No matter how good an experiment is, for most researchers there remains the nagging question of what they didn't see because there was no money, time or method to look.

Drug discovery is a case in point. While preclinical testing includes studies on drug metabolism and toxicology in addition to target validation studies, it is impossible to comprehensively predict a lead compound's effects on physiological processes, many of which are discovered during clinical trials - or after a drug has reached the market.

Those unintended effects can be bad, and then are called side effects; or they can be good, in which case they lead to off-label use or a new indication. But in either case, they are unforeseen.

Now, in the March 2005 issue of Nature Biotechnology, researchers from San Diego biotechnology company Ambit Biosciences Corp. published a method to evaluate one class of therapeutic molecules, kinase inhibitors, more comprehensively at the preclinical stage.

The researchers focused their efforts on developing an assay for inhibitors of the kinase ATP-binding site.

"If you can block the ATP site, you've essentially shut down the kinase, because all kinases require ATP to function," said David Lockhart, chief scientific officer of Ambit Biosciences and a co-author of the paper.

Also, despite the fact that every kinase has an ATP site, the binding of many inhibitors to those sites is specific. "The dogma up to very recently was that because every kinase has [an ATP binding site], targeting one kinase will target them all - and kinases are important, so you can't just shut them all down," Lockhart told BioWorld Today. "It was Gleevec that really showed that specificity is possible and a therapeutically viable path."

While targeting of other sites on the protein kinase (for example, the substrate interaction site) also is possible, most small-molecule inhibitors in clinical trials target either the ATP site directly or nearby sites that affect the ATP site.

The method is simple in principle: The kinase in question first is expressed in a T7 bacteriophage, and that phage then is used in a competitive binding assay between a "bait" kinase inhibitor linked to a support structure and the kinase inhibitor to be tested. The amount of phage binding to the bait can be quantified, and the relative strength of binding to the test ligand inferred.

Neither phage display nor competitive binding are exactly novel concepts, as Lockhart freely pointed out. The challenging part of getting the assays to work was the tinkering required with the T7 phage. The phage used in the experiments is proprietary to Ambit, and Lockhart said the profiling will not work with off-the-shelf phages.

Iressa And Tarceva: The Mystery Remains

Lockhart said one advantage of the assay is that because many inhibitors can be profiled at the same time, "you get to look broadly, and to be surprised." The researchers profiled more than 100 of the 518 known human protein kinases, and surprises came in several different forms. One unexpected result was found in the binding profile of Gleevec (imatinib), Novartis AG's chronic myeloid leukemia drug. Gleevec previously was known to bind the BCR-ABL kinase, which accounts for its effects on CML. It also was known that it does not bind tightly to the SRC kinase. However, the researchers found that Gleevec does bind another member of the SRC kinase family, known as LCK.

The LCK kinase is involved in T-cell maturation and a potential target for immunosuppression. "The fact that Gleevec does not bind SRC, but does bind a closely related and therapeutically relevant kinase - I don't think that could have been predicted" from sequence or other binding information, Lockhart said.

The researchers also investigated Tarceva (erlotinib) and Iressa (gefitinib), two epidermal growth factor receptor inhibitors approved for the treatment of non-small-cell lung cancer. Given the fact that both compounds inhibit the same receptor and Tarceva was approved based on a survival benefit relative to placebo of more than 40 percent, recent news that Iressa failed to show a survival benefit compared to placebo came as quite a surprise. (See BioWorld Today, Dec. 20, 2004.)

There has been speculation that the difference might be due to differential specificity of Iressa and Tarceva for mutations of EGFR. However, when the scientists tested Iressa and Tarceva against EGFR and nine of its known mutants, the binding profile was similar across the board, suggesting their different effects on survival must be due to other, as yet unidentified, factors. Negative data are sometimes met with less enthusiasm, as Lockhart acknowledged, saying that "if there had been differential binding, [the Iressa and Tarceva data] would have been its own paper." But such findings also are clinically relevant.

Ambit has signed deals with Bristol-Myers Squibb Co., GlaxoSmithKline plc and F. Hoffmann-La Roche Ltd. on the use of technology, and Lockhart said that additional deals are being negotiated. Lockhart added that "we are our own customer." Ambit is using the assays to identify kinase inhibitors to develop in-house. The company hopes to enter clinical trials by mid-2006.