FT. LAUDERDALE, Fla. _ What fired up cell biologist LewisCantley's audience at the 1996 Miami Bio/Technology WinterSymposium last week was not the metabolic pathways per se so muchas vistas of "a soluble oriented peptide library approach that allowscracking of protein interaction motifs in a day or two, instead ofweeks or months."
Cantley, a professor at Harvard Medical School in Boston, and itsBeth Israel teaching hospital, told the attendees here, "By ourapproach, we have been able to determine the optimal motifs forabout 15 different tyrosine kinases [TK]. These peptides," he added,"are of particular interest for oncology, because some TKs areinvolved in signaling messages to cells by growth factors, hormonesand neurotransmitters."
In his talk on "Structural basis for specificity in signaling via proteinkinases," he continued: "TKs are in cascades that tell a cell toundergo mitogenesis, and hence many of them have been identifiedas oncogenes, whose uncontrolled activation results in tumors of thebreast and other tissues, particularly epithelial."
Another type of TK, he pointed out, "the platelet-derived growthfactor receptor, is thought to be involved in development ofatherosclerosis, so it's important in smooth-muscle growth invascular tissue."
Cantley's goal is to develop inhibitors to these tyrosine kinases, "toblock progression of disease in chemotherapy." Potentially, hesuggested, "one could design peptidomimetic drugs that takeadvantage of the structure we know is critical for high-affinityinteraction, and make a non-productive pseudopeptide that would sitat that same catalytic site. It would tie up the site and block access tothe kinase, hence arrest the growth of the tissue."
Beth Israel Hospital has a patent pending on Cantley's protein kinasepeptide-library approach, but rather than push into such outright drugdiscovery, has established a collaboration with Glaxo Wellcome Inc.,in Research Triangle Park, N.C. "At this point," Cantley toldBioWorld Today, "Those approaches are best undertaken bypharmaceutical companies that have large groups of chemists whocan take advantage of the structures, and try to design an inhibitorbased on this one."
His generic approach involves interrogating soluble, oriented peptidelibraries instead of conventional phage, gene or tagged cDNAlibraries. He began in 1992, "trying to determine the optimumpeptide sequence binding to a signal transduction protein called thesrc homology 2 domain, SH2.
"We realized," Cantley recalled, "that to look at only a three-amino-acid [AA] sequence, one would have to examine all possible AAsequences. He and his colleague Zhou Songyang reasoned that "asthere are 20 AAs possible in each position, the total number ofpossibilities is 20 x 20 x 20. This is 8,000 peptides we would have tomake to compulsively scrutinize every possibility. That's ratherlaborious, and also very expensive."
Instead, they decided to make a stable mixture of peptides that had all8,000 possibilities in a single library. "We constructed this in such away that those structures were within a larger structure that wasconserved in all the peptides. That guaranteed that when theindividual peptides from that mixture bound to the domain we wereinterested in, they all bound at the same site."
The pair recognized very quickly that by sequencing the entiremixture of retained peptides, it could see common features withouthaving to look at the structure of any one individual peptide. "So wefound," Cantley went on, "that for a particular SH2 domain, all of thepeptides bound with high affinity had certain structures in common.
"In essentially a one-day experiment, we had gleaned all theimportant facets of exploring those 8,000 structures with a minimumamount of work. From a single step of purification, and a singlesequencing with multiple structures at a single position, we could seewhat that particular protein wants to bind.
"A lot of pharmaceutical companies are trying to develop suchinhibitors," Cantley said. "Competitive techniques for doing this areextremely laborious, and may take a year of many people working toget the optimum motif. In this approach, we can get it in a day, usingconventional peptide synthesizers."
Juggling 10 Trillion Domains At A Time
The largest libraries that his team has used to date contain roughly 10trillion individual structures in a single mixture. "So it's possible tostart out with roughly a milligram of peptides," he said, "that haswithin it 10 trillion individual structures with some features incommon, such as the orienting amino acid somewhere in the middle,but otherwise differ in one or another amino acid at some location.
"We can explore such a large number of structures in a single day,"he concluded, "because we don't try to identify a single structure outof that mixture of 10 trillion, but rather determine a consensusstructure of the peptides that have in common a feature designed forhigh affinity to the domain of interest."
"Part of the charm of Cantley's work," said clinical pharmacologistRichard Tsien, of Stanford University, "is the sensibleness withwhich he has gone about applying his general concept, that apopulation of candidate peptides, with slight variations among it, canbe isolated from a larger pool, and its properties isolated simply bysequencing the individual members."
Tsien, who also is research committee chairman of Neurex Corp., ofMenlo Park, Calif., told BioWorld Today, "Cantley has come up witha method that allows one to use the protein-protein interaction topurify a subclass from a library of peptides, and then, bystraightforward protein sequencing, learn a lot more about thestructural requirements of the peptide's binding site." n
-- David N. Leff Science Editor
(c) 1997 American Health Consultants. All rights reserved.