29 Jul 2005

The cyclin-dependent kinase Cdk5 is a popular target for drug developers aiming to treat a number of neurodegenerative disorders, including Parkinson (see ARF related news story) and Alzheimer (AD) diseases. In AD, the kinase has been implicated in the phosphorylation of the protein tau (see ARF related news story), which causes neurofibrillary tangles, one of the hallmarks of the disease. But Cdk5’s action on widely ranging substrates in various diseases presents a challenge—how does one hit the pathological pathways without messing up the physiological ones? And is there hope of zeroing in on Cdk5 while avoiding the dozens of other kinases in the cell with similar-looking active sites?

These questions are best answered by structure-based drug design, a process of analyzing and adjusting the molecular fit between enzyme and inhibitor to improve potency and specificity. In the July issue of Chemistry and Biology, Ken Kosik (formerly of Harvard Medical School and now at University of California in Santa Barbara) and colleagues provide fodder for the design process with their discovery of three new small molecule inhibitors that block Cdk5 phosphorylation of tau. Their x-ray crystal structures of inhibitor-enzyme complexes show that two of the compounds, like most kinase inhibitors, occupy Cdk5’s adenosine triphosphate (ATP) pocket, while the third molecule is an unusual, tau-selective inhibitor that finds a completely different home elsewhere on the kinase. Novel features of all three inhibitors suggest new approaches to solving the problems of potency and selectivity of Cdk5 inhibitors.

To search for new Cdk5 inhibitors, Kosik and colleagues, working with collaborators from MIT and the European Institute of Oncology in Milan, tested roughly 58,000 small molecules in a high-throughput screen of tau phosphorylation. Lead author Jae Suk Ahn and colleagues then characterized the top three hits kinetically (all effective inhibitors when present at μM concentrations) and structurally.

Suk Ahn found that like most kinase inhibitors, two of the compounds competed with ATP for binding to Cdk5. Structural analysis of crystals formed upon incubating the inhibitors with Cdk5 and p25 also revealed that these compounds bound in the ATP pocket; p25, a binding partner and activator of Cdk5, accumulates in the AD brain and could be a key player in AD pathology (see ARF related news story). But one inhibitor, a natural product called bellidin, made a total of eight hydrogen bonds with residues in the pocket, compared to only four usually seen for these kinds of inhibitors. The second compound, an aminothiazole, caused an unusual conformational change around residues in the pocket that has not been seen in previously characterized inhibitors. A computational analysis comparing the two new compounds to three other Cdk5 inhibitors (aloisine-A, indirubin-3’-oxime and roscovitine) exposed an additional feature—an empty bulge in the ATP binding pocket. Any of these features could be exploited to build potency and specificity into future inhibitors.

The third new compound may be the most interesting—it was competitive with tau and displayed a modest specificity for blocking tau phosphorylation over other substrates. For example, the compound didn’t block Cdk5 phosphorylation of histone H1, and it was 16 times better at inhibiting phosphorylation of tau than that of the well-characterized Cdk5 substrate FAK (IC50s for FAK and tau were 272 μM, and 17 μM, respectively). Unfortunately, the crystal structure of this inhibitor bound to Cdk5 did not show its location. What did show up was an empty ATP binding pocket and a rearranged activation loop, that part of the molecule where p25 binds. The data suggest that the enzyme had been bent into an inactive conformation by the inhibitor, which bound somewhere outside of the ATP pocket. This is the first description of a Cdk5 inhibitor that hits the enzyme away from the catalytic site—an exciting prospect because such inhibitors could have the advantages of being substrate-specific as well as more selective for Cdk5 over other kinases.—Pat McCaffrey