Sequencing of the human genome has so far revealed more than 500 genes encoding protein kinases. Many of these enzymes are directly involved in diseases such as cancer and inflammation, making them excellent targets for drug development. In the March issue of Nature Biotechnology, Lockhart, Zarrinkar and colleagues go into uncharted protein kinase territory by developing a new type of assay to determine the specificity of a number of kinase inhibitors against a panel of 119 protein kinases.

Kinase inhibitors are an important new class of anticancer drugs, and have clinical activity in tumours in which the target kinase is activated by mutation, such as the mutant kinase BCR–ABL in chronic myeloid leukaemia (CML). The success of small-molecule inhibitors such as imatinib (Gleevec; Novartis) to treat CML, and gefitinib (Iressa; Astra-Zeneca) and erlotinib (Tarceva; Genentech/OSIP), both of which target the epidermal growth factor receptor (EGFR) to treat lung cancer, has demonstrated clear proof-of-principle that this strategy is effective.

Most kinase inhibitors in clinical development target the ATP-binding site common to all kinases, and bind multiple kinases. However, it is not possible to predict binding specificity and affinity on the basis of available sequence or structural information. Conventional profiling methods that use the measurement of in vitro activity are limited by the difficulty of building and running large numbers of kinase activity assays. Such information is very valuable, both for finding new clinical uses for inhibitors and for predicting or explaining toxicity.

Fabian et al. have developed a quantitative competitive assay for measuring the binding of small molecules to the ATP-binding site of kinases. Human kinase domains are first expressed as a fusion attached to T7 bacteriophage capsid protein. Then, the test compound in solution competes with an immobilized 'bait' ligand to bind the phage expressing the kinase domain. The amount of phage bound to the bait is quantified to determine the affinity of the test compound for each kinase. If the free test compound binds the kinase and directly or indirectly blocks the ATP site, fewer protein molecules bind to the immobilized bait; if the free test compound does not bind to the kinase, the fusion proteins are able to bind to the immobilized bait.

For most of the test compounds, the tightest interaction is with the kinase that the drug was optimized to inhibit, but the difference in affinity between the primary target and other kinases varies substantially. For the inhibitors BIRB-796, VX-745, erlotinib, GW-2016 and SU11248 there is at least a tenfold difference in affinity between the intended target and off-targets, whereas for SP600125, EKB-569 and ZD-6474 there is less than a twofold difference. This indicates that optimization efforts are generally successful, but there is room for improved discrimination if necessary.

A number of CML patients develop resistance to imatinib. The authors developed assays for imatinib-resistant ABL kinases. The most dramatic finding is that the p38 inhibitor BIRB-796 binds tightly to an imatinib-resistant ABL mutant. The authors also uncovered a new target for imatinib, the SRC-family tyrosine kinase LCK, to which imatinib binds tightly. Finally, the authors showed that clinically observed mutations in EGFR do not affect the binding affinity of gefitinib or erlotinib.

The authors show that their kinase-binding assay provides results consistent with more standard in vitro results, although whether this reflects the capability of a compound to inhibit a kinase in a cell remains to be determined. Certainly, this assay is a useful tool and it will probably accelerate drug discovery and development efforts for kinase inhibitors.