DNA Damage Induced by Bleomycin, Neocarzinostatin, and ...

Vol. 269,No. 48,Issue of December 2, pp. 30587-30594, 1994 Printed in U.S.A.

JOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc. “HE

DNA Damage Induced by Bleomycin, Neocarzinostatin, and Melphalan in a Precisely Positioned Nucleosome ASYMMETRY IN PROTECTION AT THE PERIPHERY OF NUCLEOSOME-BOUND DNA* (Received for publication, July 25, 1994, and in revised form, September 26, 1994)

Bettye L. Smith$, Gwen B. Bauer, and Lawrence

E PovirkO

From the Department of Pharmacology and Toxicology, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia 23298

mately leads toboth therapeutic and toxic effects, the interactions of these drugs with purified DNA have been studied extensively. However, in eukaryotic cells, chromatin structure imposes an additional level of specificity. Since this specificity can influence the degree t o which damage is localized in essential DNA sequences or in sequences where damage is poorly repaired, itmay be an important determinantof the biological effects of these drugs. The basic subunit of chromatin is thenucleosomal core particle, which consists of approximately 146 bp’ of DNA coiled in a left-handed superhelix around a n octamer of core histones. The structural organization of DNA into nucleosomes is considered an important factor in the regulation of gene expression, and nucleosomes have been shownto inhibit transcription in vitro(van Holde, 1989).Additionally, nucleosomes have been shown to precisely position around important regulatory regions, both in vivo and in vitro (Ramsay, 1986; Rhodes, 1986; Simpson and Stafford, 1983). For example, a nucleosome is positioned around thestart site of the 5 S rRNAgene extending into the transcriptionfactor binding site ina variety of species including Lytechinus uariegatus (Moyer et al., 1989; Simpson (Hayes et aZ., 1990; and Stafford, 19831, Xenopusborelis Rhodes, 1986), andxenopus Zaeuis (Smith andMacLeod, 1993). In the presentstudy, we have taken advantageof the precisely positioned nucleosome at the transcription start site of the X. Zaevis 5 S rRNA gene to determine, at single nucleotide resolution, the degree to which nucleosome structure modulates DNA damage by three antitumor drugs, bleomycin, neocarzinostatin, and melphalan. Bleomycin and NCS are antibiotics which induce sequencespecific single and double strand breaks in DNA byfree radicalbasedmechanisms (for a review, see Dedon and Goldberg (1992)). Bleomycin is a glycopeptide which forms a metal coordination complex with Fe”. The bleomycin.Fe” complex combines with oxygen to produce a highly reactive species which Many chemotherapeutic drugs damage DNA; in fact, this specifically abstracts hydrogen from the C-4’ of the deoxyridamage is probably responsiblefor their therapeuticefficacy, as bose, ultimately leading toa strand break orabasic site. Early well as for adverse side effects such as myelosuppression and studies showed that bleomycin induces single strand breaks second malignancies in long term survivors of chemotherapy primarily at pyrimidines in G-C and G-T sequences. More re(Henne and Schmahl,1985; Pratt and Ruddon, 1994). In order cently, Steighnerand Povirk (Steighnerand Povirk,1990) to clarify how chemical modification of DNA by chemothera- showed that bistranded cleavage resulted from a strand break peutic agents affects DNA structure and function, and ultiat a primary site (which adhered to the expected G-C or G-T sequence specificity) followed by a secondary cleavage event on * This work was supported by Grant CA40615 and National Research the opposite strand. Since the secondary cleavage sites, which Service Award CA09564 from the National Cancer Institute, United occur either directly opposite the primary site or opposite the States Department of Health and Human Services. The costs of publi- base immediately following it, are rarely G-C or G-T sequences, cation of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” no true single strand cleavage occurs at most secondary sites, and thus cleavage at these sites can be taken asa measure of in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Present address: Qiagen Inc., Chatsworth,CA 91311. double strand cleavage. 5 To whom correspondence and reprint requestsshould be addressed: Medical College of Virginia, P. 0. Box 980230 MCV Station, Richmond, VA 23298. Tel.: 804-828-9640;Fax:804-828-8079. The abbreviations used are: bp, base paifis);NCS, neocarzinostatin.

The antitumor drugs bleomycin, neocarzinostatin, and melphalan all damage DNA by mechanisms which involve binding in theminor groove.In orderto examine at high resolution the modulating effects of chromatin structure on the action of these drugs, an end-labeled DNA fragment from the Xenopus Zaeuis 5 S rRNA gene was reconstituted with histone octamers to form a precisely positioned nucleosome. For each drug, DNA damage at specific sequence positions in the fragment was then compared for nucleosome-bound uersus naked DNA. Reconstitution into nucleosomes resulted in a marked inhibition of the DNA cleavage induced by bleomycin &fold)and neocarzinostatin (2.4-fold)in thecentral region of nucleosomal DNA. However, at the periphery of nucleosome-bound DNA, a distinct asymmetry was apparent, with marked inhibition of cleavage toward the upstream side, but little if any inhibition toward the downstream side, which overlaps the binding site of the transcription factor TFIIIA. In the case of melphalan, alkylation at adenine N-3 was inhibited by nearly 2-fold throughout the nucleosome, whereas alkylation at guanine N-7 was either slightly inhibited or slightly enhanced, depending onsequenceposition. None of the drugs showed the 10-base pair periodicity characteristic of hydroxyl radical-induced cleavage of nucleosomal DNA. The results are consistent with a model in which minor groove sites in nucleosome-bound DNA remain relatively accessible to small molecules, even where the minor groove facesthe histone core, and in which drug-induced DNA damage is inhibited by conformational constraints imposed onDNA by nucleosome structure. Furthermore, the degree of such constraints appears to be sequence-dependent,at least near theperiphery of nucleosome-boundDNA.

~~~~

~

30587

30588

Nucleosome Damage

by Bleomycin, Neocarzinostatin, and Melphalan

NCS is one of a family of enediyne antibiotics, whichcontains a hydrophobicchromophore as the activecomponent. Upon addition of thiol, the chromophore forms a highly reactive biradical species which can induce bistranded damageby simultaneously abstracting hydrogen from C-5' of deoxyribose in one DNA strand, and from C-1' or C-4' in the opposite strand. NCS thus induces frank double strand breaks at AGT.ACT (5' + 3l.5' "-* 3') sequences, as well as abasic sites withclosely opposed strand breaks at AGC.GC.T sequences, which rapidly decompose inchromatinto yield additional double strand breaks (Bennett and Povirk, 1992; Dedon and Goldberg, 1990; single strand breaks Povirk et al., 1988). The drug also induces at nearly all T and at many A residues, and most of these breaks have 3'-phosphate and 5"aldehyde termini (Kappen and Goldberg, 1983). Adduct formation in nucleosomal DNAby the alkylating agent melphalan was also examined, primarily to determine whether adenineN-3 (Osborne andLawley, 1993) and guanine N-7 alkylations are differentially inhibited in chromatin. Previous studies with other bulky adduct-forming agents have shown that core histones provide approximately equal protection of DNA from either benzo[a]pyrene diol epoxide, which binds in the minor groove at guanine N-2 (Smith andMacLeod, 1993) or aflatoxin B1 (Moyer et al., 19891, which binds in the major groove a n guanine N-7 (Moyer et al., 1989); in each case there was a 2-3-fold decrease in covalent binding to the nucleosomal region compared with the linker region. However, since melphalan-induced alkylation at adenine N-3, a minor groove site, occurs mostly in APT-rich tracts (Wang et al., 19911, and since APT tracts tend to be oriented with theminor groove facing the nucleosome core, selective suppression of adenine alkylation might be expected a priori. Such selective suppression could explain the relatively low frequency of A.T 4 T.A transversions induced by this drug in an endogenous gene (Austin et al., 19911, despite the predominance of such mutations ina mammalian shuttlevector system (Wangetal., 1990). In addition t o determining how chromatin structure modulates theeffects of specific DNA-damaging agents, examination of damage in chromatin may help todefine the accessibility of both major and minor groove sites in nucleosomal DNA t o interaction with exogenous molecules, as well as thesusceptibility of various domains of nucleosomal DNA to induced conformational distortion. Such information may provide insights into the interaction of larger molecules, such as DNA-binding proteins, with DNA in chromatin.

(Moyer et al., 1989; Smith and MacLeod, 1993). Briefly, core particles and linear 32P-end-labeled 5 S DNA were combined a t a napproximate molar ratioof 1OO:l in 670 pl ofTE(10 m~ Tris, pH 8,O.l mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride). 5 M NaCl was added by dropwise addition to a final concentration of 0.8 M NaCl, and the samples were allowed to reston ice for 20 min. The samples were placed in Spectropor no. 1 dialysis tubing and dialyzed uersus TE containing 0.6 M NaCl overnight a t 4 "C. The samples were then dialyzed against TE containing 0.05 M NaCl for 4 h a t 4 "C. Control samples were preparedconcurrently by performingamock-reconstitution,substitutingpurified chicken core length DNA for the core particles. Theefficiency of reconstitution was monitored by a gel mobility shift assay. Control and reconstituted samples were run on a 0.7% agarose gel at 100 I? The gels were analyzed by autoradiography. Greater than 95% of the radiolabeled fragments becamecomplexed with histones (Fig. 2). Bleomycin Deatment of Reconstituted Fragments-Control or reconstituted 5 S DNA fragments (at a final DNA concentration of 50 pg/ml) were reacted with 0, 1,or 3 p~ Fe"'.bleomycin at 37 "C for 1 h in the presence of 25 mM 2-mercaptoethanol in a total volume of 150 1.11. The reactionwasterminated by addition of EDTA to 100 m~ and the samples were cooled on ice. The saltconcentration was adjusted to 0.1 M NaCl and the drug and histones were removed by organic extractions followed by ethanol precipitation of the DNA. The cleaved products were heat-denatured and analyzed on denaturing gels (8%polyacrylamide, 8 M urea). Marker lanes were prepared according to the method of Maxam and Gilbert (Maxam and Gilbert,1980). NCS Treatment of Reconstituted Fragments-The nonprotein chromophore of NCS was added at final concentrationsof 0, 3, or 6 PMto a solution containing the reconstituted chromatin and 1 mM glutathione t o activate the drug, in a total volume of 150 pl. The reaction was incubated for 15 min at 22 "C in the dark. Since the reaction is essentially complete within a few minutes, the histones were removed by organic extractions without prior inactivation of the drug followed by ethanolprecipitation.The cleaved DNA products were analyzed on standard denaturing gels as describedabove. Melphalan Deatment of Reconstituted Fragments-Melphalan was diluted in TE and added to the reconstituted chromatin or DNA a t final . incubation a t 37 "C for drug concentrationsof 100 or 150 p ~ Following 1 h, the NaCl concentration was adjusted to 0.8 M in order to stop the reaction and help dissociate DNA and histones. (In control experiments where melphalan was added just after addition of 0.8 M NaC1, no DNA damage was detected.) After removal of the histonesby organic extractions, the DNA was ethanol-precipitated, dissolved in 0.1 ml of 50 mM Tris-HC1, 1 mM EDTA, heated at 70 "C for 1 h to depurinate thermolabile adducts, again precipitated, and heated at 90 "C for 30 min in the presence of 1 M piperidine to cleave the resulting abasic sites. The samples were then lyophilized,dissolved in formamide, and submitted to denaturing gel electrophoresis as above. RESULTS

Chromatin Reconstitution-Salt exchange reconstitution of a X . laevis somatic-type 5 S 430-bp DNA fragment containing the rRNA gene has beenshown to consistently form two precisely MATERIALSANDMETHODS Drugs"Fe"'~b1eomycin was prepared from clinical blenoxane as de- positioned nucleosomes, one extending from -95 to 50 and con1)and one extending from scribed previously (Povirk and Houlgrave, 1988)and storedat -20 "C at taining thestart site of the gene (Fig. a concentration of 6 mM. NCS chromophore was preparedby methanol about position 160 to the end of the fragment (Smith and extraction of clinical NCS (Povirk etal., 1981) and storedat -70 "C a t MacLeod, 1993). We have used the upstream (-95 to 50) in vitro a concentration of 220 PM.Prior t o each experiment, NCS was diluted on positioned nucleosome to analyze the sequence-specific interice in 20% methanol, 20 mM sodium citrate, pH4, in the dark.Melphalan was dissolved at a concentration of 10 mM in 0.1 M HCl and stored action of three chemotherapeutic drugs, bleomycin, neocarzinostatin, and melphalan, with intranucleosomalDNA. at -20 "C. The plasmid pGXlsl4 was digested with AuaI and HindIII Preparation of DNA Fragments-The plasmid pGXlsl4 was a gift from M. C. MacLeod (Smith andMacLeod, 1993) and is numbered with restriction endonucleases. The resulting 430-bp fragment was respect to the natural transcription start point of the 5 S rRNA gene as end-labeled at the AvaI site by Klenow fill in and reconstituted indicated in Fig. 1. pGXlsl4 was digested with AuaI and HindIII reinto nucleosomes as described under "Materials andMethods." striction endonucleases, and the resulting fragment was separated by As a control,DNA was purified from chickenerythrocyte monoagarose gel electrophoresis, electroeluted, and repurified by organic nucleosomes and substituted for core particles in the reconextractions and ethanol precipitation. DNAfragments were end-labeled stitution protocol. As shown in Fig. 2, the control ( C ) and reby Klenow fill-in at theAuaI site with[a-32PldCTP. Reconstitution-H1-stripped nucleosomal core particleswere iso- constituted ( R ) samples were analyzed by agarose gel lated from chicken erythrocyte nuclei as described previously (Laemmli, electrophoresis under nondenaturing conditions. Under these 1970; Smith andMacLeod, 1993). Core length DNA, which was used in conditions, DNA associated with the core histones exhibits a control experiments, was purified from core particles by proteinase K slower mobility than free DNA. Consistently, 95-100% of the digestion and organic extractions (Smith andMacLeod, 1993). end-labeled DNA in the reconstituted samples was shifted with Nucleosomes were reconstituted on 32P-end-labeled5 S DNA by the respect to the control samples. exchange of histones from chickenerythrocyte mononucleosomes

Nucleosome Damage by Bleomycin, Neocarzinostatin, Melphalan and

30589

DNA (lane C ) agrees well with the pattern ofDNA cleavage previously observed in other restriction fragments(Dedon and Goldberg, 1992). The pattern above nucleotide 11 in the lane corresponding t o reconstituted chromatin(lane R is quitesim0 0 -50 0 0 0 ilar to the pattern seenfor uncomplexed DNA. However, below A-QCCCTQC ATQGQQAQQA QCTQQQCCCC M O M Q Q C A Q CACAAQAQQA this region, bands corresponding to prominent bleomycin-inTCTCQQQACQ TACCCCTCCT CQACCCQQQQ TTCTTCCQCT QTQTTCTCCT duced cleavage sites in nakedDNA were either absent entirely or much less intense in the reconstituted samples. This region 0 0 1 0 0 0 QQAFAAQTCA QCCTTQTQTT CQCCTACQQC CACACCACCC TQAAAQTQCC is coincident with the placement of a positioned nucleosome CCTTTTCAQT ClXWACACAAQCQQATQCCQ QTQTQQTQQQ ACTTTCACQQ over the start site of the gene as described by Smith and MacLeod (1993). This distinctive difference in the banding pat0 0 50 0 tern between the control and reconstituted samples suggests CQATCTCQTC TQATCTCQQA AGCCAAGCAG GGTCGGGCCT GGTTAGTACT that thenucleosomal DNA was protected to a large extentfrom QCTAQAQCAQ ACTAQAQCCT TCGGTTCGTC CCAGCCCGGA CCAATCATGA attack by bleomycin. At the low bleomycin dose, there were some sitesnear positions 2 0 4 0 for which thebands were 100 darker in the reconstituted sample lane;however, the autoraTGGATGGGAG ACCGCCTGGG AATACCAGGT GTCGTAGGCT TTTGCACTTT diogram showed that there was also substantially more fullACCTACCCTC TGGCGGACCC TTATGGTCCA CAGCATCCGA AAACGTGAAA length DNA in that lane. In order to evaluate cleavage quanFIG.1. Partial sequence of linear 5 S DNA fragment. pGXlsl4 titatively andcorrect for such differences in sampleloading, the was digested withALWIand Hind111 restriction endonucleases and endlabeled with %'P a t t h eAuaI site. A partial sequence of this fragment, gel was analyzed witha Molecular Dynamics PhosphorImager. including the positioned nucleosome used in this study is shown. The The integrated intensity of each observable band was calcuasterisk denotes the position of the label, 0 represents positions of lated and divided by the total integrated intensity of the apmaximum hydroxyl radical cleavage (Smith and MacLeod, 1993).and fraction of total hold lettering represents the position of the nucleosome on the se- propriate lane. This number represents the molecules that were cut at that nucleotide. The cleavage frequence. Numberingis based on the start site of transcription (+1)of the 5 S RNA gene. The nucleosome extends from 95 to 50 in the sequence. quency for each sequence position in the reconstituted sample was then normalized to the cleavage frequency a t the same position in the unreconstituted sample (Drew and Calladine, 1987), and finally, this ratio was normalized to the average ratio in the linkerregion, based on the assumption thatcleavage susceptibility in the linker would be unaffected by reconstitution. A plot of this normalized ratio ( W C )as a function of nucleotide position (Fig. 4A) reveals a dramatic suppression of cleavage within the centralregion of the nucleosome. Interestingly, a portion of the fragment approximately 40 bp from the end of the nucleosome (nucleotide positions 11-50) seemed to be largely unprotected from attack by bleomycin. Although the other end of the nucleosome (-95 to -56) contains a stretch of As and Ts which tend to isolate thepredicted sites of cleavage, it is neverthelessclear that this region is moreprotected against cleavage than the region from 11 to 50, even though these two regions are approximately equidistant from the dyad axis of the nucleosome (see Fig. 5A 1. At several sites just inside and just outside the nucleosome boundary (positions 27-56), cleavage in the reconstituted sample actually appeared to be enhanced by 1.5-2-fold, although the unusually large experimental variations inWC for this region cast some doubt on the significance of this effect. In addition to single strand breaks, bleomycin is known to FIG.2. Electrophoretic mobility shift assay. The efficiency of re- induce doublestrand breaks, as well a s abasic sites with closely constitution was monitored by electrophoresis under nondenaturing opposed breaks (Povirk and Houlgrave, 1988). As noted above conditions. Complex formation resulted in a decreased electrophoretic (Introduction),single and double strand breakageby bleomycin mobility compared with the control fragment (free DNA). C = control sample subjected to mock reconstitution in the absenceof histones; R = can be determined separately by examining cleavage a t prireconstituted sample. Greater than 954. of the end-labeled fragment mary and secondary sites, respectively, and these sites can be becomes associated with core histones upon reconstitution. predicted on the basisof DNA sequence (Povirket al., 1989). To determine theinhibition of breakage at primary andsecondary Bleomycin Cleavage of Intranucleosomal DNA-In vitro re- cleavage sites in the nucleosome, we compared the integrated constituted 5 S DNA was treated with 0, 1,or 3 p~ bleomycin, intensities of each for the region of maximum suppression of and the cleavage products analyzed on standard denaturing cleavage (positions -95 to lo), the end of the nucleosome (11to polyacrylamide gels (Fig.3A ). There wasvery little cleavage in 50), and the linker region (51 to 111).The degree of suppression untreated DNA, thedarkmisshapenbandsnear positions of each primary andsecondary site isshown in Fig. 4 A , and the 60-80 in thecontrol lane apparently represent various forms of average suppression for each region is shown in Table I. There renatured or partially renatured full-length fragment, since is approximately a 5-fold decrease in cutting of the central they were much more prominent when the heat denaturation region of the nucleosome compared withthe linkerregion or the step was eliminated, and they were absent entirely in extenend of the nucleosome (11to 50). However, in these data, there sively cleaved DNA samples. The banding pattern resulting appears to be no difference between the suppression of cutting from sequence-specific bleomycin cleavage of uncomplexed a t primary versus secondary sites, implying that single and -100 0 CC GGGGATCCGG GCCCCGGCTG TCCAGQCTQQ QQTTTTTTAT CTTQQQQAAA 'CCTAGGCC CGGGGCCGAC AGGTCCQACC CCAAFAAATA QMCCCCTTT

30590

Damage Nucleosome

by Bleomycin, Neocarzinostatin, and Melphalan 1 0 A T " " 3 G CC CG RC C- RR CC RR C R

A.

B.

T

A

0

1Q(Lm

8C 60

40 20 1

FIG. 3. High resolution mapping of drug-induced cleavage of intranucleosomal DNA. A linearend-labeled fragment (Fig. 1) was reconstituted into nucleosomes. Controlandreconstituted samples were treated with 0, 1, or 3 PM bleomycin (A) or 0, 100, or 150 PM melphalan ( B ) . TC/AG are Maxam-Gilbert marker lanes; C represents control samples (subjected to same conditions as reconstituted but in the absence of core histones), and R represents samples reconstituted into nucleosomes. The numbering corresponds tothat shown inFig. 1 (relative to the start site of transcription).

80

-20

60

-40

40

20 -60

1

-80

double strand cleavage were inhibited to equal extents within the protected regions. To further analyze the patternof cleavage within thenucleosomal region, the phosphorimage intensity of the control and reconstituted samplesof the entire lane was plotted versus the relative position in the gel (Fig. 5). This intensity is proportional to the level of radioactivity in a particular band and corresponds to thefrequency of cutting at thatsite. Since bleomycin is thought tobind in theminor groove, a 10-bp periodicity in the cleavage frequencies (similar to that seen with hydroxyl radical) might have been expected in the reconstituted samples, with cleavage being preferentially inhibited a t se-

quence positions where theminor groove faces the nucleosome core. However, even in the central region, the pattern of residual cleavage in the reconstituted sample (dotted line) mimics that of the control lane (heavy line), suggesting that the reduction in cleavage seen in the reconstituted samples is not due to mere accessibility of the minor groove of DNA in the nucleosome. Thus, it appears that the sequence specificity of bleomycin cleavage supersedes minor groove accessibility and that, remarkably, cleavage sites are suppressed to approximately equal extents whether the minor groove is facing outward or inwardwith respect to the nucleosome at any particular site.

30591

Nucleosome Damage by Bleomycin, Neocarzinostatin, and Melphalan

1

nucleosome

0

FIG.4. Quantitative comparison of drug-induced cleavage. The integrated phosphorimage intensity of each observable band was divided by the total intensity in the appropriate lane to give the fraction of the total represented by each band (the fractionalintensity).At eachposition, the ratio of the fractional intensity for the reconstituted sample (R)to that for the control sample (C)is plotted uersus nucleotideposition. A, bleomycintreated nucleosomes. The open bars represent secondary cleavage sites, and the solid bars representprimary cleavage sites. B , NCS-treated nucleosomes. C,

melphalan-treated nucleosomes. Open and solid bars represent adenineand guanine residues, respectively.Each data point represents the average WC ratio from four experiments, normalized tothe average WC for the linker region. The bar above each graph represents the nucleosome position, and the 0 represents positions of maximum hydroxyl radical cleavage.

8.

25

1

0

I

nucleosome

0

C

I

nucleosome

I

1

SequencePosition

of NCS Cleavage of Intranucleosomal DNA-In vitro reconsti- duced-cleavage in the reconstituted samples paralleled that the control samples (Fig. 5B). Consistent with the results obtuted nucleosomal core particles or control sampleswere treated with 0, 3, or 6 p~ NCS in the presence of glutathione tained with bleomycin, and in contrast to those obtained with and analyzedon a standard denaturinggel as described above. hydroxyl radicals (Smith and MacLeod, 19931, there was no Suppression of cleavage again was apparent in theregion co- indication of a 10-bp periodicity in the protection of nucleosoincident within the central region of the nucleosome (positions mal DNAfrom NCS. It ispossible that a weak periodicity might have been obscured by the presence of both 5”phosphate and -96 to 10, data not shown). 5”aldehyde-terminated fragments and by the fact that NCSTo analyze this phenomenon quantitatively, the gels were scanned as described above, using the PhosphorImager. The induced breaks tend t o be clustered in the relativelym - r i c h pattern of NCS-induced cleavage is complicated by that fact tracts where the minorgroove faces the nucleosome. It is nevof the nucleosome that thenucleoside 5”aldehyde moieties present at the termini ertheless apparent that near the center groove faces of most but not all NCS-induced breaks retard themobility of cleavage is inhibitedeven at sites where the minor the resulting fragments by the equivalent of two nucleotide outward andthat in all regions cleavage continues tobe domipositions with respectt o the 5”phosphate-terminated markers nated by the inherent sequence specificity of the drug rather (Kappen and Goldberg, 1983). However, since it is not possible than by the rotational positioning of the minor groove on the to predict with certainty which bands correspond to 5“phos- nucleosome. phate andwhich to 5‘-aldehyde termini, the data are presented Melphalan Adduction in Intranucleosomal DNA-The same according t o apparent bandposition, without correction for the end-labeled reconstituted and unreconstitutedDNA fragments aldehyde termini (Fig. 4B). Despite this complication, it is ap- were treated with the alkylating agent melphalan. Following parent that cleavage was suppressed in the central region (55 heat depurination of thermolabile adducts andcleavage of the to 10) and in the upstream periphery (-95 to -56) of the nu- resulting abasic sites in alkali, strand breaks were analyzed as cleosome. As with bleomycin, the region of the nucleosome from above. Like most other damaging agents, melphalanproduced 11 to 50 exhibited no protection from cleavage by NCS. The less damage in the region of DNA bound to the core particle calculated average WC values were 0.429 2 0.23 for the region than in the linker region. As shown in Fig. 4C and Table 11, of greatest suppressionof cleavage (-81 t o lo), and1.140 f 0.14 adenine alkylation (presumably at N-3, a minor groove site for the region of no suppression of cleavage (11to 50). Thus, (Osborne and Lawley, 1993)) was suppressed by about 40%, there was approximately a 2.4-fold decrease in the frequency of with relatively little variation between individual adenines in cleavage in the central region and the upstream periphery of the sequence. There was less suppression of guanine N-7 alkythe nucleosome, as compared with the linker. lation, with a n average decrease of about 16% and more variTo examine the intranucleosomal region of DNA, the entire ation between individual nucleotide positions; in fact, certain lane was scanned and the relative intensity plotted as a func- guanine residues were more susceptible to cleavage in nucleotion of position in thegel as described in Fig. 5. Within each of somal thanin naked DNA. In contrast to results obtained with the regions described above, the pattern of residual NCS in- bleomycin and NCS, there was little difference between the

Nucleosome Damage by Bleomycin, Neocarzinostatin, and Melphalan

30592 A

4000

1

0

77

-30

200

-63

800

1000

5

-27

-52

600

800

-96

1200

1400

1600

1800

2000

1800

2000

B. 55

-71

-82

-98

1000

B .ti

600 400

200 0

0

200

direction ofelectrophoresis

400

t

loo0

1200

1400

1600

relative position

FIG.5. Cleavage pattern in free uersus nucleosomal regions of DNA. Following electrophoresis on a sequencing gel, the radioactivity profiles for control(solid line)and reconstituted (dotted line)samples from oneexperiment were quantitated using software supplied by Molecular Dynamics (ImageQuant), which constructs line graphs equivalent to densitometer tracings. A , bleomycin-treated samples (3p ~ ) B. , NCS-treated samples (6PM).Electrophoresis is from left to right (as denoted by the arrow), and positions of selected bands are noted.

TABLE I

known of the influence of chromatin structure on the nature, extent, andgenomic localization of drug-induced damage. However, the finding that active genes more are vulnerable tobleoNucleotide position Primary sites Secondary sites mycin-induced damage than nontranscribed DNA (Kuo, 1981) suggests that chromatin structure affords at least some protec-c 0.11 0.235& 0.150.187 -95 to 10 1.360& 0.281.531 f 0.67 11 to 50 tion from DNA damage by this drug. In addition, numerous 0.985 f 0.520.897 f 0.15 51 to 111 lines of evidence, including thecleavage of chromatin into oligonucleosome size fragments by both bleomycin and NCS TABLEI1 (Bennett et al., 1993; Kuo and Hsu, 19781, and the decreased Average ratios (RIC) for adenines and guanines in melphalan-treated susceptibility of DNA in cells where the nucleosomal repeat nucleosomal DNA length has been reduced (Lonn et al.,19891, indicate that these Nucleotide position As Gs drugs preferentially target internucleosomal linker DNA, implying that DNA withinthe nucleosome core is relatively 0.887 -c 0.060 -95 to 10 0.602 2 0.028 0.609f 0.044 0.758 & 0.050 11 to 50 protected. 51 to 81 0.879 f 0.075 1.160 -c 0.115 Nucleosome structure has been studiedextensively and has been shown toplay an importantrole in gene regulation, trancentral region of the nucleosome and the region near the pe- scription, andDNA repair (Hayes andWolffe, 1992a). In many riphery of nucleosomal DNA (11to 50) in the degreeof protec- cases, the rotational and translational positioning of nucleosomes on DNA is sequence-dependent. In general, rotational tion against melphalan-induced alkylation. positioning is determined by preferential placement of short DISCUSSION runs of (A, T)with the minorgroove facing toward the octamer of (G, C) with the minor groove facing away from The cytotoxicity of chemotherapeutic drugs is thought be toa and short runs consequence of DNA damage, eithercovalent DNAmodification the octamer (Drew and Travers, 1985).Translational positionor direct strand breaks. Although the interactions of these ing dependsat least insome cases on specific sequences around drugs withisolated DNA have been studiedextensively, less is the dyad(Ramsay, 1986). The 5 S rRNA gene from a variety of Average ratios (RIG') for primary and secondary cleavage sites in bleomycin-treated nucleosomal DNA

Nucleosome Damage by Bleomycin, Neocurzinostatin, and Melphalan

30593

species has beenshown to precisely position a nucleosome McMurray et al., 1991; Nieto and Palacian, 1988). It may be around thestart site of the gene bothin vivo and in vitro (Drew significant that theregion which exhibits the lack of protection andCalladine, 1987; Rhodes, 1986;Simpson and Stafford, from bleomycin- and NCS-induced cleavage within the nucleo1983; Smith and MacLeod, 1993). We have used a fragment some is the end closest to the TFIIIA binding site (positions containing thisgene in which the rotational and translational 45-91), the internalcontrol region of the gene. Previous studies position of DNA on a nucleosome has been determined previ- have shown that bindingof the octamer over this region inhibously (Smith and MacLeod, 1993) to examine the binding and its transcription in vitro, but that when a tetramer is bound cleavage patterns of bleomycin, neocarzinostatin, and melpha- (without dimers), transcription proceeds (Hayes and Wolffe, lan, at single nucleotide resolution. 1992b). Furthermore, transcription also proceeds when TFIIIA When control and reconstituted samples were treated with is added prior t o nucleosome formation, suggesting that the equalconcentrations of bleomycin, a 5-fold suppression of presence of the dimer impedes binding of the transcription cleavage was observed in the centralregion of the nucleosome, factor. Thus, it is possible that the asymmetry in cleavage at which extends from -56 to +lo, according to the translational the ends of the nucleosome is due toa weaker binding of DNA and rotational positioning determined by Smith and MacLeod to theH2A/H2B dimer in thisregion (11to 501, compared with (1993). Interestingly,one end of the nucleosome exhibited little its binding t o the other dimer on the opposite end of the nuor no suppression of cleavage (11to 501, whereas the opposite cleosome (-95 to -56). This weaker binding may be necessary to end (-95 to -56) showed suppression comparable with thecen- prevent displacement of TFIIIA by the nucleosome. tral region. Similarly, NCS-induced cleavage showed suppresIt is remarkable that of the several agentsfor which damage sion within the central region of the nucleosome (approxi- to nucleosomal DNA has beenexamined(aflatoxin B1, dimately 2.4-fold compared with the linkerregion) and the same methyl sulfate, benzo[alpyrene diol epoxide, bleomycin, melasymmetric pattern of cleavage on the ends. For the central phalan, and NCS), only one, hydroxyl free radical, induces region, the degree of suppression of NCS-induced cleavage is damage with a 10-bp periodicity corresponding to the rotasimilar to the suppressionobserved previously for adduct fortional positioning of DNAon the nucleosome. Although damage mation by aflatoxin B1 (Moyer et al., 1989) and benzo[alpyrene by NCS, bleomycin, and benzo[alpyrene diol epoxide is supdiol epoxide (Smith andMacLeod, 19931, whereas the suppres- pressed within the central region of nucleosomal DNA, the sion of bleomycin-induced cleavage is substantially greater. sequence specificity of the residual damage closely mimics that However, even within the centralregion, the residualcleavage seen with nakedDNA. Since all threeof these agents are relaof nucleosomal DNA byboth bleomycin and NCS preserved the tively small (much smaller thana helical turn of DNA) and all sequence specificity seen in naked DNA, a similar result was attack DNA from the minor groove, it is quite surprising that obtained with benzo[a]pyrene diol epoxide (Smith and damage is not preferentially suppressed at sites where the MacLeod, 1993). minor groove faces the histonecore and equally surprising that The lack of protection against bleomycin- and NCS-induced damage is suppressed at all at sites where the minor groove cleavage seen at one end of the nucleosomal DNA is consistent faces outward. with the results of Bennett et al. (1993) who found that alTo explain the finding that the extentof damage is not dicthough treatment of reconstituted chromatin with either of tated simply by the accessibility of the minor groove, several these drugsproduced a typical nucleosomal ladder, in each case mechanisms may be considered. First, structural distortions the bands were less well defined than those generated by miinduced by bending DNA around nucleosomes may disrupt the crococcal nuclease digestion. Drug-induced cleavage just inside normal interactions of these agents with the minor groove. the boundaries of nucleosome-bound DNA (e.g. positions 11 to However, if this were the case, substantial alterations in the 50 in the5 S sequence) could account for the smearingof these sequence specificity of the residualcleavage would be expected, bands andcould also explainthe NCS-induced cleavage of chroand this is clearly not observed. Second, interaction of these matin into fragments of -125 bp, slightly shorter than the agents with DNA may involve a n initial nonspecific binding in length of DNA bound by the nucleosome core (McHugh et al., the minor groove, followed by sliding along the groove until a 1982; van Holde, 1989). This region of unprotected DNAwithin high-affinity binding site isfound. Since most of the agents are the nucleosome is also roughly coincident with a transitional region described by Smith andMacLeod (1993) (positions 2%44), positively charged, the partial neutralization ofDNA phosin which benzo[a]pyrene diolepoxide binding was less suppressed phate charge by histone amines could reduce the initial nonthan in the central region of the nucleosome. A similar transi- specific binding step withoutaffecting the "sliding" step, which tional region was reported for aflatoxin B1 binding (Moyer et al., determines sequence specificity. However, this proposal would 19891, but in neither case was there such a distinct asymmetry in not explain the similar results obtained with benzo[a]pyrene diol epoxide, which is uncharged. A third possibility is that protection as that seen with bleomycin and NCS. The apparent susceptibility of the unprotected region (11to binding of the drugs to DNA prior to cleavage may involve 50) to damage can be correlated to the weaker interactions of induced local changes in DNA conformation and that steric DNA with theH2A-H2B dimers at the ends of the nucleosomes. constraints imposed by nucleosome structure may prevent The intact nucleosome is a tripartite structure in which the these conformational changes. In the case of NCS, the critical periphery of nucleosomal DNA (approximately 20-25 bp on conformational change could be intercalation of the naphthoate each end) bound is to theH2A-H2B dimers, whereas the centralmoiety of the chromophore, which appears tobe involved in the DNA (Galat andGoldberg, 1990; Povirk 100 bp is bound, much more tightly, t o a central (H3-H4), tet- binding of this agent to ramer (Thoma, 1992). Transcription in the presence of bound and Goldberg, 1981). By analogy to the ethidiumbromide studtetramers is inhibitedby addition of dimers, and it is believed ies mentioned above, it would be expected that any reaction that the dimers are displaced duringtranscription in vivo involving intercalative bindingwould be severely suppressed in (Hayes and Wolffe, 1992a). The loosely bound dimers can be nucleosomal DNA and that any suchbinding would be largely preferentially released from core particles by treatment with restricted to regions near the ends of nucleosomal DNA. The dimethylmaleic anhydride or high concentrations of ethidium apparently weaker bindingof DNA to histones at one end the is released more nucleosome (positions 11-50), as discussed above, may make bromide, and in each case the first dimer readily than the second (McMurray andvan Holde, 1991; this region more susceptible to reactions involving intercala-

30594

Nucleosome Damage

by Bleomycin, Neocarzinostatin, and Melphalan

tion or other conformational changes. teractions with other molecules than does accessibility. MoreFor bleomycin, no such simple model can be proposed, since over, the degree of conformational constraints appears to be its mode of binding to DNA is much less certain. Although early sequence-dependent, at least at the periphery of nucleosomeevidence suggested that intercalation of the bithiazole rings bound DNA. was involved in binding(Povirk et al., 1979), more recent modAcknowledgment-We thank M. C. MacLeod for providing the els have favored binding of the bithiazole in the minorgroove, following the curvature of the groove in a manner similar to plasmid pGXlsl4. distamycin. Modeling studies indicatedthat this mode of bindREFERENCES ing could be accommodated with very little perturbation of DNA helical structure (Kuwahara and Sugiura, 1988). How- Austin, M. J. F., Han, Y.-H., and Povirk, L. F. (1991)Proc. Am. Assoc. Cancer Res. 32,101 ever, it is difficult to reconcile this proposal with the severe Bauer, G. B., Wang, P., Kellogg, G . E.,Abraham, D. J., and Povirk, L. F. (1993)Proc. suppression of bleomycin-induced DNA damage which is seen Am. Assoc. Cancer Res. 34, 136 a t every potential cleavage site in the central region of the Bennett, R. A. O., and Povirk, L. F. (1992)Proc. Am. Assoc. Cancer Res. 33, 175 R.A. O., Swerdlow, P. S., and Povirk, L. F. (1993)Biochemistry 32, nucleosome, especially since this suppression is greater that Bennett, 3188-3195 reported for any other low molecular weight DNA-damaging Dedon, P. C., and Goldberg, I. H. (1990)J. Biol. Chem. 266, 14713-14716 that bleomycin-induced DNA Dedon, P. C., and Goldberg, I. H. (1992)Chem. Res. Toxicol. 6,311-332 agent.Theseresultssuggest H. R., and Calladine, C. R. (1987)J. Mol. Biol. 196, 143-173 damage may in fact require intercalation or some type of in- Drew, Drew, H. R., and Travers,A. A. (1985)J. Mol. Biol. 188, 773-790 duced conformational change in DNA prior to hydrogen ab- Fox, K. R., and Grigg, G. W. (1988)Nucleic Acids Res. 16,2063-2075 straction. The finding that bleomycin binding renders certain Galat, A,, and Goldberg, I. H. (1990)Nucleic Acids Res. 18, 2093-2099 Hayes, J. J., and Wolffe, A. P. (1992a)Biwssays 14, 597-603 adjacent DNA bases more susceptible tomodification by dieth- Hayes, J. J., and Wolffe,A. P. (1992b)Proc. Natl. Acad. Sci.U. S. A. 89, 1229-1233 ylpyrocarbonate (Fox and Grigg, 1988) also supports this pro- Hayes, J., Tullius, D., and Wolffe, A. P. (1990)Proc. Natl. Acad. Sci. U. S. A . 87, 7405-7409 posal. The finding that double and single strand cleavage are Henne, T., and Scbmahl, D. (1985)Cancer P e a t . Reu. 12, 77-94 inhibited to equal extents (Table I) confirms results obtainedby Kappen, L. S., and Goldberg, I. H. (1983)Biochemistry 22,4872-4878 Kuo, M. T. (1981)Cancer Res. 41, 2439-2443 directly quantitating single and double strandbreaksin M. T., and Hsu, T. C. (1978)Chromosoma ( B e d . )68,229-240 minichromosomes reconstituted from core histones and super- Kuo, Kuwahara, J.,and Sugiura,Y. (1988)Proc. Natl. Acad. Sci.U. S. A. 86,245%2463 coiled plasmids (Bennettet al., 1993) and is consistent with the Laemmli, U. K. (1970)Nature 227,680-685 view that the two breaks involved in double strand cleavage Lonn, U., Grin, S., Nylen, U., and Windblad, G. (1990)Biochem. Pharmacol. 39, 101-107 occur during a single bleomycin binding event (Povirk et al., Maxam, A. M., and Gilbert, W. (1980)Methods Euymol. 66,499-560 1989; Steighner and Povirk, 1990). McHugh, M. M., Woynarowski, J., and Beerman,T. (1982)Biochim. Biophys. Acta 696,7-14 In the case of melphalan, major induced conformational McMurray, C. T., and van Holde, K. E. (1991)Biochemistry 30,5631-5643 changes in DNA seem unlikely, but nucleosome structure may McMurray, C. T., Small, E. W., and van Holde, K. E. (1991)Biochemistry 30, also reduce random thermal motions of the helix which may be 5644-5652 required t o accommodate melphalan. Modeling studies of Moyer, R., Marien, K., van Holde, K., and Bailey, G. (1989)J . Biol. Chem. 264, 12226-12231 Bauer et al. (1993) suggest that melphalan binding to adenineNieto, M. A,, and Palacian, E. (1988)Biochemistry 27, 5635-5640 N-3 requires a slight wideningof the minorgroove. In thecase Osborne, M. R., and Lawley, P. D. (1993)Chem. Biol. Interact. 89,4940 L. F., and Goldberg, I. H. (1981)Biochemistry 20,40074013 of guanine N-7 alkylation, the degreeof protection afforded by Povirk, Povirk, L. F., and Houlgrave, C. W. (1988)Biochemistry 27,385s3857 nucleosome structure appears to reflect the bulkiness of the Povirk, L. F., Hogan, M., and Dattagupta, N. (1979)Biochemistry 18,96110 alkylator (aflatoxin B1 > melphalan > dimethyl sulfate). Al- Povirk, L. F., Dattampta, N., Wad, B. C., and Goldberg,I. H. (1981)Biochemistry 20,40074014 though adenine alkylation by melphalan is suppressed more Povirk, L. F., Houlgrave, C. W., and Han, Y.-H. (1988)J. B i d . Chem. 263,1926319266 than guanine alkylation, this difference alone is not nearly L. F., Han, Y.-H., and Steighner, R. J. (1989)Biochemistry 28,8508-8514 sufficient to explain the difference between the endogenous Povirk, Pratt, W. B., and Ruddon, R. W. (1994)The Anticancer Drugs, Oxford University aprt gene and the pZ189 shuttle vector in the proportion of Press, New York melphalan-induced mutations which occur a t A.T base pairs; Ramsay, N. (1986)J . Mol. Biol. 189, 178-188 Rhodei,'D. (1986)EMBO J. 4,3473-3482 thus, other factors such as differential DNA repair must be Simpson, R. T.,and Stafford, D. W. (1983)Proc. Natl. Acad. Sei.U. S. A . 80.51-55 Smit.h. 1... and MacLeod. M. C. (1993)J. B i d . Chem. 268. 20620-20629 .~~~ "", R . ", involved. R. J., and Povirk, L. F. (199OjProc. Natl. Acad. Sci. U. S. A . 87, 835s In conclusion, the view of nucleosome structure which is Steighner, 8354 emerging from studies of DNA-damaging agents is one in which Thoma, F. (1992)Biochim. Biophys. Acta 1130, 1-19 both themajor and minorgrooves are remarkablyaccessible to van Holde, K. E. (1989)Chromatin, Springer-Verlag, New York Wang, P., Bennett, R. A. O., and Povirk, L. F. (1990)Cancer Res. 60,7527-7531 small molecules. Resistance of nucleosomal DNA to conforma- W a g , P., Bauer, G. B., Bennett, R. A. 0..and Povirk, L. (1991) F. Biochemistry 30, 11515-11521 tional change seems to be a more important constraint on in"

"

~~~

~~~~