Discovery of N-phenyl-4-(thiazol-5-yl)pyrimidin-2-amine aurora kinase inhibitors.

Through cell-based screening of our kinase-directed compound collection, we discovered that a subset of N-phenyl-4-(thiazol-5-yl)pyrimidin-2-amines were potent cytotoxic agents against cancer cell lines, suppressed mitotic histone H3 phosphorylation, and caused aberrant mitotic phenotypes. It was subsequently established that these compounds were in fact potent inhibitors of aurora A and B kinases. It was shown that potency and selectivity of aurora kinase inhibition correlated with the presence of a substituent at the aniline para-position in these compounds. The anticancer effects of lead compound 4-methyl-5-(2-(4-morpholinophenylamino)pyrimidin-4-yl)thiazol-2-amine (18; K(i) values of 8.0 and 9.2 nM for aurora A and B, respectively) were shown to emanate from cell death following mitotic failure and increased polyploidy as a consequence of cellular inhibition of aurora A and B kinases. Preliminary in vivo assessment showed that compound 18 was orally bioavailable and possessed anticancer activity. Compound 18 (CYC116) is currently undergoing phase I clinical evaluation in cancer patients.

Truncation and optimisation of peptide inhibitors of cyclin-dependent kinase 2-cyclin a through structure-guided design.

Peptides that inhibit cyclin-dependent kinase 2 by blocking the macromolecular substrate recruitment site of cyclin A were simplified, for example, by replacement of dipeptide units with beta-amino acids. The smallest inhibitor retaining activity was a tripeptide, whose binding mode was confirmed by X-ray crystallography. This result suggests that nonpeptidic cyclin groove inhibitors may be feasible therapeutic agents.The cyclin-dependent kinase 2-cyclin A complex is an important regulator of the DNA-synthesis phase of the mammalian cell cycle, which is frequently deregulated in cancer. Rather than blocking the ATP-binding site of the apparently redundant kinase subunit, targeting the binding site for macromolecular substrates and regulatory proteins of cyclin A represents a promising strategy to enforce tumour-selective apoptosis. The cyclin-binding groove can be blocked with comparatively small synthetic peptides, which indirectly leads to inhibition of kinase function, but these peptides are metabolically labile and membrane impermeable. As part of our ongoing effort to develop more druglike peptidomimetics derived from cyclin-groove-binding peptides, we report the results of our studies aimed at a detailed understanding of the structural determinants required for effective binding. Using a combination of peptide synthesis, biochemical assays and X-ray crystallography, we show that it is possible to simplify peptide structures through the replacement of dipeptide units in which one of the residues is not directly involved in binding, through the introduction of beta-amino acid residues that retain only the dipeptide residue side chain that is important for binding. This approach also allowed us to probe spatial constraints in general, as well as the importance of peptide backbone hydrogen-bonding functions. Our identification of potent beta-homoleucine-containing tetrapeptide inhibitors, as well as the finding that an optimised N-terminally acetylated tripeptide retains some cyclin A-binding affinity, suggest that the pharmacological targeting of the cyclin A binding groove may be feasible.

REPLACE: a strategy for iterative design of cyclin-binding groove inhibitors.

We describe a drug-design strategy termed REPLACE (REplacement with Partial Ligand Alternatives through Computational Enrichment) in which nonpeptidic surrogates for specific determinants of known peptide ligands are identified in silico by using a core peptide-bound protein structure as a design anchor. In the REPLACE application example, we present the effective replacement of two critical binding motifs in a lead protein-protein interaction inhibitor pentapeptide with more druglike phenyltriazole and diphenyl ether groups. These were identified through docking of fragment libraries into the volume of the cyclin-binding groove of CDK2/cyclin A vacated through truncation of the inhibitor peptide-binding determinants. Proof of concept for this strategy was obtained through the generation of potent peptide-small-molecule hybrids and by the confirmation of inhibitor-binding modes in X-ray crystal structures. This method therefore allows nonpeptide fragments to be identified without the requirement for a high-sensitivity binding assay and should be generally applicable in replacing amino acids as individual residues or groups in peptide inhibitors to generate pharmaceutically acceptable lead molecules.

Design, synthesis, biological activity and structural analysis of cyclic peptide inhibitors targeting the substrate recruitment site of cyclin-dependent kinase complexes.

Inhibition of cyclin A- and cyclin E-associated cyclin-dependent kinase-2 (CDK2) activities is an effective way of selective induction of apoptotic cell death via the E2F pathway in tumour cells. The cyclin groove recognition motif (CRM) in the natural CDK-inhibitory (CDKI) tumour suppressor protein p27KIP1 was used as the basis for the design and synthesis of a series of cyclic peptides whose biological activity and structural characterisation by NMR and X-ray crystallography is reported. Whereas linear p27KIP1 sequence peptides were comparatively ineffective, introduction of side chain-to-tail constraints was found to be productive. An optimal macrocyclic ring size for the conformational constraint was determined, mimicking the intramolecular H-bonding system of p27. Molecular dynamics calculations of various macrocycles suggested a close correlation between ring flexibility and biological activity. Truncated inhibitor peptide analogues also confirmed the hypothesis that introduction of a cyclic conformational constraint is favourable in terms of affinity and potency. The structural basis for the potency increase in cyclic versus linear peptides was demonstrated through the determination and interpretation of X-ray crystal structures of complexes between CDK2/cylin A (CDK2A) and a constrained pentapeptide.

Insights into cyclin groove recognition: complex crystal structures and inhibitor design through ligand exchange.

Inhibition of CDK2/CA (cyclin-dependent kinase 2/cyclin A complex) activity through blocking of the substrate recognition site in the cyclin A subunit has been demonstrated to be an effective method for inducing apoptosis in tumor cells. We have used the cyclin binding motif (CBM) present in the tumor suppressor proteins p21(WAF1) and p27(KIP1) as a template to optimize the minimal sequence necessary for CDK2/CA inhibition. A series of peptides were prepared, containing nonnatural amino acids, which possess nano- to micromolar CDK2-inhibitory activity. Here we present X-ray structures of the protein complex CDK2/CA, together with the cyclin groove-bound peptides H-Ala-Ala-Abu-Arg-Ser-Leu-Ile-(p-F-Phe)-NH(2) (peptide 1), H-Arg-Arg-Leu-Ile-Phe-NH(2) (peptide 2), Ac-Arg-Arg-Leu-Asn-(m-Cl-Phe)-NH(2) (peptide 3), H-Arg-Arg-Leu-Asn-(p-F-Phe)-NH(2) (peptide 4), and H-Cit-Cit-Leu-Ile-(p-F-Phe)-NH(2) (peptide 5). Some of the peptide complexes presented here were obtained through the novel technique of ligand exchange within protein crystals. This method may find general application for obtaining complex structures of proteins with surface-bound ligands.