Single-stranded oligonucleotide adducts formed by Pt complexes favoring left-handed base canting: steric effect of flanking residues and relevance to DNA adducts formed by Pt anticancer drugs

Platinum anticancer drug binding to DNA creates large distortions in the cross-link (G*G*) and the adjacent XG* base pair (bp) steps (G* = N7-platinated G). These distortions, which are responsible for anticancer activity, depend on features of the duplex (e.g., base pairing) and of the cross-link moiety (e.g., the position and canting of the G* bases). The duplex structure stabilizes the head-to-head (HH) over the head-to-tail (HT) orientation and right-handed (R) over left-handed (L) canting of the G* bases. To provide fundamental chemical information relevant to the assessment of such duplex effects, we examine (S,R,R,S)-BipPt(oligo) adducts (Bip = 2,2'-bipiperidine with S,R,R,S chiral centers at the N, C, C, and N chelate ring atoms, respectively; oligo = d(G*pG*) with 3'- and/or 5'-substituents). The moderately bulky (S,R,R,S)-Bip ligand favors L canting and slows rotation about the Pt-G* bonds, and the (S,R,R,S)-BipPt(oligo) models provide more useful data than do dynamic models derived from active Pt drugs. All 5'-substituents in (S,R,R,S)-BipPt(oligo) adducts favor the normal HH conformer (～97%) by destabilizing the HT conformer through clashes with the 3'-G* residue rather than through favorable H-bonding interactions with the carrier ligand in the HH conformer. For all (S,R,R,S)-BipPt(oligo) adducts, the S pucker of the 5'-X residue is retained. For these adducts, a 5'-substituent had only modest effects on the degree of L canting for the (S,R,R,S)-BipPt(oligo) HH conformer. This small flanking 5'-substituent effect on an L-canted HH conformer contrasts with the significant decrease in the degree of R canting previously observed for flanking 5'-substituents in the R-canted (R,S,S,R)-BipPt(oligo) analogues. The present data support our earlier hypothesis that the distortion distinctive to the XG* bp step (S to N pucker change and movement of the X residue) is required for normal stacking and X·X' WC H bonding and to prevent XG* residue clashes.     

A rare example of three abundant conformers in one retro model of the cisplatin-DNA d(GpG) intrastrand cross link. Unambiguous evidence that guanine O6 to carrier amine ligand hydrogen bonding is not important. possible effect of the Lippard base pair step adjacent to the lesion on carrier ligand hydrogen bonding in DNA adducts

Guanine O6 to carrier ligand hydrogen bonding is a central feature of many hypotheses advanced to explain the anticancer activity of cis-type anticancer drugs, cis-PtA(2)X(2) (A(2) = diamine or two amines). Early structural evidence suggested that cis-Pt(NH(3))(2)(d(GpG)) (the cross-link model for the key cisplatin-DNA adduct) and other cis-PtA(2)(d(GpG)) adducts exist exclusively or mainly as the HH1 conformer with head-to-head (HH) bases. The dynamic motion of the d(GpG) in these adducts is too rapid to permit definitive characterization of both the conformation and the H-bonding. Hence, we use retro models having A(2) ligands designed to slow the motion. Here, we employ Me(2)ppz (N,N'-dimethylpiperazine), which lacks NH groups. Me(2)ppz is unique in having sp(3) N-methyl groups directly in the coordination plane, allowing the coexistence of multiple conformers but hindering dynamic motion in Me(2)ppzPt(d(GpG)) and Me(2)ppzPt(GpG) retro models. Dynamic processes are decreased enough in Me(2)ppzPt(d(GpG)) to permit HPLC separation of three abundant forms. After HPLC separation, the three re-equilibrate, proving that the three forms must be conformers and that Me(2)ppz has little influence on conformer distribution. This marks the first reported characterization of three abundant conformers for one cis-PtA(2)(d(GpG)) adduct. From NMR evidence, the Me(2)ppzPt(d(GpG)) HH1 conformer has uncanted bases. Another conformer, one of two recently discovered conformer types, has head-to-tail (HT) bases with Delta chirality. For this Delta HT1 form, several lines of evidence establish that the dinucleotide moieties have essentially identical structures in d(GpG) (and GpG) adducts of Me(2)ppzPt and other cis-PtA(2) complexes. For example, the shifts of the highly structure-sensitive G H8 NMR signals are almost identical for the Delta HT1 form of all adducts. In previous models, the stabilization of the Delta HT1 form could be attributed to G O6 H-bonding to A(2) NH groups. Such H-bonds are not possible for Me(2)ppz. The unambiguous conclusions are that G O6 H-bonding is weak and that neither canting nor H-bonding is essential in HH forms. These two features are present in almost all other small models but are essentially absent in the cross-link base pair (bp) step in duplexes. We conclude from our work that the forces favoring canting and H-bonding are weak, and we hypothesize that steric effects within the Lippard bp step adjacent to this cross-link bp step easily overcome these forces.     

Relationship of solution and protein-bound structures of DNA duplexes with the major intrastrand cross-link lesions formed on cisplatin binding to DNA.

DNA bases in the three-base-pair (3bp) region of duplexes with the two major lesions of cisplatin (cis-PtCl(2)(NH(3))(2)) with DNA, namely d(XGG) and d(XAG) ( = N7-platinated base), differ in their relative positions by as much as approximately 3.5 A in structures in the literature. Such large differences impede drug design and assessments of the effects of protein binding on DNA structure. One recent and several past structures based on NMR-restrained molecular dynamics (RMD) differ significantly from the reported X-ray structure of an HMG-bound XGG 16-mer DNA duplex (Ohndorf, U.-M.; Rould, M. A.; He, Q.; Pabo, C. O.; Lippard, S. J. Nature 1999, 399, 708). This 16-mer structure has several significant novel and unique features (e.g., a bp step with large positive shift and slide). Hypothesizing that novel structural features in the XGG or XAG region of duplexes elude discovery by NMR methods (especially because of the flexible nature of the 3bp region), we studied an oligomer with only G.C bp's in the XGGY site by NMR methods for the first time. This 9-mer gave a 5'-G N1H signal with a normal shift and intensity and showed clear NOE cross-peaks to C NHb and NHe. We assigned for the first time (13)C NMR signals of a duplex with a GG lesion. These data, by adding NMR-based criteria to those inherent in NOESY and COSY data, have more specifically defined the structural features that should be present in an acceptable model. In particular, our data indicated that the sugar of the X residue has an N pucker and that the GG cross-link should have a structure similar to the original X-ray structure of cis-Pt(NH(3))(2)(d(pGpG)) (Sherman S. E.; Gibson, D.; Wang, A. H.-J.; Lippard, S. J. J. Am. Chem. Soc. 1988, 110, 7368). With these restrictions added to NOE restraints, an acceptable model was obtained only when we started our modeling with the 16-mer structural features. The new X-ray/NMR-based model accounted for the NOESY data better than NOE-based models, was very similar in structure to the 16-mer, and differed from solely NOE-based models. We conclude that all XGG and XAG (X = C or T) duplexes undoubtedly have structures similar to those of the 16-mer and our model. Thus, protein binding does not change greatly the structure of the 3bp region. The structure of this region can now be used in understanding structure-activity relationships needed in the design of new carrier ligands for improving Pt anticancer drug activity.     

Modification of second-sphere communication, leading to an unusually high abundance of the head-to-head conformer of cisplatin cross-link retro models

Rapid atropisomerization of cisplatin-DNA cross-link models, cis-PtA(2)G(2) (A(2) = two amines or a diamine, G = guanine derivative, bold font indicating a guanine not linked to another guanine), makes their NMR spectra uninformative. The conformers [two head-to-tail (DeltaHT and LambdaHT) conformers, one head-to-head (HH) conformer] are detected in (CCC)PtG(2) retro models (CCC = chirality-controlling chelates designed to reduce rotation around the G N7 to Pt bond by destabilizing the transition state). Clear trends are found with four CCC ligands, 2,2'-bipiperidine (Bip) and N,N'-dimethyl-2,3-diaminobutane (each with S,R,R,S and R,S,S,R configurations at the chelate ring N, C, C, and N atoms, respectively). S,R,R,S ligands favor left-handed G base canting and the LambdaHT form; R,S,S,R ligands favor right-handed canting and the DeltaHT form. The HH conformer is normally negligible unless G = 5'-GMP. However, understanding this 5'-phosphate effect is complicated by possible interligand interactions of the 5'-phosphate with the N1H of the cis-5'-GMP and a CCC NH; these interactions are referred to as second-sphere (SSC) and first-to-second-sphere (FSC) communication, respectively. We now investigate the four (CCC)PtG(2) models with 1-Me-5'-GMP, a G lacking the N1H needed for SSC. The phosphate location makes FSC possible in the major but not the minor HT form. The major form should increase from pH 3 to pH 7 because the phosphate is deprotonated at pH 7. However, the major DeltaHT form for the R,S,S,R pair did not change in abundance, and the major LambdaHT form for the S,R,R,S pair actually decreased. Thus, FSC is weak. At pH approximately 7 the HH conformer of the S,R,R,S pair had an abundance (40-44%) higher than that in any reported cis-PtA(2)G(2) adduct. FSC involving one 1-Me-5'-GMP could play a role. The high HH abundance and use of a pH jump experiment with (S,R,R,S)-BipPt(1-Me-5'-GMP)(2) allowed us to obtain the first deconvoluted CD spectrum for a cis-PtA(2)G(2) HH conformer. The CD features for the HH conformer are much weaker than for the HT conformers. Our findings are best interpreted to indicate that FSC is not important in aqueous solution, especially for the HT form. Weak FSC is consistent with recent models of the cross-link in duplexes. In contrast, crystals of fluxional models often reveal FSC, but not the more important SSC. SSC was unrecognized until our retro model studies, and the new results reinforce the value of studying retro models for identifying interactions in solution.     

Neomycin-neomycin dimer: an all-carbohydrate scaffold with high affinity for AT-rich DNA duplexes

A dimeric neomycin-neomycin conjugate 3 with a flexible linker, 2,2'-(ethylenedioxy)bis(ethylamine), has been synthesized and characterized. Dimer 3 can selectively bind to AT-rich DNA duplexes with high affinity. Biophysical studies have been performed between 3 and different nucleic acids with varying base composition and conformation by using ITC (isothermal calorimetry), CD (circular dichroism), FID (fluorescent intercalator displacement), and UV (ultraviolet) thermal denaturation experiments. A few conclusions can be drawn from this study: (1) FID assay with 3 and polynucleotides demonstrates the preference of 3 toward AT-rich sequences over GC-rich sequences. (2) FID assay and UV thermal denaturation experiments show that 3 has a higher affinity for the poly(dA)·poly(dT) DNA duplex than for the poly(dA)·2poly(dT) DNA triplex. Contrary to neomycin, 3 destabilizes poly(dA)·2poly(dT) triplex but stabilizes poly(dA)·poly(dT) duplex, suggesting the major groove as the binding site. (3) UV thermal denaturation studies and ITC experiments show that 3 stabilizes continuous AT-tract DNA better than DNA duplexes with alternating AT bases. (4) CD and FID titration studies show a DNA binding site size of 10-12 base pairs/drug, depending upon the structure/sequence of the duplex for AT-rich DNA duplexes. (5) FID and ITC titration between 3 and an intramolecular DNA duplex [d(5'-A(12)-x-T(12)-3'), x = hexaethylene glycol linker] results in a binding stoichiometry of 1:1 with a binding constant ～10(8) M(-1) at 100 mM KCl. (6) FID assay using 3 and 512 hairpin DNA sequences that vary in their AT base content and placement also show a higher binding selectivity of 3 toward continuous AT-rich than toward DNA duplexes with alternate AT base pairs. (7) Salt-dependent studies indicate the formation of three ion pairs during binding of the DNA duplex d[5'-A(12)-x-T(12)-3'] and 3. (8) ITC-derived binding constants between 3 and DNA duplexes have the following order: AT continuous, d[5'-G(3)A(5)T(5)C(3)-3'] > AT alternate, d[5'-G(3)(AT)(5)C(3)-3'] > GC-rich d[5'-A(3)G(5)C(5)T(3)-3']. (9) 3 binds to the AT-tract-containing DNA duplex (B* DNA, d[5'-G(3)A(5)T(5)C(3)-3']) with 1 order of magnitude higher affinity than to a DNA duplex with alternating AT base pairs (B DNA, d[5'-G(3)(AT)(5)C(3)-3']) and with almost 3 orders of magnitude higher affinity than a GC-rich DNA (A-form, d[5'-A(3)G(5)C(5)T(3)-3']).     

Marked dependence on carrier-ligand bulk but not on carrier-ligand chirality of the duplex versus single-strand forms of a DNA oligonucleotide with a series of G-Pt(II)-G intrastrand cross-links modeling cisplatin-DNA adducts

The N7-Pt-N7 adjacent G,G intrastrand DNA cross-link responsible for cisplatin anticancer activity is dynamic, promotes local "melting" in long DNA, and converts many oligomer duplexes to single strands. For 5'-d(A1T2G3G4G5T6A7C8C9C10A11T12)-3' (G3), treatment of the (G3)2 duplex with five pairs of [LPt(H2O)2]2+ enantiomers (L = an asymmetric diamine) formed mixtures of LPt-G3 products (1 Pt per strand) cross-linked at G3,G4 or at G4,G5 in all cases. L chirality exerted little influence. For primary diamines L with bulk on chelate ring carbons (e.g., 1,2-diaminocyclohexane), the duplex was converted completely into single strands (G3,G4 coils and G4,G5 hairpins), exactly mirroring results for cisplatin, which lacks bulk. In sharp contrast, for secondary diamines L with bulk on chelate ring nitrogens (e.g., 2,2'-bipiperidine, Bip), unexpectedly stable duplexes having two platinated strands (even a unique G3,G4/G4,G5 heteroduplex) were formed. After enzymatic digestion of BipPt-G3 duplexes, the conformation of the relatively nondynamic G,G units was shown to be head-to-head (HH) by HPLC/mass spectrometric characterization. Because the HH conformation dominates at the G,G lesion in duplex DNA and in the BipPt-G3 duplexes, the stabilization of the duplex form only when the L nitrogen adducts possess bulk suggests that H-bonding interactions of the Pt-NH groups with the flanking DNA lead to local melting and to destabilization of oligomer duplexes. The marked dependence of adduct properties on L bulk and the minimal dependence on L chirality underscore the need for future exploration of the roles of the L periphery in affecting anticancer activity.     

On the effect of covalently appended quinolones on termini of DNA duplexes

Quinolones are gyrase inhibitors that are widely used as antibiotics in the clinic. When covalently attached to oligonucleotides as 5'-acylamido substituents, quinolones were found to stabilize duplexes of oligonucleotides against thermal denaturation. For short duplexes, such as qu-T*GCGCA, where qu is a quinolone residue and T is a 5'-amino-5'-deoxythymidine residue, an increase in the UV melting point of up to 27.8 degrees C was measured. The stabilizing effect was demonstrated for all quinolones tested, namely nalidixic acid, oxolinic acid, pipemidic acid, cinoxacin, norfloxacin, and ofloxacin. The three-dimensional structure of (oa-T*GCGCA)2, where oa is an oxolinic acid residue, was solved by two-dimensional NMR spectroscopy and restrained molecular dynamics. In this complex, the oxolinic acid residues disrupt the terminal T1:A6 base pairs and stack on the G2:C5 base pairs. The displaced adenosine residues bind in the minor groove of the core duplex, while the thymidine residues pack against the oxolinic acid residues. The "molecular cap" thus formed fits tightly on the G:C base pairs, resulting in increased base-pairing fidelity, as demonstrated in UV melting experiments with the sequence oa-T*GGTTGAC and target strands containing a mismatched nucleobase. The structure of the "molecular cap" with its disrupted terminal base pair may also be helpful for modeling how quinolones block re-ligation of DNA strands in the active site of gyrases.     

Imprinting structural information from a GpG ligand into the configuration of a chiral diamine ligand through second-sphere communication in platinum(II) complexes

Cisplatin forms the cis-Pt(NH3)2(d(GpG)) cross-link with DNA. We have recently created novel d(GpG) conformations by using "retro models" (complexes having bulky carrier ligands designed to slow d(GpG) dynamic motion). Our results define four conformer classes: HH1, HH2, delta HT1, and delta HT2, with a head-to-head or head-to-tail base orientation and a phosphodiester backbone with a normal (1) or opposite (2) propagation direction. Moreover, each G residue can be syn or anti, and the base canting can be left-handed (L) or right-handed (R). Thus, 32 variants of cis-Pt(NH3)2(d(GpG)) are conceivable, but the adduct is too dynamic to study. Thus far, by using retro models, we have obtained evidence for five variants with d(GpG) but only four with GpG. We therefore selected Me2DAPPt(GpG) complexes for study by 1H and 31P NMR spectroscopy, CD spectroscopy, and molecular mechanics and dynamics (MMD) calculations. Coordinated Me2DAP (N,N'-dimethyl-2,4-diaminopentane) has N, C, C, N chiral centers designated, for example, as R,R,R,R. This ligand has greater flexibility and more readily inverted N centers than ligands used previously in GpG retro models. One goal was to determine whether the GpG ligand can control the configuration of a carrier ligand. (R,R,R,R)-Me2DAPPt(GpG) forms the anti, anti HH1 R variant almost exclusively. Equal populations of the two possible linkage isomers of (S,R,R,R)-Me2DAPPt(GpG) are formed, both favoring the anti, anti HH1 R, variant; however, the isomer with the 5'-G cis to the S nitrogen has sharper signals, suggesting that interligand interactions are more favorable. Indeed, this linkage isomer was the major product of isomerization when (R,R,R,R)-Me2DAPPt(GpG) was kept at pH approximately 9.5 to allow N center equilibration. Steric clashes between the Me2DAP C-Me groups and the G O6 atoms found by MMD calculations appear to disfavor the HH1 conformer of (S,S,S,S)-Me2DAPPt(GpG) and (S,S,S,R)-Me2DAPPt(GpG) complexes. These two complexes have a significant population of the anti, syn delta HT1 conformer, as indicated by broad 1H NMR signals and by 31P NMR and CD data. Equilibration of (S,S,S,R)-Me2DAPPt(GpG) at pH 9.5 leads to a mixture of (S,S,S,S)-Me2DAPPt(GpG) and at least one isomer of (S,S,S,R)-Me2DAPPt(GpG). Thus, second-sphere communication (hydrogen bonding and steric interligand interactions) influences both GpG conformation and Me2DAP configuration.     

Investigation relevant to the conformation of the 17-membered Pt(d(GpG)) macrocyclic ring formed by Pt anticancer drugs with DNA: Pt complexes with a Goldilocks carrier ligand

Platinum anticancer drug DNA intrastrand cross-link models, LPt(d(G*pG*)) (G* = N7-platinated G residue, L = R(4)dt = bis-3,3'-(5,6-dialkyl)-1,2,4-triazine), and R = Me or Et), undergo slow Pt-N7 bond rotation. NMR evidence indicated four conformers (HH1, HH2, ΔHT1, and ΛHT2); these have different combinations of guanine base orientation (head-to-head, HH, or head-to-tail, HT) and sugar-phosphodiester backbone propagation relative to the 5'-G* (the same, 1, or opposite, 2, to the direction in B DNA). In previous work on LPt(d(G*pG*)) adducts, Pt-N7 rotation was too rapid to resolve conformers (small L with bulk similar to that in active drugs) or L was too bulky, allowing formation of only two or three conformers; ΛHT2 was not observed under normal conditions. The (R(4)dt)Pt(d(G*pG*)) results support our initial hypothesis that R(4)dt ligands have Goldilocks bulk, sufficient to slow G* rotation but insufficient to prevent formation of the ΛHT2 conformer. Unlike the (R(4)dt)Pt(5'-GMP)(2) adducts, ROESY spectra of (R(4)dt)Pt(d(G*pG*)) adducts showed no EXSY peaks, a result providing clear evidence that the sugar-phosphodiester backbone slows conformer interchange. Indeed, the ΛHT2 conformer formed and converted to other conformers slowly. Bulkier L (Et(4)dt versus Me(4)dt) decreased the abundance of the ΛHT2 conformer, supporting our initial hypothesis that steric crowding disfavors this conformer. The (R(4)dt)Pt(d(G*pG*)) adducts have a low abundance of the ΔHT1 conformer, consistent with the proposal that the ΔHT1 conformer has an energetically unfavorable phosphodiester backbone conformation; its high abundance when L is bulky is attributed to a small d(G*pG*) spatial footprint for the ΔHT1 conformer. Despite the Goldilocks size of the R(4)dt ligands, the bases in the (R(4)dt)Pt(d(G*pG*)) adducts have a low degree of canting, suggesting that the ligand NH groups characteristic of active drugs may facilitate canting, an important aspect of DNA distortions induced by active drugs.     

DNA-mediated charge transport requires conformational motion of the DNA bases: elimination of charge transport in rigid glasses at 77 K

We have proposed that DNA-mediated charge transport (CT) is gated by base motions, with only certain base conformations being CT-active; a CT-active conformation can be described as a domain, a transiently extended pi-orbital defined dynamically by DNA sequence. Here, to explore these CT-active conformations, we examine the yield of base-base CT between photoexcited 2-aminopurine (Ap*) and guanine in DNA in rigid LiCl glasses at 77 K, where conformational rearrangement is effectively eliminated. Duplex DNA assemblies (35-mers) were constructed containing adenine bridges Ap(A)nG (n = 0-4). The yield of CT was monitored through fluorescence quenching of Ap* by G. We find, first, that the emission intensity of Ap* in all DNA duplexes increases dramatically upon cooling and becomes comparable to free Ap*. This indicates that all quenching of Ap* in duplex DNA is a dynamic process that requires conformational motion of the DNA bases. Second, DNA-mediated CT between Ap* and G is not observed at 77 K; rather than hindering the ability of DNA to transport charge, conformational motion is required. Moreover, the lack of DNA-mediated CT at 77 K, even through the shortest bridge, suggests that the static structures adopted upon cooling do not represent optimum CT-active conformations. These observations are consistent with our model of conformationally gated CT. Through conformational motion of the DNA bases, CT-active domains form and break-up transiently, both facilitating and limiting CT.     

Cisplatin--DNA cross-link models with an unusual type of chirality-neutral chelate amine carrier ligand, N,N dimethylpiperazine (Me2ppz): Me2ppzPt(guanosine monophosphate)2 adducts that exhibit novel properties

Most simple cis-PtA2G2 complexes that model the G-G cross-link DNA lesions caused by the clinically used anticancer drug cis-PtCl2(NH3)2 undergo large fluxional motions at a rapid rate (A2 = two amines or a diamine; G = guanine derivative). The carrier amine ligands in active compounds have NH groups, but the fundamental role of the NH groups has been obscured by the dynamic motion. To assess carrier ligand effects, we examine retro models, cis-PtA2G2 complexes, in which dynamic motion has been reduced by the incorporation of steric bulk into the carrier ligands. In this study we introduce a new approach employing the chirality-neutral chelate (CNC) ligand, Me2ppz (N,N'-dimethylpiperazine). Because they lie in the Pt coordination plane, the methyl groups of Me2ppz do not clash with the 06 of the base of G ligands in the ground state, but such clashes sterically hinder dynamic motion. NMR spectroscopy provided conclusive evidence that Me2ppzPt(GMP)2 complexes (GMP = 5'- and 3'-GMP) exist as a slowly interconverting mixture of two dominant head-to-tail (HT) conformers and a head-to-head (HH) conformer. Since the absence of carrier ligand chirality precluded using NMR methods to determine the absolute conformation of the two HT conformers, we used our recently developed CD pH jump method to establish chirality. The most abundant HT Me2ppzPt(5'-GMP)2 form had A chirality. Previously this chirality was shown to be favored by phosphate-cis G NIH hydrogen-bonding interligand interactions; such interactions also favor the HT conformers over the HH conformer. For typical carrier ligands, G O6 and phosphate interactions with the carrier ligand NH groups also favor the HT forms. These latter interactions are absent in Me2ppzPt(GMP)2 complexes, but the HT forms are still dominant. Nevertheless, we do find the first evidence for an HH form of a simple cis-PtA2G2 model with A2 lacking any NH groups. In previous studies, the absence of the HH conformer in cis-PtA2G2 complexes lacking carrier NH groups may be due to the presence of out-of-plane carrier ligand bulk. Such bulk forces both G O6-G O6 and G O6-carrier ligand clashes, thereby disfavoring the HH form. The major DNA cross-link adduct has the HH conformation. Thus, for anticancer activity, the small bulk of the NH group may be more important than the H-bonding interaction.     

Sequence-dependent bending of DNA induced by cisplatin: NMR structures of an A.T-rich 14-mer duplex

The NMR solution structure of the A.T rich DNA 14-mer duplex d(ATACATGGTACATA).d(TATGTACCATGTAT) is reported. This is compared with the NMR structure of the same duplex intrastrand cross-linked at the d(G*pG*) site by cis-(Pt(NH3)2?2+, derived from the anticancer drug cisplatin. The unmodified duplex has B-DNA geometry, but there is a large positive base-pair roll (roll angle 24 +/- 2 degrees) at the T9-A10 step on the 3' side of the central GG site. Platination of the DNA duplex causes the adjacent guanine bases to roll toward one another (roll angle 44 +/- 4 degrees), leading to an overall helix bend of 52 +/- 9 degrees. The platinum atom is displaced from the planes of the coordinated G7* and G8* by 0.8 A and 0.3 A, respectively. The minor groove opposite the platinum lesion is widened and flattened, with geometric parameters similar to those of A-form DNA. The unwinding of the helix at the platination site is 26 degrees. Platination causes the DNA duplex to bend toward the 3'-end (with respect to the G*G* strand), in contrast to G C-rich structures reported previously, which bend toward the 5'-end. This difference can be attributed to the predisposition of the A.T rich duplex toward bending in this region. Protein recognition of bent platinated G*G* lesions may therefore exhibit a strong dependence on the local DNA structure.     

2.4 A crystal structure of an oxaliplatin 1,2-d(GpG) intrastrand cross-link in a DNA dodecamer duplex

(1R,2R-Diaminocyclohexane)oxalatoplatinum(II) (oxaliplatin) is a third-generation platinum anticancer compound that produces the same type of inter- and intrastrand DNA cross-links as cisplatin. In combination with 5-fluorouracil, oxaliplatin has been recently approved in Europe, Asia, and Latin America for the treatment of metastatic colorectal cancer. We present here the crystal structure of an oxaliplatin adduct of a DNA dodecanucleotide duplex having the same sequence as that previously reported for cisplatin (Takahara, P. M.; Rosenzweig, A. C.; Frederick, C. A.; Lippard, S. J. Nature 1995, 377, 649-652). Pt-MAD data were used to solve this first X-ray structure of a platinated DNA duplex derived from an active platinum anticancer drug other than cisplatin. The overall geometry and crystal packing of the complex, refined to 2.4 A resolution, are similar to those of the cisplatin structure, despite the fact that the two molecules crystallize in different space groups. The platinum atom of the [Pt(R,R-DACH)](2+) moiety forms a 1,2-intrastrand cross-link between two adjacent guanosine residues in the sequence 5'-d(CCTCTGGTCTCC), bending the double helix by approximately 30 degrees toward the major groove. Both end-to-end and end-to-groove packing interactions occur in the crystal lattice. The latter is positioned in the minor groove opposite the platinum cross-link. A novel feature of the present structure is the presence of a hydrogen bond between the pseudoequatorial NH hydrogen atom of the (R,R)-DACH ligand and the O6 atom of the 3'-G of the platinated d(GpG) lesion. This finding provides structural evidence for the importance of chirality in mediating the interaction between oxaliplatin and duplex DNA, calibrating previously published models used to explain the reactivity of enantiomerically pure vicinal diamine platinum complexes with DNA in solution. It also provides a new kind of chiral recognition between an enantiomerically pure metal complex and the DNA double helix.     

Cisplatin Adducts on a GGG Sequence within a DNA Duplex Studied by NMR Spectroscopy and Molecular Dynamics Simulations

The antitumor drug cisplatin (cis‐[PtCl₂(NH₃)₂]) reacts with cellular DNA to form GG intrastrand adducts between adjacent guanines as predominant lesions. GGG sites have been shown to be hotspots of platination. To study the structural perturbation induced by binding of cisplatin to two adjacent guanines of a GGG trinucleotide, we examined here the decanucleotide duplex d[(G₁C₂C₃ G₆T₇‐ C₈G₉C₁₀) ? d(G₁₁C₁₂G₁₃A₁₄C₁₅C₁₆C₁₇G₁₈‐ G₁₉C₂₀)] ( dsCG*G*G ) intrastrand cross‐linked at the G* guanines by cis‐{Pt(NH₃)₂}²⁺ using NMR spectroscopy and molecular dynamics (MD) simulations. The NMR spectra of dsCG*G*G were found to be similar to those of previously characterized DNA duplexes cross‐linked by cisplatin at a pyG*G*X site (py=pyrimidine; X=C, T, A). This similarity of NMR spectra indicates that the base at the 3′‐side of the G*G*–Pt cross‐link does not affect the structure to a large extent. An unprecedented reversible isomerization between the duplex dsCG*G*G (bearing a –Pt chelate) and duplex dsGG*G*T (bearing a –Pt chelate) was observed, which yielded a 40:60 equilibrium between the two intrastrand GG–Pt cross‐links. No formation of interstrand cross‐links was observed. NMR spectroscopic data of dsCG*G*G indicated that the deoxyribose of the 5′‐G* adopts an N‐type conformation, and the cytidines C₃, C₁₅, and C₁₆ have average phase angles intermediate between S and N. The NMR spectroscopic chemical shifts of dsGG*G*T showed some fundamental differences to those of pyG*G*–platinum adducts but were in agreement with the NMR spectra reported previously for the DNA duplexes cross‐linked at an AG*G*C sequence by cisplatin or oxaliplatin. The presence of a purine instead of a pyrimidine at the 5′‐side of the G*G* cross‐link seems therefore to affect the structure of the XG* step significantly.     

New base pairing motifs. The synthesis and thermal stability of oligodeoxynucleotides containing imidazopyridopyrimidine nucleosides with the ability to form four hydrogen bonds

The synthesis and thermal stability of oligodeoxynucleotides (ODNs) containing imidazo[5',4':4,5]pyrido[2,3-d]pyrimidine nucleosides 1-4 (N(N), O(O), N(O), and O(N), respectively) with the aim of developing two sets of new base pairing motifs consisting of four hydrogen bonds (H-bonds) is described. The proposed four tricyclic nucleosides 1-4 were synthesized through the Stille coupling reaction of a 5-iodoimidazole nucleoside with an appropriate 5-stannylpyrimidine derivative, followed by an intramolecular cyclization. These nucleosides were incorporated into ODNs to investigate the H-bonding ability. When one molecule of the tricyclic nucleosides was incorporated into the center of each ODN (ODN I and II, each 17mer), no apparent specificity of base pairing was observed, and all duplexes were less stable than the duplexes containing natural G:C and A:T pairs. On the other hand, when three molecules of the tricyclic nucleosides were consecutively incorporated into the center of each ODN (ODN III and IV, each 17mer), thermal and thermodynamic stabilization of the duplexes due to the specific base pairings was observed. The melting temperature (T(m)) of the duplex containing the N(O):O(N) pairs showed the highest T(m) of 84.0 degrees C, which was 18.2 and 23.5 degrees C higher than that of the duplexes containing G:C and A:T pairs, respectively. This result implies that N(O)and O(N) form base pairs with four H-bonds when they are incorporated into ODNs. The duplex containing N(O):O(N) pairs was markedly stabilized by the assistance of the stacking ability of the imidazopyridopyrimidine bases. Thus, we developed a thermally stable new base pairing motif, which should be useful for the stabilization and regulation of a variety of DNA structures.     

Exploring the universality of unusual conformations of the 17-membered Pt(d(G*pG*)) macrochelate ring. Dependence of conformer formation on a change in bidentate carrier ligand from an sp3 to an sp2 nitrogen donor

Early studies on cis-PtA2(d(G*pG*)) (A2 = diamine or two amines, G = N7-platinated G) and cis-Pt(NH3)2(d(G*pG*)) models for the key cisplatin-DNA cross-link suggested that they exist exclusively or mainly as the HH1 conformer (HH1 = head-to-head G bases, with 1 denoting the normal direction of backbone propagation). These dynamic models are difficult to characterize. Employing carrier A2 ligands designed to slow dynamic interchange of conformers, we found two new conformers, DeltaHT (head-to-tail G* bases with a Delta chirality) and HH2 (with 2 denoting the backbone propagation direction opposite to normal). However, establishing that the non-HH1 conformations exist as an intrinsic feature of the 17-membered Pt(d(G*pG*)) ring requires exploring a range of different carrier ligands. Here we employ the planar aromatic sp(2) N-donor 5,5'-Me(2)bipy (5,5'-dimethyl-2,2'-bipyridine) ligand, having a shape very different from those of previously used nonplanar sp(3) N-donor bidentate carrier ligands, which often bear NH groups. The 5,5'-Me(2)bipy H6 and H6' protons project toward the d(G*pG*) moiety and hinder the dynamic motion of 5,5'-Me(2)bipyPt(d(G*pG*)). We again found HH1, HH2, and DeltaHT conformers with typical properties, supporting the conclusions that the new DeltaHT and HH2 conformers exist universally in dynamic cis-PtA2(d(G*pG*)) adducts, including cis-Pt(NH3)2(d(G*pG*)), and that the carrier ligand typically has little influence on the overall structure of the Pt(d(G*pG*)) macrocyclic ring of a given conformer. The sizes of the G H8 to H6/H6' NOE cross-peaks indicate little base canting in all 5,5'-Me(2)bipyPt(d(G*pG*)) conformers, suggesting that carrier-ligand NH groups favor the canting of one G base in the HH1 and HH2 conformers of typical cis-PtA2(d(G*pG*)) adducts.     

NMR studies of models having the Pt(d(GpG)) 17-membered macrocyclic ring formed in DNA by platinum anticancer drugs: Pt complexes with bulky chiral diamine ligands

The highly distorted Pt(d(G*pG*)) (G* = N7-platinated G) 17-membered macrocyclic ring formed by cisplatin anticancer drug binding to DNA alters the structure of the G*G* base pair steps, canting one base, and increases dynamic motion, complicating solution structural studies. However, the ring appears to favor the HH1 conformation (HH1 denotes head-to-head guanine bases, 1 denotes the normal direction of backbone propagation). Compared to cisplatin, analogues with NH groups in the carrier ligand replaced by bulky N-alkyl groups are more toxic and less active and form less dynamic adducts. To examine the molecular origins for the biological effects of steric bulk, we evaluate Me(4)DABPt(d(G*pG*)) models; the bulk and chirality of Me(4)DAB (N,N,N',N'-tetramethyl-2,3-diaminobutane with S,S or R,R configurations at the chelate ring carbons) impede dynamic motion and enhance the utility of NMR methods for identifying and characterizing conformers. Unlike past studies of adducts with such bulky carrier ligands, in which no HH conformer was found, the Me(4)DABPt(d(G*pG*)) adducts did form the HH1 conformer, providing compelling evidence that the sugar-phosphate backbone can impose constraints sufficient to overcome the alkyl-group steric effects. The HH1 conformer exhibits no significant canting. The (S,S)-Me(4)DABPt(d(G*pG*)) adduct has the least amount of the "normal" HH1 conformer and the greatest amount of the ΔHT1 conformer (ΔHT1 = head-to-tail G* bases with Δ chirality) ever observed (88% under some conditions). Thus, our results lead us to hypothesize that the low activity and high toxicity of analogues of cisplatin having carrier ligands with N-alkyl groups arise from the low abundance and minimal canting of the HH1 conformer and possibly from the adverse effects of an abundant ΔHT1 conformer. The new findings advance our understanding of the chemistry of the Pt(d(G*pG*)) macrocyclic ring and of the effects of carrier-ligand steric bulk on the properties of the ring.     

Fast interstrand cross-linking of cisplatin-DNA monoadducts compared with intrastrand chelation: a kinetic study using hairpin-stabilized duplex oligonucleotides

The antitumor drug cisplatin forms two kinds of guanine-guanine cross-links with DNA: intrastrand, occurring mainly at GG sites, and interstrand, formed at GC sites. The former are generally more abundant than the latter, at least in experiments with linear duplex DNA. The formation of interstrand cross-links requires partial disruption of the Watson-Crick base pairing, and one could therefore expect the cross-linking reaction to be rather slow. In contrast with this expectation, kinetic measurements reported here indicate that interstrand cross-linking is as fast as intrastrand, or even faster. We have investigated the reactions between two hairpin-stabilized DNA duplexes, containing either a d(TGCA)(2) sequence (duplex TGCA) or a d(G(1)G(2)CA)-d(TG(3)CC) sequence (duplex GGCA), and the diaqua form of cisplatin, cis-[Pt(NH(3))(2)(H(2)O)(2)](2+), in an unbuffered solution kept at pH 4.5 +/- 0.1 and 20 degrees C. Using HPLC as the analytical method, we have determined the platination (first step) and chelation (second step) rate constants for these reaction systems. Duplex TGCA, in which the two guanines are quasi-equivalent, is found to be platinated very slowly (k=0.5 +/- 0.1M(-1)s(-1)) and to form the final interstrand cross-link very rapidly (k=13 +/- 3 x 10(-3) s(-11)). For GGCA, we find that G(1) is platinated rapidly (k=32 +/- 5M(-1)s(-1)) to form a long-lived monoadduct, which is only slowly chelated (k=0.039 +/- 0.001 x 10(-3) s(-1)) by G(2) (intrastrand), while G(2) is platinated one order of magnitude more slowly than G(1) (k=2.0 +/- 0.5M(-1)s(-1)) and chelated fairly rapidly both by G(1) (intrastrand: k=0.4 +/-0.1 x 10(-3) s(-1)) and G(3) (interstrand: k=0.2 +/- 0.1 x 10(-3) s(-1)); finally, G(3) is platinated at about the same rate as G(2) (k=2.4 +/- 0.5M(-1)s(-1)) and chelated very rapidly by G(2) (interstrand: k=10 +/- 4 x 10(-3) s(-1)). These results suggest that the low occurrence of interstrand cross-links in cisplatinated DNA is due to an extremely slow initial platination of guanines involved in d(GC)(2) sequences, rather than to a slow cross-linking reaction.     

B-DNA Helix Stability in a Solvent-Free Environment

B-DNA is the most common DNA helix conformation under physiological conditions. However, when the amount of water in a DNA solution is decreased, B-to-A helix transitions have been observed. To understand what type of helix conformations exist in a solvent-free environment, a series of poly d(CG)(n) and mixed sequence DNA duplexes from 18 to 30 bp were examined with circular dichroism (CD), ESI-MS, ion mobility, and molecular dynamics. From the CD spectra, it was observed that all sequences had B-form helices in solution. However, the solvent-free results were more complex. For the poly d(CG)(n) series, the 18 bp duplex had an A-form helix conformation, both A- and B-helices were present for the 22 bp duplex, and only B-helices were observed for the 26 and 30 bp duplexes. Since these sequences were all present as B-DNA in solution, the observed solvent-free structures illustrate that smaller helices with fewer base pairs convert to A-DNA more easily than larger helices in the absence of solvent. A similar trend was observed for the mixed sequence duplexes where both an A- and B-helix were present for the 18 bp duplex, while only B-helices occur for the larger 22, 26, and 30 bp duplexes. Since the solvent-free B-helices appear at smaller sizes for the mixed sequences than for the pure d(CG)(n) duplexes, the pure d(CG)(n) duplexes have a greater A-philicity.     

Synthesis and characterization of DNA duplexes containing an N(4)C-ethyl-N(4)C interstrand cross-link

Short DNA duplexes containing an N(4)C-ethyl-N(4)C interstrand cross-link, C-C, were synthesized on controlled pore glass supports. Duplexes having two, three, or four A/T base pairs on either side of the C-C cross-link and terminating with a C(4) overhang at their 5'-ends were prepared. The cross-link was introduced using a convertible nucleoside approach. Thus, an oligonucleotide terminating at its 5'-end with O(4)-triazoyl-2'-deoxyuridine was first prepared on the support. The triazole group of support-bound oligomer was displaced by the aminoethyl group of 5'-dimethoxytrityl-3'-O-tert-butyldimethylsilyl-N(4)-(2-aminoethyl)deoxycytidine to give the cross-link. The dimethoxytrityl group was removed, and the upper and lower strands of the duplex were extended from two 5'-hydroxyl groups of the cross-link using protected nucleoside 3'-phosphoramidites. The tert-butyldimethylsilyl group of the resulting partial duplex was then removed, and the chain was extended in the 3'-direction from the resulting 3'-hydroxyl of the cross-link using protected nucleoside 5'-phosphoramidites. The cross-linked duplexes were purified by HPLC and characterized by enzymatic digestion and MALDI-TOF mass spectrometry. Duplexes with three or four A/T base pairs on either side of the C-C cross-link gave sigmoidal shaped A(260) profiles when heated, a behavior consistent with cooperative denaturation of the A/T base pairs. Each cross-linked duplex could be ligated to an acceptor duplex using T4 DNA ligase, a result that suggests that the C-C cross-link does not interfere with the ligation reaction, even when it is located only two base pairs from the site of ligation. The ability to synthesize duplexes with a defined interstrand cross-link and to incorporate these duplexes into longer pieces of DNA should enable preparation of substrates that can be used for a variety of biophysical and biochemical experiments, including studies of DNA repair.