Determination of the structural features of (+)-CC-1065 that are

Journal of the American Chemical Society 2011 133 (44), 17641-17651 ... Ecteinascidin 743: A Minor Groove Alkylator That Bends DNA toward the Major ...
4 downloads 0 Views 4MB Size
Download PDF
Chem. Res. Tonicol. 1991,4, 203-213

203

Determination of the Structural Features of (+)-CC-1065 That Are Responsible for Bending and Winding of DNA Chong-Soon Lee, Daekyu Sun, Ryoichi Kizu,t and Laurence

H. Hurley*

Drug Dynamics Institute, Division of Medicinal Chemistry, College of Pharmacy, The University of Texas at Austin, Austin, Texas 78712 Received December 26, 1990

Analysis of the anomalous migration in electrophoretic mobilities of (+)-CC-lO65-modified oligomers following ligation reveals that (+)-CC-1065 induces DNA bending and winding of the helix. (+)-CC-1065 is a potent antitumor antibiotic produced by Streptomyces zelensis. This drug selectively bonds covalently to N3 of adenine and lies in the minor groove of DNA, reacting in a highly sequence-selective manner. Structurally, (+)-CC-1065 consists of three subunits: two identical pyrroloindole units (subunits B and C) and a third subunit containing the DNA-reactive cyclopropane ring (subunit A). While the bonding reaction is the main determinant of DNA sequence selectivity of (+)-CC-1065, binding interactions between the inside edge substituents of the B and C subunits and the floor of the minor groove of DNA can modulate or fine tune this sequence selectivity [Hurley, L. H., Lee, C.-S., McGovren, J. P., Mitchell, M. A., Warpehoski, M. A., Kelly, R. C., & Aristoff, P. A. (1988) Biochemistry 27,3886-38921. The A subunit of (+)-CC-1065is responsible for the bending of DNA, and close van der Waals contacts between the inside edge of (+)-CC-1065 and the floor of the minor groove of DNA cause winding equivalent to about 1 base pair per alkylation site and stiffening of DNA. The magnitude of DNA bending induced by (+)-CC-1065 and related compounds is about 14-19’, which is equivalent to that produced by an adenine-thymine tract of about 5-6 base pairs in length. Experiments using oligomers containing both an adenine tract and a unique (+)-CC-1065 bonding site approximately one helix turn apart demonstrate that the directionality of drug-induced bending is in toward the minor groove and the locus of bending is about 2-3 base pairs to the 5‘-side of the covalently modified adenine. A circularization efficiency assay shows that the optimum size of circles produced by (+)-CC-1065 and related drugs is between 168 and 180 base pairs. These results are discussed in relation to the molecular basis of the DNA sequence selectivity of (+)-CC-1065, and the (+)-CC-1065-induced DNA bending is compared with the intrinsic bending associated with adenine tracts. Since (+)-CC-1065 induces effects on local DNA structure that appear similar to those produced naturally by adenine tracts and certain DNA binding proteins, the relevance of this phenomenon to biological effects of (+)-CC-1065 and related drugs is considered.

I ntroductlon It has been demonstrated that bending occurs naturally in DNA fragments containing short runs of adenine (Atracts)’ (1-3), and it can be induced by site-specific DNA binding protein such as CAP (4,5), RNA polymerase (6), and Escherichia coli integration host factor (7-9). Several models have been proposed to explain the phenomenon of DNA curvature at the molecular level (1,10-13)) but the junction-bend model (I,14) appears to have the most support. The anomalies in electrophoretic mobilities of bent DNA on nondenaturing polyacrylamide gels have been used to investigate bending induced by intrastrand platinum cross-linking (15), interstrand psoralen crosslinking (16), and UV-induced thymine dimer (17) in duplex DNA. The major groove direction of the platinum-induced bends was determined by copolymerizing a duplex containing the platinum adducts with a series of synthetic decamers containing an As tracts (15). CC-1065 is an extremely potent antitumor antibiotic produced by Streptomyces zelensis (18-20). It is active ~

*Address correspondence to this author. ‘On leave from Faculty of Pharmaceutical Sciences, Kanazawa University, 13-1, Takara-machi, Kanazawa 920, Japan.

against several experimental murine tumors in vivo and is about 100 times more potent than adriamycin against a broad spectrum of tumors in the cloning assay (21). CC-1065 bonds covalently through N3 of adenine (22,23) and lies within the minor groove of DNA, covering a 3 and 1 bp region to the 5’- and 3‘-side, respectively, of the covalently modified adenine (24). Upon thermal treatment of CC-1065-(N3-adenine)-DNA adducts, cleavage of the N-glycosidic linkage and subsequent backbone breakage occur to the 3’-side of the covalently modified adenine to leave a 5’-phosphate on the 3‘-side of the break and, presumably,a modified deoxyribose on the 5’-side (25) (Figure 1). This strand breakage assay has been used to determine the DNA sequence selectivity of CC-1065: two subsets of consensus bonding sites [5’PuNTTA* and 5‘AAAAA*, where (*) indicates the covalently modified adenine and N indicates any of the four bases in DNA] were identified in the first study (25), and in the second study (26) the consensus sequence 5’(A/T)(A/T)A* was found. The Abbreviations: CAP, catabolite activator protein; bp, base pair($; EDTA, ethylenediaminetetraacetic acid; TEMED, NJV,N’JV’-tetramethylethylenediamine; DSC,15 mM NaCl and 1.5 mM sodium citrate, pH 7.4; Tris, tris(hydroxymethy1)aminomethane; A-tracts, adenine tracts; Ab, five adenines; &, six adenines; Pu, purine nucleotide; TBE, Trisborate-EDTA; N, any nucleotide.

0893-228~/91/21O4-O203$O2.5O/O 0 1991 American Chemical Society

Lee et al.

204 Chem. Res. Toxicol., Vol. 4, No. 2,1991

n

Subunit

CC-1065

n

0

CC-1065

A 5”’

AB’C’

SPECIES “B”

+

I

Ntl,

+? SPECIES “A”

‘e.

I OH

I

3’

3’ Figure 1. Reaction of CC-1065 with N3 of adenine in DNA and products from thermal cleavage reaction. Species “A” is the product produced by thermal treatment, and species “B” is the product produced by thermal and subsequent piperidine treatment. The exact nature of the species generated on the 5’-side of the strand break is unknown (25).

construction of a site-directed CC-10654N3-adenine)DNA adduct in a 117-bp fragment of M13mpl DNA (27) allowed determination of the effect of covalent attachment on local DNA structure. DNase I footprinting and restriction enzyme analysis demonstrated that CC-1065 adduct formation produces an apparent strand-selective and unidirectional effect on DNA structure, which extends more than one helix turn to the 5’-side of the covalent bonding site (28). The CC-1065 molecule (Figure 1)consists of three repeating pyrroloindole subunits, one of which (the A subunit) contains a DNA-reactive cyclopropyl function. Naturally occurring CC-1065 and synthetic analogues having the same stereochemical configuration of the cyclopropane ring in the A subunit are designated (+)-enantiomers.2 Analogues containing the alkylating subunit linked to acetyl or one or two indole subunits are designated A, AB, and ABC, respectively. An analogue having the same structure of CC-1065 except that it lacks the hydroxyl and methoxyl substituents on the B and C subunits is termed AB’C’. The structures of naturally occurring CC-1065 and selected synthetic (+)-enantiomeric analogues used in this investigation are shown in Figure 2. Facilitated by the extensive array of analogues synthesized by Upjohn scientists (29-31), structure-activity relationships have been sharply defined (32). Of particular

* Since only (+)-enantiomers [i.e., those having the same stereochemistry of the cyclopropyl ring as the natural product (+)-CC-lOfiB] were used in this study, the designation (+) or (-1 is omitted.

H

O

A

Figure 2. Structures of CC-1065 and its synthetic analogues used in this investigation. Compounds shown all have the same stereochemistry as the naturally occurring (+)-CC-1065.2

significance is our observation that the A subunit of CC-1065 has sufficient structural information to elicit the sequence specificity of ABC, although the precise structural nature of the B and C subunits can modulate or fine tune this specificity (33). In previous publications (19,33) we have argued that the primary basis for the sequence selectivity of CC-1065 is a sequence-dependent catalytic activation and/or a sequence-dependent conformational flexibility. In contrast to this conclusion, we have recently demonstrated that the sequence selectivity of (-)-CC-1065, the synthetic enantiomer, is largely determined by noncovalent binding interactions (26). The purpose of the work described here was to further define the effect of CC-1065 covalent bonding reactivity and binding interactions on local DNA structure and, more specifically,the possible ability to cause or entrap bending of DNA. For these studies, CC-1065 and selected synthetic analogues were %reactedwith a series of oligodeoxynucleotides containing a unique drug bonding site and then ligated into multimers to amplify any structural changes in the double helix ( I ) . The anomalous migration in electrophoretic mobility of ligation products was analyzed by nondenaturing gel electrophoresis to detect structural anomalies such as bends. The results show that CC1065-induced DNA bending occurs as a consequence of

CC-1065-Induced DNA Bending Table I. Sequences of the Synthetic Oligonucleotides Used in This Studya A.

B.

17 bp :

GAGCCATGATTACGGAT GGTACTAAGCCTACTC

19 bp :

GAGACCATGATTACGGAm TGGTACTAATGCCI'AACTC

20 bp :

GAGACCATGATTACOGAT TGGTACTAATGCCTAAGCTC

2 1 bp :

GAGGACCATGATTACGGATTC CEGTACTAATGCCTAAGCTC

23 bp :

GAGTGACXATGATTACGGATTCA ACXGTACTAATGCCTAAGTCTC

21-A-I bp:

AAAAACCIGATI'ACGAGAlTC TTGGACTAAVXNTAAGTIT

21-A-II bp :

AAAAACCATGATTACGGAlTC TEGTACTAATGCCTAAGTIT

21-A-m bp :

AAAAACCATCGATTAGGATTC TEGTAGCTAATCCTAAGTIT

21-A-IV bp:

AAAAACCATCGGATTAGA'ITC TEGTAGCCTAATCTAAGTIT

a Duplexes were designed to expose three nucleotide asymmetric overhangs so as to ensure both head-to-tail ligation and that a unique bonding site (5'GATTA*) for drugs was contained in the center of the sequences. (A) 17,19,20,21,22, and 23 bp oligomers were used to study the effect of varying the distance between CC-1065 bonding sites on electrophoretic mobility, and the 20- and 21-mer were also used to determine the percentage of circular vs noncircular DNA in CC-1065- and ABC-modified oligomers, respectively. (B)21-A-I, 21-A-11, 21-A-111, and 21-A-IV bp oligomers were used to determine the direction and locus of ABC- and CC1065-induced bends relative to A,-tract bends.

covalent bonding reaction of the A subunit with DNA. Furthermore, the directionality of the CC-1065-induced DNA bending is in toward the minor groove of DNA. Interactions between the inside edge substituents of CC-1065 and the floor of the minor groove of DNA lead to winding of the DNA helix. The possible importance of CC-1065-induced DNA bending to sequence selectivity, relationship to bending in naturally occurring A-tracts, and biological activity are discussed. We have recently shown, using hydroxyl radical cleavage and two-dimensional 'H NMR studies, that the CC-1065-induced DNA bending appears to have structural characteristics (i.e., narrowing of the minor groove) that are similar to those associated intrinsically with A-tracts (34).

Materials and Methods Chemicals a n d Enzymes. CC-1065 was isolated from the fermentation broth of S. zelensis (18). The synthetic analogues (Figure 2) used in this study were obtained from J. Patrick McGovren a t the Upjohn Co., Kalamazoo, MI (29-31). Electrophoretic reagents [acrylamide, TEMED, ammonium persulfate, and bis(acrylamide)] were purchased from Bio-Rad. T4 polynucleotide kinase and T 4 DNA ligase were from Bethesda Research Laboratories. [ T - ~ ~ P ] A Twas P from ICN. X-ray film, intensifying screens, and developing chemicals were from Kodak. Preparation of Oligonucleotides. A series of oligonucleotides (17, 19, 20, 21, 22, and 23 bp) (Table IA) and 21-A-I, 21-A-11, 21-A-111, and 21-A-IV were synthesized on an automated DNA synthesizer (Applied Biosystems 381A) by the phosphoramidite method. The oligomers were then deprotected separately with saturated ammonium hydroxide at 55 "C overnight. Solvent was evaporated a t room temperature. K i n a s e Reaction of Oligonucleotides. Approximately 10 pg of each oligonucleotide of complementary sequence was mixed in 23 rL of solution containing 70 mM Tris-HC1 (pH 7.6), 10 mM MgCl,, and 5 mM dithiothreitol and 5'-labeled with 30 pCi of

Chem. Res. Tonicol., Vol. 4, No. 2, 1991 205 [ Y - ~ ~ P ] Aand T P 6 units of T4 polynucleotide kinase a t 37 "C for 20 min. After labeling with [@']ATP, 1 pL of 0.1 M cold ATP and an additional 6 units of T4 polynucleotide kinase were added, and kinasing reaction was continued in the same buffer at 37 "C for a n additional 1 h. Purification of Kinased Oligonucleotides. Kinased oligonucleotides were heated to 55 "C and cooled slowly to 4 "C to form hybridized duplexes. Duplexes were electrophoresed on the 8% nondenaturing polyacrylamide gel until the bromophenol blue marker had migrated 23 cm in a 30 cm X 36 cm X 0.8 mm gel. The ratio of monoacrylamide to bis(acry1amide) was 2 9 1 (w/w). The electrophoresis buffer was 50 mM Tris-borate (pH 8.3) and 1 mM EDTA (TBE buffer). The gels were electrophoresed at room temperature at 7 V/cm. After the gels were exposed to X-ray film, duplexes were excised from the gel, minced with a blade, and extracted with 400 pL of annealing buffer [lo0mM NaCl, 10 mM Tris-HC1 (pH 8.0)]. DNA S t r a n d Breakage Assay. The DNA strand containing the CC-1065alkylation site on 21 bp oligomers was 5'-end-labeled, hybridized with the unlabeled complementary strand, and purified as described above. Aliquots of purified 21 bp duplexes were modified with 2.8 pM CC-1065 as described above. Drug-modified duplexes were resuspended in 100 pL of DSC and heated a t 90 "C for 30 minutes to produce strand breakage at the covalent modification site (25). The DNA samples were denatured by heating at 90 "C for 30 min in formamide (80%)-NaOH (10 mM) solution and then subjected to 12% denaturing gel electrophoresis in parallel with DNA sequencing reactions. DNA Sequencing. Purine- and pyrimidine-specificsequencing reactions were carried out according to the methods of Maxam and Gilbert (35),and an adenine-specific sequencing reaction was carried out according to the method of Iverson and Dervan (36). D r u g Binding, Ligation, a n d Gel Analysis. Both of the complementary oligonucleotides (approximately 10 pg) were labeled with [@?]ATP (Amenham) by T4 polynucleotide kinase as described before, and the duplex DNA was gel purified by a n 8% nondenaturing polyacrylamide gel electrophoresis. For drug modification, duplex oligomers (10 pL) were incubated with the same volume of 2.8 mM A, 0.28 mM AB, and 28 pM ABC, AB'C', and CC-1065 a t room temperature for 3 days, followed by a n ethanol precipitation to remove unbound drug molecules. Drug-modified and nonmodified duplexes were self-ligated to produce multimers in 25 pL of ligation buffer with 1 unit (1pL) of T4 DNA ligase at room temperature overnight. The ligation buffer contains 25 mM Tris-HC1 (pH 7.6), 5 mM MgClz, 2.5% (w/v) poly(ethy1ene glycol) 8000, 0.5 mM ATP, and 0.5 mM dithiothreitol. Ligated multimen were electrophoresed on an 8% nondenaturing polyacrylamide gel and the radiolabeled bands located by autoradiography. C i r c u l a r i z a t i o n Experiment. In the first-dimensional electrophoresis, ligated multimers of drug-modified or control oligomer were electrophoresed on an 8% nondenaturing polyacrylamide gel until the bromophenol blue marker reached the bottom of gel, and the gel fragment containing the DNA was cut from the gel. After equilibration of the gel fragment with T B E buffer containing 50 pg/mL chloroquine phosphate for 15 min, it was laid down on top of a glass plate and allowed to polymerize with 8% polyacrylamide containing 50 pg/mL chloroquine phosphate. After second-dimensional electrophoresis in the presence of the same concentration of chloroquine as the gel, the radiolabeled spots representing both DNA families were removed from the gel and crushed, and DNA was extracted with doubledistilled water from the gel. The solution containing the eluted DNA was lyophilized and dissolved in 90% (w/v) formamide in 0.1 M Tris-HC1 (pH 7.8), boiled for 30 minutes, and cooled in ice, before 8% denaturing polyacrylamide gel electrophoresis (7 M urea) was used to determine lengths of DNA in each radioactive spot. The lengths of DNA in each radioactive spot (linear DNA and circular DNA) were measured from comparison with the 20 bp ladder-size markers, which were produced by ligation of the 20 bp oligomer duplex. Chloroquine U n w i n d i n g Experiments. T o probe the winding effect of 02-1065 and its analogues, chloroquine was added to the polyacrylamide gel and running buffer at concentrations between 0.05 and 2 pg/mL during the electrophoresis. In the presence of different concentrations of chloroquine, the

206 Chem. Res. Toxicol., Vol. 4, No.2,1991 A 1

AG 2

TC 3

C 4

D 5

C

1 1 A

G G C

n 1 1 A

G T A

C C A

G

G

Figure 3. Autoradiogram of the thermally induced strand breakage assay of CC-1065-modified 21 bp oligomers. Lanes 1, 2, and 3 reactions are A, AG, and T C specific reactions, respectively. Lanes 4 and 5 contain the non-drug-modified oligomers and CC-1065-modified oligomer (0.28 pM), respectively, following thermal cleavage. The DNA samples were prepared as described under Materials and Methods and electrophoresed on a 12% denaturing polyacrylamide gel until the bromophenol blue marker had migrated 10 cm in a 30 cm X 36 cm X 0.4 mm gel. changes in the electrophoretic migratory rate of ligated multimers of drug-modified 21-mer oligomers (Table IA) were analyzed by determining RL values.

Results Characterization of Drug-Oligonucleotide Adducts by Gel Electrophoresis and DNA Strand Breakage Assay. A comparison of the electrophoretic mobility of CC-1065, AB'C', ABC, and non-drug-modified monomers revealed that while non-drug-modified monomers of 15, 17, and 19 bp oligomers are melted to single-stranded DNA by incubation a t room temperature for 3 days, drugmodified monomers of 15,17, and 19 bp oligomers maintain duplex DNA structure (unpublished results). These results are consistent with the marked stabilizing effect of CC-1065 (27,37-38). The absence of detectable single-stranded DNA in drug-modified monomers suggests that all the duplexes were saturated with CC-1065. Nondrug-modified monomers of 21 bp oligomers are not melted under the same reaction conditions and have almost the identical electrophoretic mobility as drug-modified monomers. The covalent modification site of CC-1065 on 21 bp oligomers was analyzed on a 12% sequencing gel by the thermally induced strand breakage assay (25) (Figure 3). DNA sequencing reactions of A (lane l),AG (lane 2), and TC (lane 3) reveal the expected sequences of the covalently modified strand. Thermal treatment of CC-1065-DNA adducts (lane 5) produces the major product of DNA strand breakage (arrow A; species "A" in Figure 1)corresponding to a fragment that still contains the apurinic acid and therefore migrates approximately 1.5 bp higher than the expected adenine residue within the 5'TTA sequence. The minor product (arrow B; species "B" in Figure 1)that comigrates with the product of an adeninespecific reaction arises from the presumed second @-eliminationreaction, which leads to loss of the modified deoxyadenosine.

Lee et al.

Subsequent piperidine treatment produces quantitative conversion of the major product (arrow A) to the minor product (arrow B) (unpublished results). In addition, virtually all the full-length covalently modified strands are converted into shorter DNA fragments after thermal treatment, indicating that all the duplexes were saturated with CC-1065 (lane 5). Effect of Varying the Distance between CC-1065 Bonding Sites on the Electrophoretic Mobility of Ligated Multimers under Nondenaturated Conditions. When A-tracts occur in phase with the helical turn, the individual bends add up to a large overall bend in ligated multimers, which shows an increasing change in RL values (ratio of apparent size to true size) at higher molecular weight. In contrast, when A-tracts occur 180° out of phase with the helical turn, the bends cancel out each other, resulting in a constant change in R L values at higher molecular weight (I). In order to study the effect of varying the distance between CC-1065 bonding sites on electrophoretic mobility, 17,19,20,21, and 23 bp oligomers were modified with CC-1065 and ligated into multimers, and their electrophoretic mobility was analyzed on an 8% nondenaturing polyacrylamide gel. The autoradiogram of ligation products of non-drugmodified (C lanes) and CC-1065modified (D lanes) 17,19, 20,21, and 23 bp oligomers is shown in Figure 4A. While drug-modified ligation products of 20 bp oligomers (lane 6) and 21 bp oligomers (about 2 times the normal helical turn of 10.5 bp) (lane 8) show considerable retardation in electrophoretic mobility compared to control ligation products (lanes 5 and 7), drug-modified ligation products of 17, 19, and 23 bp oligomers (lanes 2,4, and 10) show little retardation in electrophoretic mobility compared to control ligation products (lanes 1,3, and 9). The ratio of apparent size to true size (RL) for each of the ligation products in Figure 4A was calculated and plotted against molecular weight (Figure 4B). The increasing change in RL values of ligation products of 20 and 21 bp oligomers indicates that CC-1065-induced bends are amplified as a consequence of coherent addition of in-phase CC-1065 induced bends. Ligation products of 17, 19, and 23 bp oligomers produce little change in RL values because the bends are presumably out of phase. These results indicate that increasing change in RL value is caused by CC-1065induced bending of DNA, not by localized frictional effects of CC-1065-DNA adducts or the positive charge on the 6-amino tautomeric form (22)of the covalently modified adenine in the CC-1065-DNA adducts. The maximum retardation in electrophoretic mobility is achieved in 20 bp oligomers, thus indicating that the CC-1065-modified 20 bp oligomers are the most closely phased with the helical turn after drug bonding to DNA. This suggests that CC-1065 induces not only DNA bending but also DNA winding equivalent to about 1bp per covalent modification site, because maximum bending occurs in 20 bp oligomers (2 X 10) rather than 21 bp oligomers (2 X 10.5). Effect of Chloroquine on the RL of 11 Multimers of a 21 bp Oligomer Modified by CC-1065, AB'C', and ABC. Experiments were designed using chloroquine, a known intercalator that unwinds DNA, to further examine the question of CC-1065-induced winding that is produced in addition to the DNA bending demonstrated in the previous section. If ligation products of the 21 bp oligomer have reduced RLs relative to the 20-mer when modified with CC-1065, because of drug-induced winding, then incremental addition of chloroquine should first increase the RLof ligation products of the 21-mer as the unwinding by chloroquine annuls the CC-1065induced winding. Beyond

CC- 1065-Induced DNA Bending

Chem. Res. Toxicol., Vol. 4, No.2,1991 207

17e~ 1 9 e ~ n n C D C D

1.5

I

-5

. 0.0

0.5

1 .o

1.5

2.0

Chloroquine Conc. (ug/mlJ

Figure 5. Plot of RLvalues against increasing concentrationof chloroquine of the 11 multimer of the 21 bp oligomer modified with CC-1065, AB’C’, and ABC. (Filled circles) 11 multimers of the 21 bp oligomer modified with CC-1065;(open squares) AB%’; (open circles) ABC.

4M 1

2

3

4

5

7

6

8

910

,

.

B 1.6

-

1.4

-

1.2

-

-

d) 3

9

0.8; 0

.

, 50

.

, 100

.

150

, 200

.

,

I

250

Length in Base Pairs

Figure 4. Panel A Autoradiogram of the ligation products of 17,19,20,21, and 23 bp oligomers modified with CC-1065 on an 8% nondenaturing polyacrylamide gel. Lanes 1,3,5, 7, and 9 are control ligation products (C), and lanes 2,4,6,8, and 10 are ligation products of 28 pM CC-1065 modified oligomers (D). The ligation products were prepared as described under Materials and Methods. Arrowheads indicate the location of the monomer (M) and dimer (D). Panel B: Plot of RL values vs total length of oligomers in base pairs from the ligation products of 17,19,20, 21, and 23 bp oligomers modified with CC-1065. R L values were obtained from the results of experiments shown in panel A.

a certain concentration of chloroquine, when the amount of chloroquine-induced unwinding of DNA exceeds the winding effect of CC-1065, the RL will decrease as the CC-1065-induced bends are no longer phased. The results of these experiments, which are summarized in Figure 5, show exactly this predicted pattern (i.e., initial increase in RL followed by a subsequent decrease) when an analysis of the R L s of the 11 multimers of the CC-1065- and AB’C’-modified 21 bp oligomer a t increasing chloroquine concentrations is carried out. Significantly, while AB’C’ mimics CC-1065 in showing an increase in RL value, prior to a decrease as the chloroquine concentration is increased, ABC does not show the initial increase in R b 3 We in-

terpret this to imply that it is the inside edge substituents that are unique to CC-1065 and AB’C’ which are responsible for the observed winding effect. Effect of Varying the Structure of the B and C Subunits of CC-1065 on DNA Bending. To further examine the effect of the fine structure of the B and C subunits of CC-1065 on DNA bending, the electrophoretic mobilities of ligation products modified with CC-1065, AB’C‘, and ABC in 20 and 21 bp oligomers were compared (Figure 6A). While CC-1065- and AB’C‘-modified ligation products produce more retardation in electrophoretic mobility in 20 bp oligomers (lanes 2 and 3) than in 21 bp oligomers (lanes 6 and 7), the reverse is true for ABC (lanes 4 and 8). A histogram of the RL- 1values of octamers of 19,20, 21, and 23 bp oligomers modified with CC-1065, AB’C’, and ABC is shown in Figure 6C-A. While CC-1065 and AB’C’ produce a higher RL - 1value in 20 bp oligomers than in 21 bp oligomers, ABC produces a higher RL - 1value in 21 bp oligomers than in 20 bp oligomers. Since the maximum amount of bending is approximately the same for CC-1065, AB’C’, and ABC but is achieved in either ligated 20 or 21 bp oligomers, these results suggest that the binding interactions between the B and C subunits and DNA modulate the amount of DNA winding but do not greatly effect the magnitude of DNA bending. Determination of the Importance of the Covalent Bonding Reaction in Contributing to DNA Bending Produced by ABC. Since the A subunit alone has sufficient structural information to mediate the sequence selectivity of the entire ABC molecule (33), it was important to determine whether the covalent bonding of the A subunit to the N3 of adenine is sufficient to mediate the DNA bending. A comparison of the electrophoretic mobility of ligation products modified with A, AB, and ABC in 20 and 21 bp oligomers is shown in Figure 6B. The A subunit modified ligation products of 21 bp oligomers (lane 8) produce significant retardation in electrophoretic mobility compared to those of control 21 bp oligomers (lane 5). Furthermore, the A subunit modified ligation products of 21 bp oligomers (lane 8) produce more retardation in electrophoretic mobility than AB (lane 7) and ABC (lane 6). These results show that DNA bending occurs as a consequence of the covalent bonding reaction of the A Furthermore, high concentrations of chloroquine (over 50 pg/mL) completely abolished the retardation of electrophoretic mobility, indicating that the phasing property of bent DNA can be annulled by the DNA unwinding effect on DNA of chloroquine intercalation (data not shown).

208 Chem. Res. Toxicol., Vol. 4, No.2,1991

Lee et al.

u

M-

,

21 bp

20 bp

p

u

0.5

,

i

A II

0 0.4

-

0.3

-

R L - ~

.

0.2

-

(+)-CC-l065 E (+)-AB’C‘ I (+)-ABC

0.5

, B

(+)-A (+)-AB I (+)-ABC

H 0.4

1

0.1

0.0 19

20

21

23

Oligomer Size in Base Pairs

19

20

21

23

Oligomer Size in Base Pairs

Figure 6. Panel A: Autoradiogram of ligation products of 20 and 21 bp oligomers modified with CC-1065, AB’C’, and ABC on an 890 nondenaturing polyacrylamide gel. Lanes 1and 5 are control ligation products. Lanes 2 and 6 are 28 p M CC-1065 modified ligation products, lanes 3 and 7 are 28 pM AB’C’ modified ligation products, and lanes 4 and 8 are 28 p M ABC modified ligation products. The ligation products were prepared as described under Materials and Methods. Arrowheads indicate the location of monomer (M) and dimer (D).Panel B: Autoradiogram of ligation products of 20 and 21 bp oligomers modified with ABC, AB, and A on an 8% nondenaturing polyacrylamide gel. Lanes 1and 5 are control ligation products. Lanes 2 and 6 are 28 p M ABC modified ligation products, lanes 3 and 7 are 0.28 mM AB modified ligation products, and lanes 4 and 8 are 2.8 mM A modified ligation products. The ligation products were prepared as described under Materials and Methods. Arrowheads indicate the location of the monomer (M) and the dimer (D).Panel C: Histogram of the R L - 1values of octamers of 19,20,21,and 23 bp oligomers modified with 2.8 mM A, 0.28 mM AB, and 28 pM ABC, AB’C’, and CC-1065. (A) R L - 1values for CC-1065, AB’C’, and ABC. (B) R L - 1 values for A, AB, and ABC.

subunit with DNA. Consequently, the A subunit alone has sufficient structural information to mediate the DNA bending produced by ABC. In contrast to the results with CC-1065and AB’C’ in Figure 6A, ABC, AB, and A subunit modified ligation products produce more retardation in electrophoreticmobility in 21 bp oligomers (lanes 6-8)than 20 bp oligomers (lanes 2-4); see also a histogram of the RL - 1values of octamers of 19,20,21, and 23 bp oligomers modified with A, AB, and ABC in Figure 6C-B. This result is consistent with the suggestion made previously that the inside edge substituents of CC-1065 and AB’C’ are associated with winding of DNA, since A, AB, and ABC lack these substituents.

Determination of the Magnitude of the Bending Angles Produced by CC-1065, AB’C‘, ABC, AB, and A from RL Values. The estimate of bending angles in the 5’GATTA* (A* is the covalent modification site) sequence produced by CC-1065, AB’C’, ABC, AB, and A was calculated by using a calibration equation for gel mobility anomalies derived by Koo and Crothers (39). The 140 bp (heptamers of 20 bp oligomers) for CC-1065 and AB’C‘ and 147 bp (heptamers of 21 bp oligomers) for A, AB, and ABC were chosen because they are within the optimal range for the application of a calibration equation, and maximum retardation in electrophoretic mobility is achieved in 20 bp oligomers for CC-1065 and AB’C’ and in 21 bp oligom-

Chem. Res. Toxicol., Vol. 4, No. 2, 1991 209

CC-1065-Induced DNA Bending

Table 11. Calculation of DNA Bending Angles Produced by A, AB, ABC, AB'C', and CC-1065 in the 5'GATTA* Sequence Contained in the 20- and 21-mer Oligomers Listed in Table I optimum bending angle, size of RL bending size for deg, from the compound oligomers valueo angle, de$ circularization, bp circularizationd A AB ABC AB'C' CC-1065

21 21 21 20 20

1.310 1.262 1.262 1.257 1.257

14.9-19.8 13.7-18.2 13.7-18.2 14.5-19.2 14.5-19.2

C

c

c 168 c 180

C

17.7 C

14.0

O R L values were calculated from 147 bp ligation products for 21 bp oligomers and from 140 bp ligation products for 20 bp oligomers. *The estimate of bend angles of each compound was calculated by using a calibration equation for gel mobility anomalies derived by Koo and Crothers (39). cNot determined. dThe bond angles of CC-1065 and ABC from the circularization experiment were determined by the method described in Husain et al. (17).

em for A, AB, and ABC. The sizes of oligomers, RL values, and calculated bend angles for each compound are shown in Table 11. An estimate of AB- and ABC-induced bend angles (14-18') is smaller by 1-2' than that of the A subunit induced bend angle (15-20°), suggesting that attachment of the B and C subunits to the A subunit reduces slightly the magnitude of the bending angle. AB'C' and CC-1065 apparently produce the same magnitude of DNA bending as the A subunit. Determination of the DNA Bending Produced by CC-1065 and ABC by Analysis of the Two-Dimensional Gel Electrophoresis for Circular DNA in Drug-Modified Oligomers following Ligation. In addition to bent DNA, other unusual DNA structures, such as cruciforms, lariats, and gaps, can cause anomalous electrophoretic mobility (40-43). To confirm whether ABC- and CC-1065-induced DNA bending is responsible for the observed anomalous retardation in electrophoretic mobility, two-dimensional gel electrophoresisexperiments were carried out (3, 17). Figure 7 shows typical autoradiograms of a two-dimensional gel containing the ligation products of drug-nonmodified oligomer and the 21 bp oligomer (Table IA) modified with ABC. The radioactive spots corresponding to both circular and noncircular DNA in these gels were cut out, and the DNA was extracted separately. The structure and size of DNA in each isolated radioactive spot was determined by analysis on denaturing gels, as described under Materials and Methods. The upper families of radioactive spots were determined to be circular DNA (slower migration in second-dimensional electrophoresis than the lower family), and the lower families were noncircular? Ligation products of nondrug-modified oligomers could only be shown to produce linear DNA in the molecular weight range that could be evaluated (Figure 7A), while there were significant amounts of small circular DNAs in ligated products of ABC-modified oligomers in the same molecular weight range. A graphical representation of circularization efficiency vs the molecular weight of multimers of control- and ABC-modified oligomers is shown in Figure 8. For the ABC-modified oligomer, the percentage of radioactivity in a circle vs the total in circular and linear DNA is plotted against the length of DNA in base pairs for each of the ligated products. From the optimal size of DNA required for circularizationby ABC, the approximate value for DNA The upper family of DNA molecules (Figure 7) was determined to be circular species from an analysis on sequencing gels with appropriate marker standards (data not shown). Although these DNA molecules were small-sized,they showed remarkably slow mobility in comparison to the same-sized noncircular DNA molecules on 8% nondenaturingpolyacylamide gel. Furthermore, these molecules showed little change in mobility on electrophoresis in nondenaturingpolyacrylamide gels that contained chloroquine (50 pg/mL), while noncircular bent DNA molecules were distorted into out-of-phase bends and showed faster migration with the addition of chloroquine (Figure 7).

B

1s1 Dlmension ( eo&Polyacrylamide)

Figure 7. Analysis of ligation products of 21 bp oligomers (Table IA) without drug modification (A) or following modification with ABC (B) by two-dimensional gel electrophoresis. In (B), the radioactive spots in the upper family are circular 6-9 multimer of the 21 bp oligomer, and those in the lower family are noncircular multimers of 21 b p oligomer. The numbers indicate the multiplicity of the oligomers in the radioactive spots, which were determined as described under Materials and Methods.

bending magnitude can be determined as follows for ABC, by dividing 360' by 168' (the number of base pairs in the circle), an average 2.14' of bending per base pair was obtained. Since the inherent flexibility of DNA is about 1.3'/bp (2,3,44), we determine that for the eight druginduced bends introduced at optimum circularization, a total of (2.14 - 1.3) X 168 = 141.1 of bending, or about

Lee et al.

210 Chem. Res. Toxicol., Vol. 4, No. 2, 1991

t

I

0 0

50

100

150

200

Multimers Size (bp)

Figure 8. Circularization efficiency for ABC, as expressed by the percentage of radioactivity in the circular DNA, to the total amount of radioactivity in the ligation products of the same size. The radioactivity of each spot was measured by scintillation counting, and the percentage of circular DNA at the given length of DNA in the total species of DNA was calculated. 17.7' per bend, is calculated. This number agrees well with the bending magnitude value calculated from the RL value (see Table 11). A similar two-dimensional and subsequent one-dimensional gel analysis was carried out for CC-1065 but using the 20-mer (Table IA) rather than 21-mer, which contained the same bonding sequence (5'GATTA*) as was used for ABC. In comparison to the ABC-modified oligomer, the optimum size for circularization is increased from 168 to 180 bp for the CC-1065-modifiedoligomer, and the circularization efficiency is slightly lower for CC-1065 than ABC (unpublished results). For the CC-1065-induced bend, the circularization experiment provides a calculated bending angle of about 14.0', which is somewhat less than that obtained from consideration of RL values (see Discussion). Determination of t h e Direction a n d Locus of the Drug-Induced Bends in DNA. To determine the directionality of bending of DNA by CC-1065 and related drugs, and also the locus of the drug-induced bending, the four oligomers shown in Table IB were used. It has been previously shown that the direction of A-tract-induced bending is in toward the minor groove and that the center of the bend corresponds to the center of the A-tract ( I , 39). The four oligomers I, 11, 111, and IV, listed in Table IB, were designed to vary in distance between the center of the A-tract and the following drug covalent modification site by 10, 11, 12, and 13 bp, respectively. Nondenaturing gel electrophoresis of the oligomers 1-IV (Table IB) shows that modification with ABC and CC-1065 leads to increased retardation of gel mobility in comparison to non-drug-modified oligomers (unpublished results). Since oligomers I, 11,111,and IV contain an A-tract that is separated by about one helical turn from the drug bonding site, we conclude that both bends are in the same direction, i.e., in toward the minor groove of DNA. A comparison of RL values for a 5 multimer of these four oligomers modified with CC-1065 and ABC leads to some important conclusions (Figure 9). For ABC-modified oligomers the maximum bending occurs in oligomer 111, in which 12 bp separate the center of the A-tract from the covalent modification site. This shows that optimum inphase bending arises when one helical turn separates the center of the A-tract from the position between the two thymidines within the ABC recognition sequence

I

Distance between Center of A tract and Alkylation site(bp). Figure 9. Effect of distances in base pairs between the center of A5-tractand the drug covalent modification site on RL values of 5 multimer of 21-A-I, 21-A-II,21-A-III,and 21-A-IV modified with CC-1065 and ABC. These oligomers were designed to vary in distance between the center of the A-tract and the following drug covalent modification site by 10, 11, 12, and 13 bp, respectively. o = ABC-modified oligomers; @ = CC-1065-modified oligomers.

(GATTA*) and consequently provides compelling evidence for the locus of drug-induced bending between two thymidine within the recognition sequence. However, for the CC-1065-induced bend, optimum bending occurs when 11 bp separate the center of the A-tract from the adjacent drug-DNA adduct site; Le., oligomer I1 (Figure 9) produces maximum retardation. This is exactly what is expected, taking into account the earlier observation that CC-1065 induces the winding of DNA by the equivalent of about 1 bp per covalent modification and consequently reduces the base pair number per two helical turns to 20.0. The overall decrease in RL values of all oligomers modified with CC-1065 and the relatively small difference of RL values between oligomer I1 and I11 modified with CC-1065 compared to those modified with ABC can be explained as follows. First, the overall decrease in RL of all oligomers modified with CC-1065 relative to ABC is because CC-1065 produces a decrease in base pairs per helical turn between the A-tracts and consequently creates out-of-phase bending of A-tracts. Second, the relatively small difference in RL between oligomers I1 and I11 for CC-1065, relative to ABC, is because oligomer I1 is less out of phase compared to oligomer 111, since there is 1 bp more per helical turn after CC-1065 modification than before. Most importantly, these results taken together show that the locus of bending is at the same site in the recognition sequence &e., between the two thymidines) for both ABC and CC-1065.

Discussion Studies from this laboratory in collaboration with our colleagues at the Upjohn Co. are aimed toward understanding the molecular basis for the potent biological effects of CC-1065 and its synthetic analogues (19). Structure-activity relationships clearly define DNA as the molecular target for the most potent cytotoxic and antitumor properties of CC-1065 (32). The ability to alkylate DNA is crucial for both the potent cytotoxic effects and antitumor activity of CC-1065 analogues. However, some biological effects, such as delayed death, require other structural features of CC-1065 in addition to the alkylating potential of the drug molecule (31). One of the most

Chem. Res. Toxicol., Vol. 4, No. 2, 1991 211

CC-1065-Induced DNA Bending

surprising properties of the CC-1065 molecule is that its DNA sequence selectivity can be mediated by the alkylating subunit alone (33). The B and C subunits serve primarily to increase the kinetics of the alkylation reaction. We have argued elsewhere (19) that the molecular basis for the DNA sequence selectivity is due to a sequencedependent conformational flexibility and perhaps a sequence-dependent catalytic role of DNA. The results in this paper allow us to assign CC-1065-induced changes in DNA structure, such as bending and winding, to interaction between DNA and specific regions of the CC-1065 molecule. In addition, the information we have gained from the experiments described in the present paper extends the known relationship between the effects of covalent bonding of CC-1065 on local DNA structure and both the molecular basis for DNA sequence selectivity and, more tentatively, biological effects. Structural Origin of Drug-Induced Bending and Winding of DNA in the CC-1065 Molecule. The CC1065-induced bending of DNA is clearly associated with the A subunit and is therefore a consequence of the covalent bonding reaction. For compounds such as A, AB, and ABC, which lack the inside edge substituents common to the B and C subunits of AB’C’ and CC-1065 (Figure 2), maximum retardation of gel mobility occurs when the covalent modification sites are separated by 10.5 bp. However, for AB’C’ and CC-1065, maximum retardation of gel mobility and, consequently, maximum coherent addition of in-phase bending occur when the covalent modification sites are separated by 10 bp, indicating that the inside edge substituents, which are unique to AB’C’ and CC-1065, wind the DNA helix by the equivalent of about 1.0 bp per alkylation site. This translates into a winding angle of 34’, but this number could easily vary by as much as &8’, since the accuracy of the experiment is limited by the availability of only whole-number oligomers for analysis. While the DNA bending angles calculated from RL values and circularization efficiency are in good agreement for ABC, in the case of CC-1065 the RL value determined appears to overestimate the bending angle, as calculated according to the circularization efficiency assay, by about 2-3’. We believe the most likely reason for this discrepancy is that the CC-1065 molecule stiffens the DNA helix; and consequently, the inherent flexibility of DNA is reduced, resulting in an increase of the optimum size for the circularization and underestimate of bending magnitude, because circularization is dependent upon the inherent flexibility of DNA as well as the local bending magnitude contributed by the drug bonding to DNA. Since the DNA stiffening effects are also found with AB’C’ (unpublished results) we conclude that, just like winding of the helix, DNA stiffening is also a property associated with the inside edge substituents of the B and C subunits. These relationships between the structural characteristics of CC-1065 and the conformational/dynamic effects on DNA are summarized in Figure 10. Molecular Origin of CC-1065-Induced Bending in DNA and Comparison to the Intrinsic Bending Associated with A-Tracts. There are a number of structural similarities between the bends intrinsically associated with A-tracts and those induced by CC-1065 and related drugs. The most obvious similarities are the directionality of DNA bending, which in both cases is into the minor groove, and the magnitudes of DNA bending. For CC-1065 and related drugs, in the experiments described here the bending angle is about 14-19’, and for Ab- or A6-tracts it is about 12-16’ and 19-22’, respectively (39). The experimental determination of the exact magnitude of

DNA Winding and Stiffening DNA Bendina

f

/

/

\ .NH,

Figure 10. Diagram representing the correlation between the bending and winding effects and structural features of CC-1065.

bending by CC-1065 and related drugs is in some cases (e.g., CC-1065 and AB’C’) complicated by the associated winding and stiffening effects documented here. Furthermore, the magnitude of bending is dependent not only upon the immediate bonding sequence (e.g., AGTTA* vs GATTA*) but also by the flanking regions (45). To examine in more detail the structural basis for the DNA bending produced by CC-1065, we have used hydroxylradical footprinting and one- and two-dimensional proton NMR techniques (34). The conclusions drawn from the results of both techniques are in agreement and support the idea that compression of the minor groove, sandwiched between widened regions, is responsible for the observed drug-induced bending of DNA. This is analogous to that found for A-tracts (47). In addition, the compression and widening of the minor groove common for both A-tracts and CC-1065-induced bends is associated with a discontinuity at the 3‘-end of the A-tract (39) or between the covalently modified base pair and the base pair to the 3’-side on the covalently modified strand, respectively (34). Maximum compression of the minor groove by CC-1065 is localized between the two thymidines in the 5‘-TTA* bonding sequence (34),in accord with the results reported here for the locus of DNA bending produced by CC-1065. Relationships between CC-1065-InducedBending and Winding of DNA and DNA Sequence Selectivity. CC-1065 causes bending of DNA, which appears in overall respects to be analogous to A-tracts, although the precise details may differ (34). The A subunit of CC-1065 has sufficient structural information to mediate both the bending of DNA and the sequence selectivity. Structurally, the A subunit contains the alkylating ability of CC-1065 (Figure 1). Therefore, it is the covalent bonding reaction that is responsible for both the bending of DNA and the sequence selectivity. It was an unexpected result to find that the A subunit had sufficient structural information to mediate the sequence selectivity of the entire CC-1065 molecule. Our observation that the sequence selectivity of CC-1065 is primarily determined, not by sequencerecognizing noncovalent binding but by the sequence requirements of the bonding step, has led to our postulate of a sequence-dependent conformational flexibility allowing adenine in certain environments to approach the electrophilic center sufficiently closely for bond formation. It might also be postulated to involve sequence-dependent electrophilic activation of the A subunit simultaneously with adenine N3 nucleophilic attack, in a process similar to the bifunctional catalysis observed in many enzymesubstrate reactions (48). The bent DNA structure that we observe after covalent adduct formation with CC-1065 would likely require the inherent sequence-dependent conformational flexibility that we have postulated is important in sequence recognition. Furthermore, sequences such as A-tracts, if their overall bent structure is similar

Lee et al.

212 Chem. Res. Toricol., Vol. 4, No. 2, 1991

to that induced by CC-1065, should be optimum bonding sites for CC-1065. Indeed, our observations that bonding of CC-1065 to A-tracts produces only a modest overall increase in bending of DNA and that these are one of the subsets of consensus bonding sequences for CC-1065 are in accord with this postulate (34). Moreover, the overall direction of bending and magnitude of bending induced by CC-1065 corresponds well to an A,- or A,-tract. However, it is important to recognize that because sequences such as 5’TTA are also equally, if not more, reactive to CC-1065 than A-tracts, and since these sequences do not have an intrinsic bent DNA structure, other factors, such as catalytic actiuation, may also play an important role in sequence recognition. We have recently demonstrated that a critically placed hydrogen-bonded water molecular may play a role in this catalysis (49). The precise structure of the B and C subunits can modulate or fine tune the sequence selectivity of CC-1065 analogues (33). While new covalent bonding sites are not created over that observed with the A subunit, the binding interactions of the B and C subunits increase or decrease the rate of reaction at individual bonding sites. Our present observation that the precise structure of the B and C subunits can cause winding (CC-1065 and AB’C’) or modulate bending (ABC and AB) provides a possible rationale for these fine-tuning effects. Recently, Powers and Gorenstein (50) published a model based upon molecular dynamics refinement for the structure of the AB’C’-DNA adduct, in which the druginduced bending of DNA is both quantitatiuely and qualitatively different from that reported here for (+)CC-1065. Their model shows a 60’ kink in the DNA with an associated widening of the minor groove of DNA. Clearly, both of these predictions are in sharp disagreement with the experimental results reported in this paper and previously published by us (34),where the bending of DNA is estimated to be within the range of 17-22’ and is associated with a distinct narrowing of the minor groove. Biological Implications. The covalent bonding reaction of CC-1065 is crucial for the potent biological effects, such as cytotoxicity and antitumor activity. Trisubunit compounds, which lack the ability to covalently bond to DNA, show cytotoxic activity several orders of magnitude less than CC-1065 and negligible antitumor activity. On the other hand, the A subunit, although considerably less cytotoxic, shows moderate antitumor activity at elevated doses (32). Since we have now demonstrated that CC-1065 and the A subunit produce bending of DNA into the minor groove in quite a similar manner to a naturally occurring A-tract, it is interesting to speculate how this may be responsible for the cytotoxic and antitumor activity of CC-1065 and its synthetic analogues. Bends in DNA are found naturally or may be induced by DNA binding proteins (for an insightful view of the biological significance of bends in DNA, see ref 51). These occur at the origin of replication (52),in a site-specific recombination site (7, 9, 53), and in transcriptional regulatory regions of DNA ( 4 , 5 , 5 3 , 5 4 ) .The covalent bonding of CC-1065 at such regions in DNA may mimic DNA binding protein effects on DNA or may entrap bent DNA structures that are normally in equilibrium with nonbent structures. This may have profound effects on one or more of the replication, recombination, and transcriptional events and perhaps therefore be responsible for some of the potent biological effects of CC-1065.

Acknowledgment. Supported by grants from the National Cancer Institute (CA-49751),the Welch Foundation, and the Burroughs Wellcome Fund. We are

grateful to David M. Bishop for preparation of the manuscript.

References (1) Koo, H.-S., Wu, H. M., and Crothers, D. M. (1986) DNA bending at adenine-thymine tracts. Nature (London) 320, 501-506. (2) Griffith, J., Bleyman, M., Rauch, C. A., Kitchin, P. A., and Englund, P. T. (1986) Visualization of the bent helix in kinetoplast DNA by electron microscopy. Cell 46, 717-724. (3) Ulanovsky, L., Bodner, M., Trifonov, E. N., and Choder, M. (1986) Curved DNA: Design, synthesis, and circularization. h o c . Natl. Acad. Sci. U.S.A. 83, 862-866. (4) Wu, H. M., and Crothers, D. M. (1984) The locus of sequence directed and protein-induced DNA bending. Nature (London) 308, 509-513. (5) Liu-Johnson, H.-N., Gartenberg, M. R., and Crothers, D. M. (1986) The DNA binding domain and bending angle of E. coli CAP protein. Cell 47, 995-1005. (6) Kuhnke, G., Fritz, H. J., and Ehring, R. (1987) Unusual properties of promoter-up mutations in the Escherichia coli galactese operon and evidence suggesting RNA polymerase-induced DNA bending. EMBO J. 6, 507-513. (7) Robertson, C. A., and Nash, H. A. (1988) Bending of the bacteriophage X attachment site by Escherichia coli integration host factor. J. Biol. Chem. 263, 3554-3557. (8) Thomson, J. F., and Landy, A. (1988) Empirical estimation of protein-induced DNA bending angles: Application to A site-specific recombination complex. Nucleic Acids Res. 16, 9687-9705. (9) Snyder, U. K., Thompson, J. F., and Landy, A. (1989) Phasing of protein induced DNA bends in a recombination complex. Nature (London) 341, 255-257. (10) Haran, T. E., and Crothers, D. M. (1989) Cooperativitiy in A-tract structure and bending properties of composite TnAn Blocks. Biochemistry 28, 2763-2767. (11) Ulanovsky, L. E., and Trifonov, E. N. (1987) Estimation of wedge components in curved DNA. Nature (London) 326, 720-722. (12) Burkhoff, A. M., and Tullius, T. D. (1988) Structural details of an adenine tract that do not cause DNA to bend. Nature (London) 331, 455-466. (13) Calladine, C. R., Drew, H. R., and McCall, M. J. (1988) The intrinsic curvature of DNA in solution. J.Mol. Biol. 201, 127-137. (14) Zinkel, S. S., and Crothers, D. M. (1987) DNA bend direction by phase sensitive detection. Nature (London) 328, 178-181. (15) Rice, J. A., Crothers, D. M., Pinto, A. L., and Lippard, S. J. (1988) The major adduct of the antitumor drug cis-diamminedichloroplatinum(I1) with DNA bends the duplex by =40° toward the major groove. Proc. Natl. Acad. Sci. U.S.A. 85, 4158-4161. 6) Haran, T. E., and Crothers, D. M. (1988) Phased psoralen cross links do not bend the DNA double helix. Biochemistry 27, 6967-6971. 7) Husain, I., Griffith, J., and Sancar, A. (1988) Thymine dimers bend DNA. Proc. Natl. Acad. Sci. U.S.A. 85, 2558-2562. 8) Hanka, L. J., Dietz, A., Gerpheide, S. A., Kuentzel, S. L., and Martin, D. G. (1978) CC-1065 (NSC-298223), a new antitumor antibiotic. Production in vitro biological activity, microbiological assays and taxonomy of the producing microorganisms. J . Antibiot. 31, 1211-1217. (19) Warpehoski, M. A., and Hurley, L. H. (1988) Sequence selectivity of DNA covalent modification. Chem. Res. Toricol. 1, 315-333. (20) Reynolds, V. L., McGovren, J. P., and Hurley, L. H. (1986) The chemistry, mechanism of action, and biological properties of CC-1065, a potent antitumor antibiotic. J.Antibiot. 39,319-334. (21) Bhuyan, B. K., Newell, K. A., Crampton, S. L., and VonHoff, D. D. (1982) CC-1065 (NSC 298223), a most potent antitumor agent: Kinetics of inhibition of growth, DNA synthesis, and cell survival. Cancer Res. 42, 3532-3537. (22) Lin, C. H. and Hurley, L. H. (1990) Determination of the major tautomeric form of the covalently modified adenine in the (+)CC-1065-DNA adduct by ‘H and 16N NMR studies. Biochemistry 29, 9503-9507. (23) Scahill, T. A., Jensen, R. M., Swenson, D. H., Hatzenbuhler, N. T., Petzold, G. L., Wierenga, W., and Brahme, N. D. (1990) An NMR study of the covalent and noncovalent interactions of CC-1065 and DNA. Biochemistry 29, 2852-2860. (24) Hurley, L. H., Reynolds, V. L., Swenson, D. H., Petzold, C. L., and Scahill, T. A. (1984) Reaction of the antitumor antibiotic CC-1065 with DNA. Structure of a DNA adduct with DNA se-

CC-1065-Znduced DNA Bending quence specificity. Science 226,843-844. (25) Reynolds, V. L., Molineux, I. J., Kaplan, D. J., Swenson, D. H., and Hurley, L. H. (1985) Reaction of the antitumor antibiotic CC-1065 with DNA. Location of the site of thermally induced strand breakage and analysis of DNA sequence specificity. Biochemistry 24,6228-6237. (26) Hurley, L. H., Warpehoski, M. A., Lee, C.-S., McGovren, J. P., Scahill, T. A,, Kelly, K. C., Wicnienski, N. A., Gebhard, I., and Bradford, V. S. (1990) Sequence specificity of DNA alkylation by the unnatural enantiomers of CC-1065 and its synthetic analogs. J. Am. Chem. Soc. 112, 4633-4649. (27) Needham-VanDevanter, D. R., and Hurley, L. H. (1986) Construction and characterization of a site-directed CC-1065 (N3adenine) DNA adduct within a 117 bp DNA restriction fragment. Biochemistry 25, 8430-8436. (28) Hurley, L. H., Needham-VanDevanter, D. R., and Lee, C.-S. (1987) Demonstration of the asymmetric effect of CC-1065 on local DNA structure using a site-directed adduct in a 117 base pair fragment from M13mpl. Proc. Natl. Acad. Sci. U.S.A. 84, 64 12-64 16. (29) Warpehoski, M. A. (1986) Total synthesis of U-71184, a potent new antitumor agent modeled on CC-1065. Tetrahedron Lett. 27, 4103-4106. (30) Kelly, R. C., Gebhard, I., Wicnienski, N., Aristoff, P. A., Johnson, P. D., and Martin, D. G. (1987) Coupling of cyclopropylpyrroindole (CPI) derivatives. The preparation of CC-1065, ent-CC-1065, and analogs. J. Am. Chem. SOC.109, 6837-6838. (31) Warpehoski, M. A,, and Bradford, V. S. (1988) Bis-des-hydroxy, bis-des-methoxy CC-1065. Synthesis, DNA binding, and biological activity. Tetrahedron Lett. 29, 131-144. (32) Warpehoski, M. A., Gebhard, I., Kelly, R. C., Krueger, W. C., Li, L. H., McGovren, J. P., Prairie, M. D., Wicnienski, N., and Wierenga, W. (1988) Stereoelectronic factors influencing the biological activity and DNA interaction of synthetic antitumor agents modeled on CC-1065. J. Med. Chem. 31, 590-603. (33) Hurley, L. H., Lee, C.-S., McGovren, J. P., Mitchell, M. A,, Warpehoski, M. A., Kelly, R. C., and Aristoff, P. A. (1988) Molecular basis for the sequence specific DNA alkylation by CC-1065. Biochemistry 27, 3886-3892. (34) Lin, C. H., Sun, D., and Hurley, L. H. (1991) (+)-CC-1065 produces bending of DNA that appears to resemble adenine/ thymine tracts. Chem. Res. Toxicol. (in press). (35) Maxam, A. M., and Gilbert, W. (1980) Sequencing end-labeled DNA with base-specific chemical cleavages. Methods Enzymol. 65, 499-560. (36) Iverson, B. L., and Dervan, P. B. (1987) Adenine specific DNA chemical sequencing reaction. Nucleic Acids Res. 15,7823-7830. (37) Swenson, D. H., Li, L. H., Hurley, L. H., Rokem, J. S., Petzold, G. L., Dayton, B. D., Wallace, T. L., Lin, A. H., and Krueger, W. C. (1982) Mechanism of interaction of CC-1065 (NSC 298223) with DNA. Cancer Res. 42, 2821-2828.

Chem. Res. Toxicol., Vol. 4, No. 2, 1991 213 (38) Wierenga, W., Bhuyan, B. K., Kelly, R. C., Krueger, W. C., Li, L. H., McGovren, J. P., Swenson, D. H., and Warpehoski, M. A. (1986) Antitumor activity and biochemistry of novel analogs of the antibiotic CC-1065. In Adoances in Enzyme Regulation (Weber, G., Ed.) Vol. 25, pp 141-155, Pergamon, New York. (39) Koo, H.-S., and Crothers, D. M. (1988) Calibration of DNA curvature and a unified description of sequence-directed bending. Proc. Natl. Acad. Sci. U.S.A. 85, 1763-1767. (40) Marini, J. C., Levene, S. D., Crothers, D. M., and Englund, P. T. (1982) Bent helical structure in kinetoplast DNA. Proc. Natl. Acad. Sci. U.S.A. 79, 7664-7668. (41) Gough, G. W., and Lilley, D. M. (1985) DNA bending induced by cruciform formation. Nature (London) 313, 154-156. (42) Konarska, M. M., Grabowski, P. J., and Sharp, P. A. (1984) Lariat RNA's intermediates and products in the splicing of messenger RNA precursors. Science (Washington, D.C.) 225, 898-903. (43) Thomas, D. C., Kunkel, T. A,, Casna, N. J., Ford, J. P., and Sancar, A. (1986) Activities and incision patterns of ABC excinuclear on modified DNA containing single base mismatches and extrahelical bases. J. Biol. Chem. 261, 14496-14505. (44) Shore, D., Langowski, J., and Baldwin, R. (1981) DNA flexibility studied by covalent closure of short fragments into circles. Proc. Natl. Acad. Sci. U.S.A. 73, 4833-4837. (45) Sun, D., and Hurley, L. H. (1990) unpublished results. (46) Burkhoff, A. M., and Tullius, T. D. (1987) The unusual conformation adopted by the adenine tracts in kinetoplast DNA. Cell 48,935-943. (47) Nadeau, J. G., and Crothers, D. M. (1989) Structural basis for DNA bending. Proc. Natl. Acad. Sci. U.S.A. 86, 2622-2626. (48) Fersht, A. (1985) in Enzyme Structure and Function, 2nd ed., W. H. Freeman, New York. (49) Lin, C. H., Beale, J., and Hurley, L. H. (1991) Structure of the (+)-CC-1065-DNA adduct. The critical role of ordered water molecules and implications for the involvement of phosphate catalysis in the covalent reaction. Biochemistry (submitted for publication). (50) Powers, R., and Gorenstein, D. G. (1990) Two-dimensional 'H and 31PNMR spectra and restrained molecular dynamics structure of a covalent CPI-CDPIpdigodeoxyribonucleotide decamer complex. Biochemistry 29,9994-10008. (51) Travers, A. A. (1990) Why bend DNA? Cell 60, 177-180. (52) Zahn, K., and Blattner, F. R. (1987) Evidence for DNA bending at the h replicating origin. Science (Washington, D.C.) 236, 416-422. (53) Goodman, S. D., and Nash, H. A. (1989) Functional replacement of a protein-induced bend in a DNA recombination site. Nature (London) 341, 251-254. (54) McAllister, C. F., and Achberger, E. C. (1989) Rotational orientation of upstream curved DNA affects promoter function in Bacillus subtilis. J. Biol. Chem. 264, 10451-10456.