Oxindole-based inhibitors of cyclin-dependent kinase 2 (CDK2): design, synthesis, enzymatic activities, and X-ray crystallographic analysis.

Two closely related classes of oxindole-based compounds, 1H-indole-2,3-dione 3-phenylhydrazones and 3-(anilinomethylene)-1,3-dihydro-2H-indol-2-ones, were shown to potently inhibit cyclin-dependent kinase 2 (CDK2). The initial lead compound was prepared as a homologue of the 3-benzylidene-1,3-dihydro-2H-indol-2-one class of kinase inhibitor. Crystallographic analysis of the lead compound bound to CDK2 provided the basis for analogue design. A semiautomated method of ligand docking was used to select compounds for synthesis, and a number of compounds with low nanomolar inhibitory activity versus CDK2 were identified. Enzyme binding determinants for several analogues were evaluated by X-ray crystallography. Compounds in this series inhibited CDK2 with a potency approximately 10-fold greater than that for CDK1. Members of this class of inhibitor cause an arrest of the cell cycle and have shown potential utility in the prevention of chemotherapy-induced alopecia.

Prevention of chemotherapy-induced alopecia in rats by CDK inhibitors.

Most traditional cytotoxic anticancer agents ablate the rapidly dividing epithelium of the hair follicle and induce alopecia (hair loss). Inhibition of cyclin-dependent kinase 2 (CDK2), a positive regulator of eukaryotic cell cycle progression, may represent a therapeutic strategy for prevention of chemotherapy-induced alopecia (CIA) by arresting the cell cycle and reducing the sensitivity of the epithelium to many cell cycle-active antitumor agents. Potent small-molecule inhibitors of CDK2 were developed using structure-based methods. Topical application of these compounds in a neonatal rat model of CIA reduced hair loss at the site of application in 33 to 50% of the animals. Thus, inhibition of CDK2 represents a potentially useful approach for the prevention of CIA in cancer patients.

The discovery of potent cRaf1 kinase inhibitors.

A series of benzylidene-1H-indol-2-one (oxindole) derivatives was synthesized and evaluated as cRaf-1 kinase inhibitors. The key features of the molecules were the donor/acceptor motif common to kinase inhibitors and a critical acidic phenol flanked by two substitutions. Diverse 5-position substitutions provided compounds with low nanomolar kinase enzyme inhibition and inhibited the intracellular MAPK pathway.

Melanocortin receptor binding determinants in the agouti protein.

The agouti protein plays an important role in the development of diabetes and obesity in rodents and has been shown to be a potent antagonist of melanocortin receptors. For this reason alanine-scanning mutagenesis was performed on the agouti protein carboxyl terminus to locate residues important for melanocortin receptor binding inhibition. When agouti residues Arg116 and Phe118 are changed to alanine, very large decreases in agouti affinity for melanocortin receptor 1, 3, and 4 result. Mutation of Phe117 to alanine causes a similar increase in agouti KI app at melanocortin receptor 4. Substitution of agouti residue Asp108 with alanine results in large increases in KI app for all three melanocortin receptors examined. All of these residues are conserved in the agouti-related transcript, ART, whose expression is up-regulated in animal models of obesity. The three-dimensional structure of the agouti carboxyl terminus was modeled, and residues which decrease receptor binding by a factor of > or = 15 when mutated to alanine localize to one side of the structure. These agouti variants with altered receptor selectivity may be useful in determining the role of melanocortin receptors in diabetes and obesity.

Evidence for the existence of a pseudoknot structure at the 3' terminus of the flavivirus genomic RNA.

The 3'-terminal nucleotides of the flavivirus genomic RNA form conserved secondary structures that may function as cis-acting signals for RNA replication. Here we provide evidence for the existence of a conserved pseudoknot structure at the 3' terminus of the flavivirus genomic RNA. A truncated version of the West Nile virus (WNV) 3'-terminal RNA sequence was used as the model for these studies. Circular dichroism spectra indicated the presence of a highly structured RNA conformation with a significant amount of A-form helix. Ribonuclease probing not only confirmed the presence of the predicted secondary structure, which consists of a long stem-loop (SL1) and a shorter stem-loop (SL2), but also suggested that base pairing occurs between nucleotides in the loop of SL2 and those in an internal loop strand located on the 5' side of SL1. Analysis of three mutant RNAs further supported the existence of pseudoknot interactions. UV-melting analysis of the WNV 3' model RNA showed three transitions with significant hyperchromicity at approximately 46, 62, and 79 degrees C. UV-melting analysis with either SL1 or SL2 RNA alone suggested that the 62 and 79 degree C transitions represent the unfolding of SL2 and SL1, respectively. The 46 degree C transition is most likely due to the opening of the proposed tertiary structure. A similar melting curve was obtained for another flavivirus (dengue-3 virus) 3'-terminal RNA, providing further support for the conservation of the structure among flaviviruses. Molecular modeling of the RNA indicated that a pseudoknot structure is a stereochemically and energetically reasonable model for the 3' terminus of flavivirus genomic RNA.

A nucleic acid triple helix formed by a peptide nucleic acid-DNA complex.

The crystal structure of a nucleic acid triplex reveals a helix, designated P-form, that differs from previously reported nucleic acid structures. The triplex consists of one polypurine DNA strand complexed to a polypyrimidine hairpin peptide nucleic acid (PNA) and was successfully designed to promote Watson-Crick and Hoogsteen base pairing. The P-form helix is underwound, with a base tilt similar to B-form DNA. The bases are displaced from the helix axis even more than in A-form DNA. Hydrogen bonds between the DNA backbone and the Hoogsteen PNA backbone explain the observation that polypyrimidine PNA sequences form highly stable 2:1 PNA-DNA complexes. This structure expands the number of known stable helical forms that nucleic acids can adopt.

Design and synthesis of conformationally constrained analogues of 4-(3-butoxy-4-methoxybenzyl)imidazolidin-2-one (Ro 20-1724) as potent inhibitors of cAMP-specific phosphodiesterase.

The synthesis and biological evaluation of cAMP-specific phosphodiesterase (PDE IV) inhibitors is described. The PDE IV inhibitor 4-(3-butoxy-4-methoxybenzyl)imidazolidin-2-one (Ro 20-1724, 2) was used as a template from which to design a set of rigid oxazolidinones, imidazolidinones, and pyrrolizidinones that mimic Ro 20-1724 but differ in the orientation of the carbonyl group. The endo isomer of each of these heterocycles was more potent than the exo isomer in an enzyme inhibition assay and a cellular assay, which measured TNF alpha secretion from activated human peripheral blood monocytes (HPBM). Imidazolidinone 4a inhibited human PDE IV with a Ki of 27 nM and TNF alpha secretion from HPBM with an IC50 of 290 nM. By comparison, Ro 20-1724 is significantly less active in these assays with activities of 1930 and 1800nM, respectively.

NMR solution structure of a peptide nucleic acid complexed with RNA.

Peptide nucleic acids (PNA) incorporating nucleic acid bases into an achiral polyamide backbone bind to DNA in a sequence-dependent manner. The structure of a PNA-ribonucleic acid (RNA) complex was determined with nuclear magnetic resonance methods. A hexameric PNA formed a 1:1 complex with a complementary RNA that is an antiparallel, right-handed double helix with Watson-Crick base pairing similar to the "A" form structure of RNA duplexes. The achiral PNA backbone assumed a distinct conformation upon binding that differed from previously proposed models and provides a basis for further structure-based design of antisense agents.

DAPI (4',6-diamidino-2-phenylindole) binds differently to DNA and RNA: minor-groove binding at AT sites and intercalation at AU sites.

The interaction of DAPI and propidium with RNA (polyA.polyU) and corresponding DNA (polydA.polydT) sequences has been compared by spectroscopic, kinetic, viscometric, Tm, and molecular modeling methods. Spectral changes of propidium are similar on binding to the AT and AU sequences but are significantly different for binding of DAPI. Spectral changes for DAPI with the DNA sequence are consistent with the expected groove-binding mode. All spectral changes for complexes of propidium with RNA and DNA and for DAPI with RNA, however, are consistent with an intercalation binding mode. When complexed with RNA, for example, DAPI aromatic protons signals shift significantly upfield, and the DAPI UV-visible spectrum shows significantly larger changes than when complexed with DNA. Slopes of log kd (dissociation rate constants) versus-log [Na+] plots are similar for complexes of propidium with RNA and DNA and for the DAPI-RNA complex and are in the range expected for an intercalation complex. The slope for the DAPI-DNA complex, however, is much larger and is in the range expected for a groove-binding complex. Association kinetics results also support an intercalation binding mode for the DAPI-RNA complex. The viscosity of polyA.polyU solutions increases significantly on addition of both propidium and DAPI, again in agreement with an intercalation binding mode for both molecules with RNA. Molecular modeling studies completely support the experimental findings and indicate that DAPI forms a very favorable intercalation complex with RNA. DAPI also forms a very stable complex in the minor groove of AT sequences of DNA, but the stabilizing interactions are considerably reduced in the wide, shallow minor groove of RNA. Modeling studies,thus,indicate that DAPI interaction energetics are more favorable for minor-groove binding in AT sequences but are more favorable for interaction in RNA.

The influence of reducing agent and 1,10-phenanthroline concentration on DNA cleavage by phenanthroline + copper.

Copper in the presence of excess 1,10-phenanthroline, a reducing agent, and molecular oxygen causes cleavage of DNA with a preference for T-3',5'-A-steps, particularly in TAT triplets. The active molecular species is commonly thought to be the bis-(1,10-phenanthroline)Cu(I) complex, (Phen)2Cu(I), regardless of the reducing agent type. We have found that (Phen)2Cu(I) is not the predominant copper complex when 3-mercaptopropionic acid (MPA) or 2-mercaptoethanol are used as the reducing agents, but (Phen)2Cu(I) predominates when ascorbate is used as the reducing agent. Substitution of ascorbate for thiol significantly enhances the rate of DNA cleavage by 1,10-phenanthroline + copper, without altering the sequence selectivity. We show that (Phen)2Cu(I) is the complex responsible for DNA cleavage, regardless of reducing agent, and that 1,10-phenanthroline and MPA compete for copper coordination sites. DNA cleavage in the presence of ascorbate also occurs under conditions where the mono-(1,10-phenanthroline)Cu(I) complex predominates (1:1 phenanthroline:copper ratio), but preferential cleavage was observed at a CCGG sequence and not at TAT sequences. The second phenanthroline ring of the (Phen)2Cu(I) complex appears essential for determining the T-3',5'-A sequence preferences of phenanthroline + copper when phenanthroline is in excess.

Modeling of nucleic acid complexes with cationic ligands: a specialized molecular mechanics force field and its application.

A potential energy force field designed for modeling nucleic acids and particularly their complexes with cationic ligands is presented. The force field is a modified version of that developed by Weiner, S.J., Kollman, P.A., Nguyen, D.T. and Case, D.A.,J. Comp. Chem. 7,230-252 (1986) and is based upon the use of a distance dependent dielectric constant, epsilon = 4rij, and partially neutralized phosphates to represent solvent and counterion. Changes from the Weiner et al. force field include additional atom types and modifications to van der Waals, electrostatic, hydrogen bonding and torsional parameters. Molecular modeling test cases of the force field are presented for a number of simple small molecules, as well as uracil and benzene dimerization, thymine-adenine and cytosine-guanine base pair formation, and adenosine/deoxyadenosine pseudorotation. Several DNA and RNA oligomers and DNA/RNA intercalation complexes with ethidium are also modeled with the force field. In all cases, the modeling results compare favorably with available experimental results. Additionally, conformational trends observed experimentally for nucleic acids by NMR and X-ray crystallographic techniques are reproduced. The modeling results for ethidium intercalation indicate a complex in which the favorable interactions are primarily van der Waals contacts, and in which electrostatic interactions are a relatively minor component. We feel the force field is particularly useful for molecular mechanics aided drug design, and an analysis of modeling results with respect to design of drugs which bind selectively to RNA is presented.

Noncovalent DNA binding of bis(1,10-phenanthroline)copper(I) and related compounds.

The noncovalent DNA binding of the bis(1,10-phenanthroline)copper(I) complex [(Phen)2CuI] was examined under anaerobic conditions by absorption and circular dichroism spectroscopy, and viscometry, as a function of phenanthroline concentration. Analyses according to the McGhee-von Hippel method indicated that binding exhibited both neighbor-exclusion and positive cooperativity effects, with a neighbor-exclusion parameter n approximately 2 and a cooperativity parameter omega approximately 4. The association constant for (Phen)2CuI binding decreased with increasing concentration of phenanthroline in excess over that required to stoichiometrically generate (Phen)2CuI, indicating that free phenanthroline was a weak competitive inhibitor of (Phen)2CuI binding. The maximal association constant for DNA binding of (Phen)2CuI in 0.2 M NaCl and 9.8% ethanol, extrapolated to zero concentration of excess phenanthroline, was 4.7 x 10(4) M-1 (DNA base pairs). The magnitude of the neighbor-exclusion parameter, the changes in spectral properties of (Phen)2CuI induced by DNA binding, and the increase in DNA solution viscosity upon (Phen)2CuI addition are consistent with a model for DNA binding by (Phen)2CuI involving partial intercalation of one phenanthroline ring of the complex between DNA base pairs in the minor groove as suggested previously [Veal & Rill (1989) Biochemistry 28, 3243-3250]. Viscosity measurements indicated that the mono(phenanthroline)copper(I) complex also binds to DNA by intercalation; however, no spectroscopic or viscometric evidence was found for DNA binding of free phenanthroline or the bis(2,9-dimethyl-1,10-phenanthroline)copper(I) complex. DNA binding of free phenanthroline may be cooperative and induced by prior binding of (Phen)2CuI.

Interaction of a macrocyclic bisacridine with DNA.

The binding of the macrocycle SDM to DNA was investigated by visible spectroscopy, stopped-flow kinetics, and NMR spectroscopy. SDM is composed of two 9-aminoacridines linked via the amino groups by a spermine side chain and via the 4-positions by a N,N'-[(methylthio)ethyl]succinamide side chain [Zimmerman, S. C., Lamberson, C. R., Cory, M., & Fairley, T. A. (1989) J. Am. Chem. Soc. 111, 6805-6809]. The visible spectrum of SDM bound to poly[d(A-T)]2 or poly[d(G-C)]2 is red-shifted relative to the spectrum of SDM alone and displays considerable hypochromicity. Results from titrations of SDM with polymer indicate a binding site size of three base pairs per macrocycle. The dissociation constant for SDM bound to either poly[d(A-T)]2 or poly[d(G-C)]2 is an order of magnitude lower than that for a similar bisacridine linked only by a spermine side chain. In addition, the dependence of the dissociation constant on ionic strength is significantly reduced. NMR studies of SDM complexes with poly[d(A-T)]2 or a tetramer, d(CGCG)2, show that intercalation is the mode of binding. The magnitudes of the chemical shift differences for SDM aromatic protons in the free and bound states support intercalation with the acridine ring systems essentially parallel to the long axis of the base pairs. Cross peaks from NOESY spectra of the SDM complex with d(CGCG)2 further support this mode of binding and provide information on the structure of the complex. The results are analyzed for consistency with each of three binding models: (i) bisintercalation with the two side chains in the same groove; (ii) bisintercalation according to the neighbor-exclusion principle with the two side chains in opposite grooves; and (iii) bisintercalation with two side chains in opposite grooves but with violation of the neighbor-exclusion principle. Model i is found to be unlikely on the basis of all evidence obtained, including preliminary modeling studies. Both models ii and iii can be reconciled with the experimental evidence and from a modeling standpoint are energetically feasible.

Crystal and solution structures of the oligonucleotide d(ATGCGCAT)2: a combined X-ray and NMR study.

A combined crystal-structure determination and NMR analysis of the octanucleotide d(ATGCGCAT)2 is reported. The X-ray analysis shows that the structure is A-form duplex in crystal state. The NMR study shows that in solution this sequence is B-type. The conformational results from each technique are presented in detail. The implications of these findings in terms of conformational flexibility and ligand binding are discussed.

Sequence specificity of DNA cleavage by bis(1,10-phenanthroline)copper(I): effects of single base pair transitions on the cleavage of preferred pyrimidine-purine-pyrimidine triplets.

The cleavage of DNA restriction fragments by bis(1,10-phenanthroline)copper(I) [[(OP)2CuI]+] is sequence dependent: the trimer TAT is most strongly preferred, while the trimer TGT and tetramers TAAT, TAGT, and CAGT are strongly to moderately preferred [Veal, J. M., & R. L. (1988) Biochemistry 27, 1822-1827]. [(OP)2CuI]+ cleavage of a series of oligonucleotide duplexes of the type 5'-CCCTPyPuPyCCCC-3'/3'-GGGAPuPyPuGGGG-5' (Py = pyrimidine; Pu = purine) was examined to determine the effects of purine substituents in the central triplet on specificity. The relative cleavage rates of different PyPuPy triplets in oligomers were similar to those observed for restriction fragments. The undecamer duplex containing the trimer TAT (TTATC) was most preferentially cleaved, predominantly at the central adenosine and the adjacent 3'-thymidine. Duplexes differing from TTATC by a single A.T----G.C transition in the central triplet were cleaved at significantly reduced rates relative to TTATC, the order of preference being TAT greater than TGT greater than TAC greater than CAT. By contrast, duplexes differing from TTATC by a single A.T----I.C transition were cleaved at rates similar to those for TTATC when the transition occurred at the 5'-pyrimidine or central purine [i.e., C(.I)AT and TIT]. A duplex containing the trimer TAC(.I) was cleaved at a reduced rate similar to the duplex containing TAC(.G). The guanine 2-amino group at positions 1 and 2, but not position 3, of a 5'-PyPuPy-3' trimer is therefore implicated as a strong inhibitor of DNA binding by the copper-phenanthroline complex.(ABSTRACT TRUNCATED AT 250 WORDS)

Sequence specificity of DNA cleavage by bis(1,10-phenanthroline)copper(I).

The bis(1,10-phenanthroline)copper(I) complex is a relatively simple molecule previously shown to cause DNA cleavage with a strong preference for gene control regions such as the Pribnow box. Sequence level mapping of sites of [(Phen)2CuI]+ cleavage in greater than 2000 bases in histone genes and the plasmid pUC9 showed that the specificity for control regions is related to a predominant preference for minor groove binding at TAT triplets, which were cleaved most strongly at the adenosine sugar ring. The related sequences TGT, TAAT, TAGPy, and CAGT (Py = pyrimidine) were moderately preferred, while CAT and TAC triplets, PyPuPuPu quartets, PuPuPuPy quartets, and CG-rich PyPuPuPy quartets were cleaved with low to average frequency. Polypurine and polypyrimidine sequences were cleaved with low frequency. The sequence preferences of [(Phen)2CuI]+ can be ascribed predominantly to (i) a requirement for binding in the minor groove at a pyrimidine 3'----5' step and (ii) stereoelectronic effects of the 2-amino group of guanine in the minor groove, which inhibit binding. Although the reagent appears primarily to recognize sequence features at the triplet or quartet level, lower than expected cleavage was observed for two TAT sequences adjacent to several other preferred sequences and higher than expected cleavage was observed at CAAGC sequences, suggesting that longer range sequence-dependent DNA conformational effects influence specificity in certain cases.