THE JOURNAL OF BIOLOGICAL CHEMISTRY 6) 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 264, No. 21, Issue of July 25, pp. 12226-12231,1989 Printed in U S A .

Site-specific Aflatoxin B1 Adduction of Sequence-positioned Nucleosome Core Particles*

(Received for publication, February 24, 1989)

Richard MoyerSg, Koenraad MarienII, Kensal van HoldeSIl, and George BaileySlI From the Departments of TBiochemistn, and Biorhysics. and ([Food Science and Technology, Oregon State University, Coruallis, Oregon 97331

. . . -

The question of how the presence of nucleosomal packing of DNA modifies carcinogen interaction at specific sites cannot be answered by studies on whole chromatin or bulk nucleosomes because of the hetero- geneity of DNA sequences in the particles. We have circumvented this problem by using nucleosomes that are homogenous in DNA sequence and hence in DNA- histone contact points. A cloned DNA fragment con- taining a sea urchin 5 S gene which precisely positions a histone octamer was employed. By using 32P end- labeled DNA and genotoxins that allow cleavage at sites of attack, the frequency of adduction at every susceptible nucleotide can be determined on sequencing gels. The small methylating agent dimethyl sulfate and the bulky alkylating agent aflatoxin B,-dichloride (AFBI-C12) were used to probe the influence of DNA- histone interactions on DNA alkylation patterns in the sequence-positioned core particle. We find dimethyl sulfate to bind with equal preference to naked or nu- cleosomal DNA. In contrast, AFBI-Cl2 binding is sup- pressed an average of 2.4-fold at guanyl sites within nucleosomes compared with AFB1-ClZ affinity at the corresponding site in naked DNA. The DNA is more accessible in regions near the particle boundary. We observe no other histone-imposed localized changes in AFB1-Cl2 sequence specificity. Further, sites of DNase I cleavage or proposed DNA bending show neither enhanced nor reduced AFBI-C12 adduction to N7-gua- nine. Since AFB1-C1, binding sites lie in the major groove, nucleosomal DNA appears accessible to AFBI- Clz at all points of analysis but with an access which is uniformly restricted in the central 100 nucleotides of the core particle. The data available do not indicate further localized or site-specific perturbations in DNA interactions with the two carcinogens studied.

In eukaryotes, the in vivo target of genotoxic mutagens and carcinogens is not naked DNA but chromatin. Experimental evidence indicates that genotoxic agents bind nonrandomly in eukaryotic chromatin (i-6). Initial studies examined the kinetics or amount of adduct released by micrococcal nuclease or DNase I, but digestion of chromatin often yielded conflict-

* This work was supported in part by Grant BC-504A from the American Cancer Society to K. v. H. and G. S. B. and by National Institute of Environmental Health Sciences Center Grants ES 00210 and ES 03850. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

8 Supported in part by a Tartar Fellowship. Present address: Dept. of Chemistry, Paul M. Gross Chemical Laboratory, Duke University, Durham, NC 27706. Tel.: 919-684-3928.

11 Recipient of an American Cancer Society research professorship.

ing results (7,8). More definitive studies on the chromosomal location of carcinogen adducts utilized more direct ap- proaches, such as the isolation of nucleosomal core particles and oligonucleosomes from carcinogen-labeled chromatin. By injecting trout with [3H]aflatoxin B1, Bailey et al. (3) con- ducted in vivo analysis of aflatoxin B1 (AFB1)’ adduct concen- trations among mono- and oligonucleosomal fractions con- taining differing percentages of core and linker DNA. These data indicated that, in the repair-deficient liver of rainbow trout, linker DNA is about 5-fold more susceptible to adduc- tion than the histone-protected core DNA. Kootstra (5), ex- amining benzo(a)pyrene diol epoxide binding in human Iung cell cultures, found linker to core binding ratios ranging from 14.5 to 1.7, depending on the class of mononucleosomes ex- amined and the time of incubation after treatment during which repair occurred.

At the gene level rRNA gene sequences believed to be non- nucleosomal (9, 10) are adducted 5-fold over bulk nuclear DNA (4). At the single gene copy level, histone H5 and @- globin genes in chicken erythrocytes are more accessible to AFB,-dichloride modification than vitellogenin or ovalbumin genes, which are inactive in these cells (6). All of these previous studies of carcinogen-chromatin interaction have measured only average relative accessibility over extended gene regions or with the whole genome.

The issue of nonrandom carcinogen-genome interaction is further complicated by the recent recognition that the primary sequence of DNA affects the frequency of adduct formation. Sequence-specific effects have been observed for many DNA binding agents displaying a variety of binding modes and sites (11-14). For example, AFB,-N7-guanine binding varies 10- ZO-fold, depending on the nucleotides flanking the adducted site. AFB1-C12, an activated AFB, analogue (Fig. l), displays in vitro the same sequence-specific binding characteristics as the parent compound (15-17). General rules describing this behavior have been enumerated (18-20) though significant exceptions to these have been reported (21). It is important to stress that such rules have been derived entirely using purified DNA, with no information on the extent to which they may apply in chromatin.

To obtain a more precise understanding of the dual effects of local chromatin structure and local DNA sequence selectiv- ity in DNA-genotoxin interaction, we have utilized a mono- disperse population of DNA sequence-positioned nucleo- somes. These reconstituted particles were prepared by histone transfer from chicken erythrocyte core particles to a cloned DNA fragment containing the sea urchin Lytechinw uarie- gatus 5 S rRNA gene and a nucleosome-positioning signal (22, 23). The advantages of this approach are 2-fold: (i) an aver-

’ The abbreviations used are: AFB,, aflatoxin €31; AFB,-CL, 8,9- dichloro-8,9-dihydro-aflatoxin Bl; DMS, dimethyl sulfate; SDS, so- dium dodecyl sulfate; bp, base pair(s).

12226

Aflatoxin B1-Nucleosomal DNA Binding 12227

0 II

H3C-O-S-O"CH3 II 0

AFBI-Clz

FIG. 1. Structures of DMS and AFBI-C12.

aged population of nucleosomes is avoided and (ii) nucleo- some-mediated changes in global and local sequence-specific patterns of DNA alkylation can be quantified at the single- base level.

We have treated these particles with dimethyl sulfate (DMS), a small methylating agent, and the bulky alkylating agent AFBI-C12 to investigate the effects of genotoxin size and DNA-histone interactions on nucleosome-carcinogen binding at specific sites. Both agents attack guanine-N7, ren- dering these sites alkali-labile, which allows precise single base-level analysis on sequencing gels. We find that the DNA- nucleosomal conformation has no effect, either global or local, on DMS W-guanine methylation. By contrast, AFB1-C12 binding is globally suppressed approximately 2.4-fold over most of the nucleosome, with a decrease in suppression near the ends of the nucleosomal DNA. We saw no evidence among available guanyl sites for unusual reactivity within this central region of the particle other than global suppression. Surpris- ingly, no correlation exists between DNase I cutting and AFBl-C12 adduction levels at adjacent sites, indicating that the bulky reagent AFB1-C12 had equal access to guanines located interior or exterior to the histone core.

MATERIALS AND METHODS

Chemicals-Dimethyl sulfate (Aldrich) and piperidine (Sigma) were purchased and used without further purification. AFBI-CI, was synthesized as described (15). Briefly, AFB, (Calbiochem) or [3H] AFB, (Moravek Biochemicals, Brea, CA) was dissolved in dichloro- methane, and a 10% molar excess of Clz (as determined by iodometric titration) was added. After 5 min at room temperature, the solvent was evaporated and the residue dissolved in dichloromethane. AFB,- C1, was separated from AFB, by normal phase high pressure liquid chromatography on a Waters (Milford, MA) pPorasil column with a 97:3 dich1oromethane:acetone solvent system.

Gel Electrophoresis-Both chick erythrocyte and reconstituted nu- cleosomes were examined by SDS or native polyacrylamide mini-slab gels (Idea Scientific, Corvallis, OR). To assess histone integrity, Laemmli-type SDS gels were run (24) and the gels stained with Coomassie Blue. Nucleosomes were analyzed on particle gels after Todd and Garrard (25) where 1-mm-thick 5% acrylamide gels, with a 1:20 bisacrylamide to acrylamide ratio, were run to 10 V/cm, 4 "C, for 1 h. For sequencing, 40-cm 8% polyacrylamide slab gels as de- scribed by Maxam and Gilbert (26) were run at 50 watts constant power for 2-4 h.

Chick Erythrocyte Mononucleosome Preparation-Nucleosome core particles were prepared from chicken erythrocytes following Lutter (27) with modifications as described by McMurray and van Holde (28) and stored on ice. SDS-gel electrophoresis revealed either trace amounts or the complete absence of histone degradation in the several preparations used.

Plasmid Preparation-Escherichia coli HBlOl containing plasmid pLv405-10 was obtained from Dr. R. T. Simpson, NIH (23). Plasmid was isolated by the methods of Sakonju et al. (29) and Clewell and Helinski (30) and then restricted with EcoRI to isolate the 260-bp

fragment containing the 5 S rRNA gene (see Fig. 2). This fragment was either 3' or 5' 32P end-labeled as described (23). For 3' labeling, a 15-min chase with cold dTTP assured flush ends. Secondary restric- tion was performed with NciI, and the resultant 195-bp fragment was purified by gel electrophoresis and then passively eluted from the excised gel strip (31). For brevity, we shall henceforth refer to this fragment as "5 S DNA."

Reconstitution-Chick erythrocyte nucleosomes and free 32P end- labeled 5 S DNA were combined in 300 p1 of 10 mM Tris-HC1 (pH 7.5), 1 mM EDTA (TE), and 0.8 M NaC1. To achieve a ratio of 1:200 to 1:250 pieces of 5 S DNA to chick core particles, 81-100 ng of the 195-bp fragment of 5 S DNA was added to 15 pg of core particles, as determined by absorbance at 260 nm on either a Cary 2200 or a Bausch and Lomb 601 spectrophotometer. After 20 min on ice, during which exchange of histones occurs between chicken nucleosomes and 5 S DNA, dialysis tubing (3500-dalton cutoff) was placed over the tube, secured, and the tube inverted in 200 ml of TE, 0.6 M NaC1, and 1 mM phenylmethanesulfonyl fluoride. Dialysis with stirring a t 4 "C was continued for at least 12 h; then the tube was transferred to 200 ml of TE, 0.05 M NaCI, 1 mM phenylmethanesulfonyl fluoride for at least 3 h. After each reconstitution, a small amount of this preparation would be run on mini 3.5% polyacrylamide gel electro- phoresis, as described above. Gels of reconstituted particles contain- ing the 5 S DNA were dried and then exposed to x-ray film for 12-16 h, usually without the aid of intensifying screens. Bands were quan- tified with a Zeineh laser scanner (Biomed Instruments Inc., Brea, CA). Care was taken to scan only within the linear range of the film. Subsequent film scanning demonstrated 90-97% incorporation of the 5 S DNA into core particles with no detectable free 5 S DNA (Fig. 3).

DNase I Cleaoage-DNase I (Worthington) reactions were per- formed in 10 mM Tris (pH 7.5), 1 mM EDTA, 3 mM MgClz at 4 "C. For digestions of free DNA, a trace amount (10-30 ng) of 32P-labeled 5 S DNA fragment was combined with 4 pg of carrier (deproteinized core particle) DNA. One unit of DNase I was added and the reaction quenched after 20 s by the addition of EDTA to 12 mM. For DNase I digestions of reconstituted core particles, identical conditions were used, excepting 4 units of DNase I for a 60-s digestion.

DNA A2kylation"Control DNA was purified from chicken eryth- rocyte nucleosomes by Proteinase K (Boehringer Mannheim) diges- tion (0.5 mg/ml), phenol and chloroform extraction, and ethanol

-10 I I I I I

0 10 20 30

AATTCCAACGAATAACTTCCAGGGATTTATAAGCCGATGACGTCATAACA GGTTGCTTATTGAAGGTCCCTAAATATTCGGCTACTGCAGTATTGT

40 50 I I I I I

6 0 70 8 0

TCCCTGACCCTTTAAATAGCTTAACTTTCATCAAGCAAGAGCCTACGACC AGGGACTGGGAAATTTATCGAATTGAAAGTAGTTCGTTCTCGGATGCTGG

90 I

100 I

110 120 130 I I I

ATACCATGCTGAATATACCGGTTCTCGTCCGATCACCGAAGTCAAGCAGC TATGGTACGACTTATATGGCCAAGAGCAGGCTAGTGGCTTCAGTTCGTCG

140 I I

150 160 I .1 I

180

ATAGGGCTCGGTTAGTACTTGGATGGGAGACCGCCTGGGAATACCGGGTG TATCCCGAGCCAATCATGAACCTACCCTCTGGCGGACCCTTATGGCCCAC

t 19 0 200 210 220 230

T T G T A G G C T T T T T T T T C T C C C C C C C C C C C C T C T T T G C T T C G C C AACATCCGAAAUAAAGAGGGGGGGGGGGGAGAAXGAAGTACTTTACGG

I I I I I

2 4 0 I

TCTTGG AGAACCTTAA

FIG. 2. Sequence of the cloned L. variegatus 5 S rRNA gene fragment. The underline shows the nucleosome-binding region, and the arrows depict the NciI site at 175.

12228 Aflatoxin B,-Nucleosomal DNA Binding

precipitation (32). For alkylation, DNA samples were dissolved in 0.2 ml of 50 mM NaCI, 10 mM Tris (pH '7.5), and 1 mM EDTA. Equivalent amounts (3 .5 pg) of DNA, either naked or in reconstituted nucleo- somes, were treated in parallel with DMS or AFBl-CI, on ice. DMS was added to a final concentration of 200 p~ and the reaction quenched after 1 min by the addition of dithiothreitol to 2 mM. AFBl- C I S in dichloromethane was added to the reaction tube, and the solvent was evaporated under a gentle stream of dry nitrogen. Upon addition of buffer and DNA or nucleosomes, the AFRl-Cl2 was dissolved by vigorous shaking. The final concentration of AFRI-C12 was 1.7 FM. Atter 5 min on ice, the reaction was quenched with dithiothreitol as above. An aliquot of each sample was analyzed on a particle gel to ensure particle integrity after alkylation (Fig. 3). Samples were then made 0.5'0 SDS, 0.5 mg/ml Proteinase K, and incubated for 30 min at 37 "C. NaCI was added to a final concentration of 0.2 M. After extractions with 1 volume each of phenol, phenol:chloroform, and chloroform, the DNA was ethanol-precipitated and then dissolved in 0.1 ml of glass-distilled H20. Each sample was made 3% (v/v) piper- idine, then placed in a 90 "C heating block. After 20 min, each tube cap was pierced with a needle, and then samples were frozen and lyophilized. Each pellet was redissolved with 10-20 pl of water and counted in a Packard Tri-Carb liquid scintillation counter. After two additional lyophilizations, sufficient 90% (v/v) formamide (2-20 pl ) was added to make each sample 10,000 cpmlpl. Equivalent volumes (usually 2 P I ) of each sample were then applied to sequencing gels for resolution of fragments.

Quantitation of DMS or AFBl-C12 Binding-Sequencing gels were fixed, dried, and then exposed to film for 12-96 h in order to obtain a range of film intensities. Care was taken to scan within the linear range of the film, as determined by a series of exposures with "'P standards. Alternately, film was preflashed as described by Laskey and Mills (33). In order to eliminate errors due to irregularities of band shape or width, films were turned 90 and scanned across the two lanes a t bands being compared. This also allowed more accurate quantitation of less intense bands, as scanner gain could be adjusted to maximize signal to noise ratios, settings which often put adjacent darker bands off scale. A few weakly adducted AFBI-C1, sites were judged insufficiently above film background to be included in the cleavage map (Fig. 6). Additionally, adjacent pairs or triplets of binding sites toward the right side of the particle (Fig. 6) beginning with 100,10! on the W strand, and 32 on the C strand were scanned and quantified as a single pooled average ratio of DNA:core binding and are depicted accordingly.

AFB,-CI?:DNA ratios were chosen so that less than one adduct formed per DNA fragment or nucleosome. This was readily estab- lished by determining densitometrically the amount of uncut DNA in AFR,-Cl?-treated samples, compared with control DNA. The prob- ability (P) that no cuts occurred in the treated samples was obtained by dividing the area of the trace from the alklyated sample by that of the unalkylated sample. The average numher of cuts/strand ( Y ) was then calcuated using the Poisson distribution function

P = exp(-Y)

as described (21, 34). At less than 0.5 cleavages/strand, Poisson statistics indicate less than 10% of the fragments cut twice, preventing a significant bias toward shorter fragments.

RESULTS

Precision of Reconstitution-For these studies to be inter- pretable, it is essential that the reconstitution be nearly quantitative, with little or no free DNA remaining, and that the positioning of the histone core on the DNA fragments be precise and uniform. Further, it is important that the proce- dures employed do not nick the DNA in sites other than those alkylated by DMS or AFB1-C12.

Fig. 3 shows a typical autoradiograph of a particle gel from a reconstitute. Such experiments showed that 90-97% (deter- mined densit.ometrically) of the labeled DNA was routinely incorporated into nucleosomes and that neither DMS or AFBI-C12 treatment caused detectable DNA dissociation. (The remaining 3-10% of the labeled fragment, which mi- grates above the core particle, is likely a core particle with additional histones attached.)

Fig. 4 is an autoradiograph of a typical sequencing gel.

1 2 3 4 5

FIG. 3. Reconstituted nucleosomes: treatment with DMS or AFB,-Cl,. Outer lanes are ."P-laheled 5 S DNA; inner 3 lanes of reconstituted particles are control, exposure to 10 mM DMS for 1 min, and exposure to 1.1 p~ AFRI-CI? for 5 min, respectively, followed by immediate loading on particle mini-gels as descrihed in text. AFBl- Cl? lane, overloaded, shows the absence of any detectable free DNA caused by dissociation.

O M S B,-C12 DNASE I

1 2 3 4 5 6 7

FIG. 4. Dimethyl sulfate, aflatoxin B,-CI,, or DNase I cleav- age of free DNA and reconstituted core particles. The DNA fragment (W strand) was prepared hv cutting plasmid pLV40.5 with EcoRI, :'lP 5' end labeling with T4 kinase, then secondary restriction with NciI. The 195-bp fragment, either as "naked" DNA (lanes 2, 4, and 6 ) or reconstituted into nucleosomes by salt exchange (lanes 3, 5 , and 7), was treated with 200 p~ DMS for 1 min (lanes 2 and 3 ) , 1.7 p~ AFR,-CI, for 5 min (lanes 4 and .5), or DNase I (lanes 6 and 7). All reactions were carried out at 4°C. Control (lane 1 ) was carried through all steps except alkylation or DNase I cleavage. DNA was phenol/chloroform-extracted from each sample and ethanol-precipi- tated hefore loading on 8 5 polyacrylamide sequencing gels.

Lanes 6 and 7 show DNase I cleavage patterns for histone- free 5 S DNA and reconstituted core particles, respectively. Protected regions (approximately 10 bp) in much of the core particle as opposed to frequent sequence selective cutting throughout the naked DNA demonstrate a uniquely posi-

Aflatoxin B1-Nucleosomal DNA Binding 12229

tioned particle. The fact that several regions of preferred cutting appear as single bands is evidence for high precision in positioning; if several alternate positions were present, these bands would be multiples. In gels run twice as long to better visualize longer fragments, DNase I cutting appears identical in both nucleosomes and naked DNA above approx- imately 145 bp (see Fig. 5 for numbering scheme). Addi- tionally, AFB1-C12 cutting frequencies reflect a boundary at about 145 bp (see below). Taken together, these results indi- cate a histone octamer located approximately between 20 and 165 bp from the labeled end in full agreement with the results reported by Simpson and Stafford (23).

To determine background cleavage levels from sources other than alkylation, DNA (Fig. 4, lane I ) was exposed to all steps, save alkylation with DMS or AFB1-Clz. Although some faint bands cleaved by piperidine alone can be seen from this length of film exposure the region in which the histone core will be placed is nearly free of such adventitious cutting.

Alkylation with Dimethyl Sulfate-Fig. 4, lanes 2 and 3, represents DMS-DNA binding and subsequent guanine-spe- cific cleavage of free DNA or DNA residing on reconstituted core particles, respectively. These results are depicted more fully in Fig. 5. For the strand which we arbitrarily label W, the average ratio of DMS-DNA binding at each guanyl site in free DNA compared with binding at that same site in core particle DNA is 0.99 + 0.07 for 25 scanned sites. The highest ratio of binding, approximately 1.06, occurs a t position 19, and the lowest, 0.87, occurs at position 3 within the particle. For the C strand (data not shown), the average ratio for 32 scanned sites is 1.04 + 0.05. The ratio of binding ranges from 1.16 at position 137 to 0.86 a t position 24. Thus we can quantitatively state that the DNA conformation in this nu- cleosomal particle neither inhibits nor promotes DMS meth- ylation of guanine-N7. McGhee and Felsenfeld (35) using “native” rooster erythrocyte core particles of heterogenous sequence also observed a generally uniform pattern of N7- guanine-DMS accessibility, though site 62 in their bulk nu- cleosomes displayed enhanced reactivity.

Alkylation with Aflatoxin B1-Dichloride-In Fig. 4, lanes 4

FIG. 5. Free DNA:core DMS- DNA binding ratios, W strand. The W strand of the 195-bp fragment was 5’ end-labeled, then reconstituted into nu- cleosome core particles or mixed with deproteinized DNA from chicken core particles before DMS treatment. Posi- tion 0 is the left boundary of the particle; the terminal 32P label (not shown) is a t -20. DMS binding ratios for each site were calculated from laser densitometry scans of three separate experiments. A ratio of 1 represents equivalence in DMS binding to free or core particle DNA a t that site. The average binding to all sites was 0.99 +- 0.07.

1.2

1

8

Ld K 0 .6

6 z 9 n

.4

.2

0

and 5 represent the results of AFBl-C12 adduction to free 5 S DNA or reconstituted core particles, respectively. Unlike DMS, AFB1-Cl2 demonstrates inhibited adduct formation to nucleosomal DNA. Surprisingly, the marked sequence speci- ficity of AFBl-C12 (18-21) is unchanged by DNA-histone contacts. Although there is an overall repression of adduction within the region of histone interaction, the profile of guanyl site selectivity within this repressed region did not differ from that in the free DNA reactions. These results are more fully depicted in Fig. 6. For the W strand, AFB1-C12 binding to free versus nucleosomal DNA averages 2.57 f 0.34 from position 2 to 121 within the particle. No guanyl site within this region departs significantly from this ratio, which indicates the ab- sence of any highly localized histone-mediated effects among the guanyl sites available in this particle. From position 126 to 145 the ratio decreases toward unity. Past 145 values are near unity, as would be predicted for reaction with non- nucleosomal or linker DNA. (Values at several guanines are unreported, where weak sequence-specific AFB1-Cl2 binding to both free and particle DNA prevented precise quantitation. Such sites are marked with triangles in Fig. 6.) For the C strand, the AFB1-Cl, binding ratio is near unity for positions -1 and 0, then averages 2.17 + 0.54 to position 118. The ratio a t site 137 is 1.07 and at site 148, just beyond the particle boundary, it is 1.00.

Comparison of AFBl-C12 Binding and DNase I Cleavage Sites-Fig. 6 also displays DNase I cutting sites and their frequencies for both W and C DNA strands. To determine if AFBI-C12 binding ratios (upper) correlated with DNase I cleavage patterns (lower), binding ratios of AFB1-ClZ to free versus nucleosomal DNA were averaged for each distance of 0, kl, +2, +3, f 4 , and 25 bases from major DNase I cleavage sites. The resultant average binding ratios for each distance from DNase I cleavage all fell within 1 S.D. of the strand average (data not shown). As there are no significant differ- ences in AFB1-Cl, adduction from 0 to +5 bases from DNase I cutting, we conclude that the steric hindrance which mod- ulates DNase I cleavage of the reconstituted nucleosomes does not affect AFB1-Cl2 nucleosomal DNA binding. The N7 posi-

r r r

12230 Aflatoxin B1-Nucleosomal DNA Binding 3 ,

i ' l : "_

3

1 ""L""""

I

0 o 0 n 0 n 0 o 0 o o u - l O n O n O n O o o n o o o ~ o m o ~ o m o * - - ~ ~ ~ n t ~ n u - l ~ ~ ~ h m m r n r n o o - - ~ ~ n n ~ ~ o ~ ~ ~ """""""

FIG. 6. Free DNA:core aflatoxin BI-Clz DNA binding ratios. The W strand of the 195-bp fragment was 5' end-labeled, or alter- nately, the C strand was 3' end-labeled, then reconstituted into core particles. Position 0 is the left boundary of the particle; the terminal

P label (not shown) is at -20. The box (center) represents the position of the nucleosomal particle on the 5 S DNA strand. Upper, AFB1-C12 binding ratios at guanyl sites for free DNA uersus nucleo- some core particle DNA. A ratio of 1 (dashed line) represents equiv- alence in AFBI-C12 binding to free DNA or particle DNA, and a ratio of 2 represents a guanyl site that is twice as likely to bind aflatoxin AFB1-C12 when in a histone-free uersus nucleosomal (histone-bound) state. Binding at each site was calculated from scans of three separate experiments. Average binding ratios for all sites within the central region of the particle: W strand = 2.57 k 0.34; C strand = 2.17 ? 0.54. Sites where AFB,-C12 was insufficiently above film background to be quantitated are represented by triangles. Lower, DNase I frac- tional cleavages of reconstituted particles, depicting the probability of DNase I cutting at each site on the W or C strands.

32

tion of guanine appears equally accessible to AFB1-C12 whether it is facing the nucleosomal histone core or the solvent.

DISCUSSION

DNase I cleavage patterns of the reconstituted 5 S DNA nucleosome particle used for these studies closely parallel those obtained by Simpson and Stafford (23). Minor differ- ences are likely due to differing extents of digestion. DMS, a relatively small methylating agent, was found in this study to bind equally efficiently to this DNA fragment, whether in a reconstituted nucleosome particle or free in solution. Our findings of DMS-nucleosomal DNA binding are generally consistent with those of McGhee and Felsenfeld (35) with

two exceptions. First, McGhee and Felsenfeld observed a clear increase in nucleosomal DNA-DMS binding near position 62 on both strands, although they did not estimate its magnitude. We find DMS-naked DNA uersus DMS-nucleosomal DNA binding ratios from 1.05 to 0.96 on both strands between sites 59 and 69. At site 62 on the C strand, the DMS binding ratio is 0.98. We do find larger differences a t other sites, from 0.86 to 1.16. These small differences may reflect data scatter rather than actual modulation of DMS binding. We cannot rule out the possibility that this difference results from the fact that we are studying a particular DNA and nucleosome, whereas McGhee and Felsenfeld were examining a broad distribution of sequences.

AFBI-C12, a large molecule compared to DMS (Fig. l), is hindered in its binding to nucleosomal DNA. Surprisingly, neither local DNA-histone contacts nor DNA helical perio- dicity seem to further modulate N7-guanine accessibility; there is no correlation between sites devoid of DNase I cutting and AFBl-C12 binding nor do major DNase I sites display enhanced binding. X-ray crystallography has revealed sites of sharp DNA bending located k10 and -+40 bp from the nucleo- somal dyad axis, e.g. a t about sites 30, 60, 80, and 110 (36). We did not observe unusual binding ratios in these regions of our particle. Furthermore, even the orientation of the DNA toward the histone core seems to be unimportant. Note for example the behavior a t sites 14 and 15 on C and 16 on W. All are repressed sites, even though they lie almost across the DNA dyad. About 5 bp further, we have sites 21 on C and 22 on W, which display the same behavior. Due to the helical periodicity of DNA, it is impossible for all of these sites to be buried against the histone core.

The 2.4-fold global protection we do observe decreases toward unity in three of the four DNA strand ends as the particle boundary is approached. The right side of the particle (as diagrammed in Fig. 6) is increasingly accessible to AFB1- Clz binding in both strands, beginning about 20 bp from the particle boundary. In native nucleosome core particles, 20-35 bp on each end melts a t a lower temperature (37, 38). At the left end of our particle, C strand binding reflects a similar transition from protected sites 14 and 15 to sites 0 and -1, which bind AFB1-C12 as well as free DNA. The W strand, at sites 2, 3, and 4, is protected from AFBI-C12 binding to a greater extent than the other three strand ends, at a level consistent with the central 100 bp of the 5 S particle. Studies of histone DNA contacts in the average nucleosomal core show 5' strand ends free of histone contacts for about 20 bp, while 3' strand ends are reported to be bound to histone H3 (39). The reconstituted particle used in these experiments displays no such symmetry of AFBl-C12 binding to 5' ends, perhaps reflecting DNA sequence-specific effects unobserva- ble in averaged particles. The 3' ends, it should be noted, do reflect symmetrical AFB,-Cl, binding patterns. In this case, however, histone H3 contact (if it exists in these specific particles) does not confer protection, but instead the binding affinity approaches that of free DNA.

A relative freedom of the DNA ends not experienced by the central IC9 bp of the particle may explain observed AFB1-C12 binding patterns. One possibility is that DNA-histone inter- action may suppress precovalent binding. AFBl has a precov- alent association with DNA characterized by strong AFB, hypochromism (40) and the observation that activated AFB, covalently binds 4-fold more strongly to double helical rather than single-stranded DNA.' This precovalent association is apparently groove binding, as AFBI, unlike the intercalating

K. Marien, unpublished data.

Aflatoxin B1-Nucleosomal DNA Binding 12231

agent ethidium bromide, will not induce positive supercoils in plasmid DNA, even at a molar ratio of 10,000:l.3

Gale et al. (41, 42) recently reported UV photoproduct formation in whole cells and in mononucleosomes which displays a striking 10.3-bp periodicity, presumably corre- sponding to sites where the DNA phosphate backbone is farthest from core histones. Both a significant change in adjacent pyrimidine orientation and local distortion of the DNA helix are required for pyrimidine dimer formation. As N7-guanine is accessible from the major groove, AFBI-CI:! binding may cause little helix distortion, accounting for the absence of periodicity in binding ratios we observe. In this connection, it should be noted that McGhee and Felsenfeld (43) found that T4 DNA, glucosylated in the major groove, readily forms nucleosomes. From the above, we conclude that the whole major groove must be quite accessible in nucleoso- mal DNA. Further, we predict that only those genotoxins which produce significant helix distortion upon DNA-adduct formation would display local site-specific modulation of DNA damage by DNA-histone interaction.

Earlier work in these laboratories (3) found a 5:l internu- cleosomal (linker) DNA to core DNA ratio of adduction in vivo in liver DNA when rainbow trout were injected intraper- itoneally with [3HJAFBl. A similar ratio was observed in AFBl adduction to ribosomal RNA genes, which were &fold more accessible than the remaining bulk nuclear DNA. Ribosomal RNA genes appear to be non-nucleosomal (9, lo) and may behave similarly to linker DNA. In our particle, the free DNA ends or ’‘linker’’ DNA are only adducted 2.4-fold more than core DNA. This discrepancy may be due to an additional level of core DNA protection conferred by chromatin condensation. The system used in this study only assesses differences of AFB1-C12 adduction due to the first level of chromatin pack- aging, the nucleosomal core particle.

I t must be remembered that homogenous sequence-posi- tioned particles are necessarily a subset of the “average” nucleosome core particle. Further, the DNA fragment we have used does not place guanyl sites in every possible nucleosomal position; hence data are not available on all nucleosomal locations. Though we cannot rule out other nucleosome con- formations where local DNA-histone contacts further modu- late AFBI-C12 or DMS binding, the major conclusion from our studies is that individual guanine bases are neither mark- edly exposed to nor shielded from alkylation by these two agents. Based on the data available, we conclude that the covalent and/or precovalent binding of AFB1-C12 to nucleo- somal DNA is suppressed by a global DNA conformation and is not further modulated by local histone interactions or localized DNA bending.

Acknowledgments-We wish to express our appreciation to Dr. R. Simpson, who provided the DNA clone utilized in these experiments and to Valerie Stanik for preparing chicken core particles and plas- mid.

REFERENCES

1. Jahn, C . S., and Litman, G. W. (1979) Biochemistry 18, 1442-

2. Koostra, A., Slaga, T. J., and Olins, D. E. (1979) Chem.-Biol.

3. Bailey, G. S., Nixon, J. E., Hendricks, J. D., Sinnhuber, R. O.,

1449

Interact. 28, 225-236

K. Marien and R. A. Moyer, unpublished data.

4.

5. 6.

7. 8.

9. 10. 11.

12. 13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24. 25.

26.

21. 28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38. 39.

40.

41.

42.

43.

and Van Holde, K. E. (1980) Biochemistry 19,5836-5842

U. S. A. 81,664-668 Irvin, T. R., and Wogan, G. N. (1984) Proc. Natl. Acad. Sci.

Kootstra, A. (1986) Nucleic Acids Res. 14, 9897-9909 Delcuve, G. P., Moyer, R., Bailey, G., and Davie, J. R. (1988)

Jack, P. L., and Brookes, P. (1982) Carcinogenesis 3, 341-344 Arrand, J. E., and Murray, A. M. (1982) Nucleic Acids Res. 10,

Scheer, V. (1980) Eur. J . Cell Biol. 2 3 , 189-196 Labhart, P., and Koller, T. (1982) Cell 28, 279-292 Boles, C. T., and Hogan, M. E. (1986) Biochemistry 25 , 3039-

Dervan, P. D. (1986) Science 232,464-471 Burns. P. A.. Gordon. A. J . E.. and Glickman. B. W. (1987) J.

Cancer Res. 48 , 7146-7149

1547-1555

3043

Mol.’ Biol. 194, 3851390 ’

Hartlev. J. A.. Gibson. N. W.. Kohn. K. W.. and Mattes. W. B. (1986j Cancer Res. 46 , 1943-1947 ’

Res. 35,3811-3823

and Humayan, M. 2. (1987) Mutat. Res. 179,89-101

Cancer Res. 48,5391-5396

chemistry 22,3351-3359

Natl. Acad. Sci. U. 9. A. 80, 6-10

(1988) Biochemistry 27,472-481

G. (1987) J. Biol. Chem. 2 6 2 , 7455-7462

Res. 8, 1839-1853

U. S. A. 80, 51-55

Swenson, D. H., Miller, J. A., and Miller, E. C. (1975) Cancer

Jacobsen, J. S., Refolo, L. M., Conley, M. P., Sambamurti, K.,

Wood, M. L., Lindsay Smith, J. R., and Colin Garner, R. (1988)

Misra, R. P., Muench, K. F., and Humayan, M. Z. (1983) Bio-

Muench, K. F., Misra, R. P., and Humayan, M. Z. (1983) Proc.

Benasutti, M., Ejadi, S., Whitlow, M. D., and Loechler, E. L.

Marien, K., Moyer, R., Loveland, P., Van Holde, K., and Bailey,

Lu, A. L., Steege, D. A., and Stafford, D. W . (1980) Nucleic Acids

Simpson, R. T., and Stafford, D. W. (1983) Proc. Natl. Acad. Sci.

Laemmli, U. K. (1970) Nature 227,680-685 Todd, R. D., and Garrard, W. T. (1977) J. Biol. Chem. 252 ,

Maxam, A. M., and Gilbert, W. (1980) Methods Enzymol. 65,

Lutter, L. C. (1978) J . Mol. Biol. 124 , 391-420 McMurray, C . T., and van Holde, K. E. (1986) Proc. Natl. Acad.

Sakonju, S., Bogenhagen, D. F., and Brown, D. D. (1980) Cell 19,

Clewell, D. B., and Helinski, D. R. (1970) Biochemistry 9,4428-

Maniatis, T., and Efstratiadis, A. (1980) Methods Enzymol. 6 5 ,

Britten, R. J., Graham, D. E., and Neufeld, B. R. (1974) Methods

Laskey, R. A,, and Mills, A. D. (1975) Eur. J . Biochem. 56,335-

4729-4738

499-560

Sci. U. S. A . 83, 8472-8476

13-25

4440

299-305

Enzymol. 29,363-418

341 Boles, T. C., and Hogan, M. E. (1984) Proc. Natl. Acad. Sci.

McGhee, J. D., and Felsenfeld, G . (1979) Proc. Natl. Acad. Sci.

- ”

U. S. A. 81, 5623-5627

U. S. A. 76,2133-2137 Richmond, T..J., Finch, J. T., Rushton, B., Rhodes, D., and Klug,

Weischet, W . O., Tatchell, K., van Holde, K. E., and Klump, H.

Simpson, R. T. (1979) J. Biol. Chem. 284,10123-10127 Shick, V. V., Belyavsky, A. V., Bavykin, S. G., and Mirzabekov,

Clifford, J. I., and Rees, K. R. (1969) Biochem. Phrmacol. 18 ,

Gale, J. M., Nissen, K. A., and Smerdon, M. J. (1987) Proc. Natl.

Gale, J. M., and Smerdon, M. 3. (1988) J. Mol. Biol. 204, 949-

McGhee, J. D., and Felsenfeld, G. (1982) J. Mol. Biol. 158, 685-

A. (1984) Nature 3 11, 532-537

(1978) Nucleic Acids Res. 5, 139-159

A. D. (1980) J. Mol. Biol. 139, 491-517

2783-2785

Acad. Sci. U . S. A. 84,6644-6648

958

698