A new class of cancer-associated PTEN mutations defined by membrane translocation defects.

Phosphatase and tensin homolog (PTEN), which negatively regulates tumorigenic phosphatidylinositol (3,4,5)-trisphosphate (PIP3) signaling, is a commonly mutated tumor suppressor. The majority of cancer-associated PTEN mutations block its essential PIP3 phosphatase activity. However, there is a group of clinically identified PTEN mutations that maintain enzymatic activity, and it is unknown how these mutations contribute to tumor pathogenesis. Here, we show that these enzymatically competent PTEN mutants fail to translocate to the plasma membrane where PTEN converts PIP3 to PI(4,5)P2. Artificial membrane tethering of the PTEN mutants effectively restores tumor suppressor activity and represses excess PIP3 signaling in cells. Thus, our findings reveal a novel mechanism of tumorigenic PTEN deficiency.

The role of PTEN signaling perturbations in cancer and in targeted therapy.

The PTEN tumor suppressor was discovered by its homozygous deletion and other mutations in cancer. Since then, PTEN has been shown to be a non-redundant, evolutionarily conserved phosphatase whose function affects diverse cellular progresses such as cell cycle progression, cell proliferation, chemotaxis, apoptosis, aging, muscle contractility, DNA damage response, angiogenesis and cell polarity. In accordance with its ability to influence multiple crucial cellular processes, PTEN has a major role in the pathogenesis of numerous diseases such as diabetes, autism and almost every cancer examined. This review will discuss the diverse ways in which PTEN signaling is modified in cancer, and how these changes correlate with and might possibly affect the action of targeted chemotherapy.

PCDH8, the human homolog of PAPC, is a candidate tumor suppressor of breast cancer.

Carcinoma is an altered state of tissue differentiation in which epithelial cells no longer respond to cues that keep them in their proper position. A break down in these cues has disastrous consequences not only in cancer but also in embryonic development when cells of various lineages must organize into discrete entities to form a body plan. Paraxial protocadherin (PAPC) is an adhesion protein with six cadherin repeats that organizes the formation and polarity of developing cellular structures in frog, fish and mouse embryos. Here we show that protocadherin-8 (PCDH8), the human ortholog of PAPC, is inactivated through either mutation or epigenetic silencing in a high fraction of breast carcinomas. Loss of PCDH8 expression is associated with loss of heterozygosity, partial promoter methylation, and increased proliferation. Complementation of mutant tumor cell line HCC2218 with wild-type PCDH8 inhibited its growth. Two tumor mutants, E146K and R343H, were defective for inhibition of cell growth and migration. Surprisingly, the E146K mutant transformed the human mammary epithelial cell line MCF10A and sustained the expression of cyclin D1 and MYC without epidermal growth factor. We propose that loss of PCDH8 promotes oncogenesis in epithelial human cancers by disrupting cell-cell communication dedicated to tissue organization and repression of mitogenic signaling.

The three-dimensional structure of the 4:1 mithramycin:d(ACCCGGGT)(2) complex: evidence for an interaction between the E saccharides.

Mithramycin and chromomycin, two antitumor drugs, each having an identical aglycone and nearly identical disaccharide and trisaccharide side chains, have differing binding properties to a small oligonucleotide, d(ACCCGGGT)(2) (M. A. Keniry et al., Journal of Molecular Biology, 1993, Vol. 231, pp. 753-767). In order to understand the forces that induce four mithramycin molecules to bind to d(ACCCGGGT)(2) instead of two drug molecules in the case of chromomycin, the structure of the 4:2:1 mithramycin: Mg(2+):d(ACCCGGGT)(2) complex was investigated by (1)H-nmr and restrained molecular dynamics. The resulting three-dimensional model showed that in order to accommodate the close approach of one neighboring mithramycin dimer, the inwardly directed CDE saccharide chain of the neighboring mithramycin dimer undergoes a conformational change such that the E saccharide no longer spans the minor groove but reorients so that the hydrophilic face of the E saccharides from the two dimers oppose each other. Two hydrogen bonds are formed between the hydroxyl groups of the two opposing E saccharide groups. The results are interpreted in terms of the differences in stereochemistry and functional group substitutions between mithramycin and chromomycin. A mithramycin dimer is able to self-associate on an oligonucleotide template because it has two hydroxyl groups on the same face of its terminal E saccharide. A chromomycin dimer is unable to self-associate because one of these hydroxyl groups is acetylated and the neighboring hydroxyl group has a stereochemistry that cannot permit close contact of the hydroxyl group with a neighbouring chromomycin dimer.

NMR solution structure of the theta subunit of DNA polymerase III from Escherichia coli.

The catalytic core of Escherichia coli DNA polymerase III contains three tightly associated subunits (alpha, epsilon, and theta). The theta subunit is the smallest, but the least understood of the three. As a first step in a program aimed at understanding its function, the structure of the theta subunit has been determined by triple-resonance multidimensional NMR spectroscopy. Although only a small protein, theta was difficult to assign fully because approximately one-third of the protein is unstructured, and some sections of the remaining structured parts undergo intermediate intramolecular exchange. The secondary structure was deduced from the characteristic nuclear Overhauser effect patterns, the 3J(HN alpha) coupling constants and the consensus chemical shift index. The C-terminal third of the protein, which has many charged and hydrophilic amino acid residues, has no well-defined secondary structure and exists in a highly dynamic state. The N-terminal two-thirds has three helical segments (Gln10-Asp19, Glu38-Glu43, and His47-Glu54), one short extended segment (Pro34-Ala37), and a long loop (Ala20-Glu29), of which part may undergo intermediate conformational exchange. Solution of the three-dimensional structure by NMR techniques revealed that the helices fold in such a way that the surface of theta is bipolar, with one face of the protein containing most of the acidic residues and the other face containing most of the long chain basic residues. Preliminary chemical shift mapping experiments with a domain of the epsilon subunit have identified a loop region (Ala20-Glu29) in theta as the site of association with epsilon.

Solution structure and backbone dynamics of long-[Arg(3)]insulin-like growth factor-I.

Long-[Arg(3)]insulin-like growth factor-I (IGF-I) is a potent analog of insulin-like growth factor-I that has been modified by a Glu(3) --> Arg mutation and a 13-amino acid extension appended to the N terminus. We have determined the solution structure of (15)N-labeled Long-[Arg(3)]-IGF-I using high resolution NMR and restrained molecular dynamics techniques to a precision of 0.82 +/- 0.28 A root mean square deviation for the backbone heavy atoms in the three alpha-helices and 3.5 +/- 0.9 A root mean square deviation for all backbone heavy atoms excluding the 8 N-terminal residues and the 8 C-terminal eight residues. Overall, the structure of the IGF-I domain is consistent with earlier studies of IGF-I with some minor changes remote from the N terminus. The major variations in the structure, compared with IGF-I, occur at the N terminus with a substantial reorientation of the N-terminal three residues of the IGF-I domain. These results are interpreted in terms of the lower binding affinity for insulin-like growth factor-binding proteins. The backbone dynamics of Long-[Arg(3)]IGF-I were investigated using (15)N nuclear spin relaxation and the heteronuclear nuclear Overhauser enhancement (NOE). There is a considerable degree of flexibility in Long-[Arg(3)]IGF-I, even in the alpha-helices, as indicated by an average ((1)H)(15)N NOE of 0.55 for the regions. The largest heteronuclear NOEs are observed in the helical regions, lower heteronuclear NOEs are observed in the C-domain loop separating helix 1 from helix 2, and negative heteronuclear NOEs are observed in the N-terminal extension and at the C terminus. Despite these data indicating conformational flexibility for the N-terminal extension, slow amide proton exchange was observed for some residues in this region, suggesting some transitory structure does exist, possibly a molten helix. A certain degree of flexibility may be necessary in all insulin-like growth factors to enable association with various receptors and binding proteins.

Quadruplex structures in nucleic acids.

DNA oligonucleotides that have repetitive tracts of guanine bases can form G-quadruplex structures that display an amazing polymorphism. Structures of several new G-quadruplexes have been solved recently that greatly expand the known structural motifs observed in nucleic acid quadruplexes. Base triads, base hexads, and quartets that contain cytosine have recently been identified stacked over the familiar G-quartets. The current status of the diverse array of structural features in quadruplexes is described and used to provide insight into the polymorphism and folding pathways. This review also summarizes recent progress in the techniques used to probe the structures of G-quadruplexes and discusses the role of ion binding in quadruplex formation. Several of the quadruplex structures featured in this review can be accessed in the online version of this review as CHIME representations.

The NMR solution structure and characterization of pH dependent chemical shifts of the beta-elicitin, cryptogein.

The NMR structure of the 98 residue beta-elicitin, cryptogein, which induces a defence response in tobacco, was determined using 15N and 13C/15N labelled protein samples. In aqueous solution conditions in the millimolar range, the protein forms a discrete homodimer where the N-terminal helices of each monomer form an interface. The structure was calculated with 1047 intrasubunit and 40 intersubunit NOE derived distance constraints and 236 dihedral angle constraints for each subunit using the molecular dynamics program DYANA. The twenty best conformers were energy-minimized in OPAL to give a root-mean-square deviation to the mean structure of 0.82 A for the backbone atoms and 1.03 A for all heavy atoms. The monomeric structure is nearly identical to the recently derived X-ray crystal structure (backbone rmsd 0.86 A for residues 2 to 97) and shows five helices, a two stranded antiparallel beta-sheet and an omega-loop. Using 1H,15N HSQC spectroscopy the pKa of the N- and C-termini, Tyr12, Asp21, Asp30, Asp72, and Tyr85 were determined and support the proposal of several stabilizing ionic interactions including a salt bridge between Asp21 and Lys62. The hydroxyl hydrogens of Tyr33 and Ser78 are clearly observed indicating that these residues are buried and hydrogen bonded. Two other tyrosines, Tyr47 and Tyr87, show pKa's > 12, however, there is no indication that their hydroxyls are hydrogen bonded. Calculations of theoretical pKa's show general agreement with the experimentally determined values and are similar for both the crystal and solution structures.

NMR structure refinement and dynamics of the K+-[d(G3T4G3)]2 quadruplex via particle mesh Ewald molecular dynamics simulations.

The solution structure and dynamical properties of the potassium-stabilized, hairpin dimer quadruplex formed by the oligonucleotide d(G3T4G3) have been elucidated by a combination of high-resolution NMR and molecular dynamics simulations. Refinement calculations were carried out both in vacuo, without internally coordinated K+ cations, and in explicit water, with internally coordinated K+ cations. In the latter case, the electrostatic interactions were calculated using the particle mesh Ewald (PME) method. The NMR restraints indicate that the K+ quadruplex has a folding arrangement similar to that formed by the same oligonucleotide in the presence of sodium, but with significant local differences. Unlike the Na+ quadruplex, the thymine loops found in K+ exhibit considerable flexibility, and appear to interconvert between two preferred conformations. Furthermore, the NMR evidence points toward K+-stabilized guanine quartets of slightly larger diameter relative to the Na+-stabilized structure. The characteristics of the quartet stem are greatly affected by the modeling technique employed: caged cations alter the size and symmetry of the quartets, and explicit water molecules form hydration spines within the grooves. These results provide insight into those factors that determine the overall stability of hairpin dimer quadruplexes and the effects of different cations in modulating the relative stability of the dimeric hairpin and linear, four-stranded, quadruplex forms.

The contribution of thymine-thymine interactions to the stability of folded dimeric quadruplexes.

The loop of four thymines in the sodium form of the dimeric folded quadruplex [d(G3T4G3)]2 assumes a well-defined structure in which hydrogen bonding between the thymine bases appears to contribute to the stability and final conformation of the quadruplex. We have investigated the importance of the loop interactions by systematically replacing each thymine in the loop with a cytosine. The quadruplexes formed by d(G3CT3G3), d(G3TCT2G3), d(G3T2CTG3) and d(G3T3CG3) in the presence of 150 mM Na+ were studied by gel mobility, circular dichroism and 1H NMR spectroscopy. The major species formed by d(G3CT3G3), d(G3TCT2G3) and d(G3T3CG3) at 1 mM strand concentration at neutral pH is a dimeric folded quadruplex. d(G3T2CTG3) has anomalous behaviour and associates into a greater percentage of linear four-stranded quadruplex than the other three oligonucleotides at neutral pH and at the same concentration. The linear four-stranded quadruplex has a greater tendency to oligomerize to larger ill-defined structures, as demonstrated by broad 1H NMR resonances. At pH 4, when the cytosine is protonated, there is a greater tendency for each of the oligonucleotides to form some four-stranded linear quadruplex, except for d(G3T2CTG3), which has the reverse tendency. The experimental results are discussed in terms of hydrogen bonding within the thymine loop.

Secondary structure determination of 15N-labelled human Long-[Arg-3]-insulin-like growth factor 1 by multidimensional NMR spectroscopy.

Insulin-like growth factors (IGFs) are a group of proteins that promote cell growth and differentiation. Long-[Arg-3]-IGF-I (Francis et al. (1992) J. Mol. Endocrinol. 8, 213-223), a potent analogue of IGF-I, which has a Glu-3 to Arg-3 substitution and a hydrophobic, thirteen amino acid N-terminal extension, has been studied by 1H,15N NMR spectroscopy. All the backbone 1H and 15N assignments and most of the 1H sidechain assignments have been completed. The secondary structure elements were identified by determining the sequential and medium range NOEs from sensitivity-enhanced 15N-NOESY-HSQC and sensitivity-enhanced 15N-HSQC-NOESY-HSQC spectra. The IGF-I domain of Long-[Arg-3]-IGF-I was found to have an almost identical structure to IGF-I. The N-terminal seven amino acid residues of the extension have very few medium range or long range NOEs but the next five amino acids form a turn-like structure that is spatially close to the beginning of helix 1 in the IGF-I domain. Hydrogen-deuterium exchange experiments show that all the slowly exchanging backbone amide protons in the IGF-I domain are either in the helical or the extended structural elements. Many of the amide protons in the N-terminal extension are also protected from the solvent although the residues in this part of the extension do not have any identifiable secondary structure. The results are interpreted in terms of the increased biological potency of Long-[Arg-3]-IGF-I and the decreased binding to insulin-like growth factor binding proteins.

Solution structure of the Na+ form of the dimeric guanine quadruplex [d(G3T4G3)]2.

The solution structure of the DNA quadruplex formed by the association of two strands of the DNA oligonucleotide, d(G3T4G3), in NaCl solution has been determined by 1H two-dimensional NMR techniques, full relaxation matrix calculations and restrained molecular dynamics. The refined structure incorporates the sequences 5'-G1sG2AG3AT4AT5AT6AT7AG8sG9AG10A-3' and 5'-G11sG12AG13AT14AT15AT16AT17AG18sG19sG20A-3' (where S and A denote syn and anti, respectively) in a three-quartet, diagonal-looped structure that we [Strahan, G. D., Shafer, R. H. & Keniry, M. A. (1994) Nucleic Acids Res. 22, 5447-5455] and others [Smith, F. W., Lau, F. W. & Feigon, J. (1994) Proc. Natl. Acad. Sci. USA 91, 10546-10550] have described. The loop structure is compact and incorporates many of the features found in duplex hairpin loops including base stacking, intraloop hydrogen bonding and extensive van der Waals' interactions. The first and third loop thymines stack over the outermost G-quartet and are also associated by hydrogen bonding. The second and the fourth loop thymines fold inwards in order to enhance van der Waals' interactions. The unexpected sequential syn-syn deoxyguanosines in the quadruplex stem appear to be a direct consequence of the way DNA oligonucleotides fold and the subsequent search for the most stable loop structure. The implications of loop sequence and length on the structure of quadruplexes are discussed.

NMR studies of drug-DNA complexes.

The application of high-resolution, multidimensional NMR techniques to the problem of determining the structure of drug-DNA complexes in solution has led to substantial progress in understanding the effect of drugs on DNA at the molecular level. With the development of isotopic labeling methods applied in three- and four-dimensional experiments, we anticipate that more complex drug-DNA systems will become amenable to structural analysis. In addition to implementing these newer techniques, progress will also be made in terms of investigating the structure of drug complexes with more unusual forms of DNA, such as triplexes, quadruplexes, multistranded junctions, and so forth.

Structural properties of the [d(G3T4G3)]2 quadruplex: evidence for sequential syn-syn deoxyguanosines.

Two-dimensional 1H NMR studies on the dimeric hairpin quadruplex formed by d(G3T4G3) in the presence of either NaCl or KCl are presented. In the presence of either salt, the quadruplex structure is characterized by half the guanine nucleosides in the syn conformation about the glycosidic bond, the other half in the anti conformation, as reported for other similar sequences. However, 1H NOESY and 1H-31P heteronuclear correlation experiments demonstrate that the deoxyguanosines do not strictly alternate between syn and anti along individual strands. Thus we find the following sequences with regard to glycosidic bond conformation: 5'-G1SG2SG3AT4AT5A-T6AT7AG8SG9AG10A-3' and 5'-G11SG12AG13AT14AT1 5AT16AT17AG18SG19SG20A-3', where S and A denote syn and anti, respectively. This represents the first experimental evidence for a nucleic acid structure containing two sequential nucleosides in the syn conformation. The stacking interactions of the resulting quadruplex quartets and their component bases have been evaluated using unrestrained molecular dynamics calculations and energy component analysis. These calculations suggest that the sequential syn-syn/anti-anti and syn-anti quartet stacks are almost equal in energy, whereas the anti-syn stack, which is not present in our structure, is energetically less favorable by about 4 kcal/mol. Possible reasons for this energy difference and its implications for the stability of quadruplex structures are discussed.

Polycytosine regions contained in DNA hairpin loops interact via a four-stranded, parallel structure similar to the i-motif.

Thermal denaturation profiles of an oligodeoxynucleotide that forms a hairpin structure with a cytidine-rich loop show an unexpected transition at 60 degrees C at pH 5.0 but not at pH 8.0. Analytical ultracentrifugation shows that this transition reflects dimer formation via the interaction of loops from two molecules to form a novel structure termed the h-dimer. The dependence of this structure on low pH implies the formation of cytosine-protonated cytosine base pairs. NMR spectroscopy, thermal denaturation and ultraviolet absorption spectral analysis suggest a similarity to the i-motif structure recently proposed for the interaction of deoxycytidine oligomers. The use of hairpin loops to form i-motif-like structures may prove useful in searches for cognate proteins and possibly in the production of antibodies.

The solution conformation of a trisdecanucleotide containing the consensus binding site of the dnaA initiation protein.

The solution structure of a trisdecanucleotide, d(CCTGTGGATAACA).d(TGTTATCCACAGG) containing the consensus binding site of the dnaA initiation protein has been determined by two-dimensional NMR techniques and restrained molecular dynamics calculations. Interproton distances were obtained by an iterative complete relaxation matrix algorithm, MARDIGRAS. During molecular dynamics runs, the backbone was restricted with the assistance of experimentally derived distance constraints. A family of refined structures with small pairwise root-mean-square deviation values (approximately 0.08 nm) was obtained. All but one of the pyrimidines were found to adopt the C1'-exo conformation while the purines were found to adopt the C2'-endo or C1'-exo conformation. The six-membered rings of the purines were found to stack over the six-membered rings of the pyrimidines while there is virtually no overlap of the pyrimidines over the purines. 5'-purine-purine-3' and 5'-pyrimidine-pyrimidine-3' stacking resembles the observed stacking of these bases in other NMR and X-ray structures of oligonucleotides. The final refined structure exhibited a small curvature and was slightly longer than canonical B-DNA. The variation of twist angle, proposed as a recognition element for proteins, exhibited symmetry about the centre of the consensus binding site.

Nuclear magnetic resonance comparison of the binding sites of mithramycin and chromomycin on the self-complementary oligonucleotide d(ACCCGGGT)2. Evidence that the saccharide chains have a role in sequence specificity.

A comprehensive two-dimensional 1H nuclear magnetic resonance spectral analysis of the ternary 4:2:1 mithramycin-Mg2+-d(A1C2C3C4G5G6G7T8)2 complex and the ternary 2:1:1 chromomycin-Mg(2+)-d(A21C2C3C4C4G5G6G7T8)2 complex is presented. The self-complementary oligonucleotide is found to bind two dimers of mithramycin in two identical off-center binding sites such that the twofold symmetry of the oligonucleotide is retained. In contrast, the same oligonucleotide binds only one dimer of chromomycin in a single but distinct off-center binding site. Two-dimensional nuclear Overhauser spectroscopy experiments show that the aglycone binding site of the drug dimer in each complex extends over almost four base-pairs and is similar in length to other complexes between chromomycin or mithramycin and oligonucleotides. The data demonstrate that the chromomycin dimer binding site is offset by one base-pair step from the dimer binding site in the mithramycin complex. This preferred binding site prevents two dimers of chromomycin binding to d(ACCCGGGT)2 for steric reasons and lends further support to previous work that showed the 5'-CG base-pair site is less favored by these drugs compared to the 5'GC and 5'-GG,5'-CC sites. Evidence is presented that suggests mithramycin may occupy either of two distinct binding sites on d(ACCCGGGT)2 when the drug concentration is not saturating. The nuclear magnetic resonance data demonstrate that the saccharide chains of this family of drugs do have a role in determining the binding site on nucleotides and as a consequence the CDE trisaccharide chain may alter its conformation to fulfil this role. Titration of mithramycin up to a drug-duplex ratio of 7:1 reveals further association of mithramycin with the complex but no new drug-oligonucleotide nuclear Overhauser enhancement contacts.

NMR investigation of the interaction of mithramycin A with d(ACCCGGGT)2.

The binding of mithramycin A to d(ACCCGGGT)2 has been investigated by one- and two-dimensional 1H NMR spectroscopy. Titration of the drug into the octamer solution results in loss of the oligonucleotide C2 symmetry at stoichiometric ratios less than 4 drug molecules per duplex. However, at a ratio of 4:1 (drug/duplex), the C2 symmetry of the oligonucleotide is restored. From these data it is evident that more than one complex forms at ratios less than 4:1 while only one complex predominates at the ratio 4:1. This is the first report of a DNA octamer which binds 4 large drug molecules. These results are compared to those we have recently reported for mithramycin binding to d(ATGCAT)2, where only a single, bound complex is observed, with a stoichiometry of 2:1.

NMR investigation of mithramycin A binding to d(ATGCAT)2: a comparative study with chromomycin A3.

The binding of mithramycin A to the d(A1T2G3C4A5T6) duplex was investigated by 1H NMR and found to be similar to that of its analogue chromomycin A3. In the presence of Mg2+, mithramycin binds strongly to d(ATGCAT)2. On the basis of the two-dimensional NOESY spectrum, the complex formed possesses C2 symmetry at a stoichiometry of two drugs per duplex (2:1) and is in slow chemical exchange on the NMR time scale. NOESY experiments reveal contacts from the E-pyranose of mithramycin to the terminal and nonterminal adenine H2 proton of DNA and from the drug hydroxyl proton to both G3NH2 protons, C4H1' proton, and A5H1' proton. These data place the drug chromophore and E pyranose on the minor groove side of d(ATGCAT)2. NOE contacts from the A-, B-, C-, and D-pyranoses of mithramycin to several deoxyribose protons suggest that the A- and B-rings are oriented along the sugar-phosphate backbone of G3-C4, while the C- and D-rings are located along the sugar-phosphate backbone of A5-T6. These drug-DNA contacts are very similar to those found for chromomycin binding to d(ATGCAT)2. Unlike chromomycin, the NOESY spectrum of mithramycin at the molar ratio of one drug per duplex reveals several chemical exchange cross-peaks corresponding to the drug-free and drug-bound proton resonances. From the intensity of these cross-peaks and the corresponding diagonal peaks, the off-rate constant was estimated to be 0.4 s-1. These data suggest that the exchange rate of mithramycin binding to d(ATGCAT)2 is faster than that of chromomycin.

NMR studies of the interaction of chromomycin A3 with small DNA duplexes. Binding to GC-containing sequences.

The interaction of chromomycin A3 with the oligodeoxyribonucleotides 1, d(ATGCAT), 2, d(ATCGAT), 3, d(TATGCATA), and 4, d(ATAGCTAT), has been investigated by 1H and 31P NMR. In the presence of Mg2+, chromomycin binds strongly to the three GC-containing oligomers 1, 3, and 4 but not to the CG-containing oligomer 2. The proton chemical shift changes for 1 and 3 are similar, and these DNA duplexes appear to bind with a stoichiometry of 2 drugs:1 Mg2+:1 duplex. The same stoichiometry of 2 drugs:1 duplex is confirmed with 4; however, proton chemical shift changes differ. An overall C2 symmetry is exhibited by the drug complex with 1, 3, and 4. At a molar ratio of 2.0 (drugs:duplex), no free DNA proton NMR signals remain. Two-dimensional nuclear Overhauser exchange spectroscopy (NOESY) of the saturated chromomycin complex with 1 and 3 positions both chromomycinone hydroxyls and the E carbohydrates in the minor groove and provides evidence suggesting that the B carbohydrates lie on the major-groove side. This is supported by several dipolar coupling cross-peaks between the drug and the DNA duplex. Drug-induced conformational changes in duplex 1 are evaluated over a range of NOESY mixing times and found to possess some characteristics of both B-DNA and A-DNA, where the minor groove is wider and shallower. A widening of the minor groove is essential for the DNA duplex to accommodate two drug molecules. This current minor-groove model is a substantial revision of our earlier major-groove model [Keniry, M.A., Brown, S.C., Berman, E., & Shafer, R.H. (1987) Biochemistry 26, 1058-1067] and is in agreement with the model recently proposed by Gao and Patel [Gao, X., & Patel, D. J. (1989a) Biochemistry 28, 751-762].