combination with micrococcal nuclease probing, we were able to demonstrate that a relative loss of repair patches in micrococcal nucleaseresistant domains was accompanied by a compensatory increase in psoralen-accessible areas (and vice versa). Our results suggest that extensive rearrangements of repair patches relative of the nucleosomal array of chromatin accompany the process of DNA excision repair. Thus, these results confirm and extend the observations of the groups of Hanawalt, Cleaver, and Lieberman, who proposednucleosomal rearrangement of repair patches on the basis of their differential micrococcalnucleasedigestion kinetics relative to random DNA (Smerdon and Lieberman, 1980; Bodell and Cleaver, 1981; Zolan et al., 1982a). Recently, Cohn and Lieberman (1984) useda novel immunoaffinity probing protocol to study the repair time-dependent distribution of repair patches in total genomicDNA. These investigators suggested that theexcision repair system itself might operate in an inherently nonrandom manner. In accordance with this concept, our results suggest that extensive changes of chromatin organization are an integral part of this presumed repair mode. The possibility of a functional link is further substantiated by our recent observation that all of these processes depend similarly on nuclear poly(ADPribose) formation induced byDNAdamage.3 This process affects other chromatin functions in cultured hepatocytes

G.Mathis and F. R.Althaus, unpublished results.

Structural Changes of Chromatin Accompany Excision

Repair

5765

which arealso associated with changes of chromatin structure Cook, P. R., Brazell, I. A., and Jost, E. (1976) J. Cell Sei. 2 2 , 303324 (Althaus et aL,1982b). Friedberg, E. C. (1985) DNA Repair, W. H. Freeman and Company, Acknowledgments-We thank Dr. Miriam C. Poirier,National Institutes of Health, Bethesda, MD, for the analysis of DNA adducts and Drs. J. M. Sogo and Th. Koller, ETH Zurich, Switzerland, for sharing some of their results prior to publication. We are grateful to Ralph Eichenberger for excellent technical assistance. REFERENCES Althaus, F. R., Lawrence, S. D., Sattler, G. L., and Pitot,H. C. (1982a) J. Biol. Chem. 257,5528-5535 Althaus, F. R., Lawrence, S. D., He, Y.Z., Sattler, G.L., Tsukada, Y., and Pitot, H. C. (1982b) Nature 300,366-368 Althaus, F. R., Lawrence, S. D., Sattler, G. L., Longfellow, D. G., and Pitot, H. C. (1982~)Cancer Res. 42,3010-3015 Beard, P., Kaneko, M., and Cerutti, P. A. (1981) Nature 291,84-85 Ben-Hur, E., Prager, A., and Riklis, E. (1979) Photochem. Photobiol. 29,921-924 Berkowitz, E. M., and Riggs, E. A. (1981) Biochemistry 2 0 , 72847290 Berry, M. N., and Friend, D. S. (1969) J. Cell Bid. 43,506-520 Birnboim, H. C., and Jevcak, J. J. (1981) Cancer Res. 4 1 , 1889-1892 Bodell, W. J., and Cleaver, J. E. (1981) Nucleic Acids Res. 9,203-213 Bonney, V. R., Becker, J. E., Walker, P. R., and Potter,V. R. (1974) In Vitro 9,399-413 Butler, P. J. G., and Thomas, J. 0.(1980) J. Mol. Biol. 140,505-529 Cech, T., and Pardue, M. L. (1977) Cell 11,631-640 Cech, T., Pathak, M. A., and Biswas, R. K. (1979) Biochim. Biophys. Acta 562,342-360 Cohn, S. M., and Lieberman, M. W. (1984) J. Biol. Chem. 259, 12463-12469

New York Hanawalt, P. C., Cooper, P. C., Ganesan, A. K., and Smith, C.A. (1979) Annu. Reu. Biochem. 4 8 , 783-836 Hearst, J. E. (1981) J. Inuest. Dermatol. 77,39-44 Kaneko, M., and Cerutti, P. A. (1980) Cancer Res. 40,4313-4318 Labarca, C., and Paigen, K. (1980) Anal. Biochem. 1 0 2 , 344-352 Laemmli, U. K.(1970) Nature 2 2 7 , 680-685 Lang, M.C., de Murcie, G., Mazen, A., Fuchs, R. P. P., Leng, M., and Daune, M. (1982) Chem.-Biol. Interact. 4 1 , 83-93 Lindahl, T. (1982) Annu. Rev. Biochem. 51, 61-87 Mathis, D., Oudet, P., and Chambon, P. (1980) Prog. Nucleic Acid. Res. 2 4 , 1-55 Poirier, M. C. (1981) in DNA Repair (Friedberg, E. C., and Hanawalt, P. C., eds) pp. 143-153, Marcel Dekker Inc., New York Reeves, R. (1984) Biochim, Biophys. Actu 782,343-393 Seidman, M., Slor, H., and Bustin, M. (1983) J. Biol. Chem. 2 5 8 , 5215-5220 Sinden, R.R., Carlson, J. O., and Pettijohn, D. E. (1980) Cell 21, 773-783 Smerdon, M. J., and Lieberman, M. W. (1980) Biochemistry 19, 2992-3000 Sogo, J. M., Ness, P. J., Widmer, R. M., Parish, R. W., and Koller, T. (1984) J. Mol. Bid. 1 7 8 , 897-919 Song, P. S., and Tapley, K. J. (1979) Photochem. Photobiol. 29,11771197 Weniger, P. (1979) Int. J. Radiat. Bwl. 36,197-199 Wiesehahn, G. P., Hyde, J. E., and Hearst, J. E. (1977) Biochemistry 16,925-932 Zolan, M. E., Smith, C. A., Calvin, N. M., and Hanawalt, P. C. (1982a) Nature 299,462-464 Zolan, M. E., Cortopassi, G. A,, Smith, C. A., and Hanawalt, P. C. (1982b) Cell 28,613-619