By Kevin Davies
May 15, 2004 | Over two days, a formidable group of speakers in “The Druggable Genome” presented the challenges of entering the genomic era for basic research, drug discovery, and medicine.
At the genome level, priorities must be to define all of the functional elements and variants in the sequence, and the pathways altered in diseases, said David Altshuler (Broad Institute). Understanding why more than half of the sequences conserved between humans and mice lie outside gene-coding regions requires comparative genomics. Altshuler hailed the NIH comparative genomic sequence plan as “remarkably foresightful.” Genomic profiling methods are identifying new targets in leukemia and type II diabetes, agreeing with leads from other studies. Progress in cataloging SNPs is equally dramatic. Out of a universe of about 10 million common alleles, more than 7 million have been cataloged in the dbSNP database. “Ninety to ninety-five percent of these variations occurred just once in human history, tethered to a set of flanking SNPs, or haplotypes,” Altshuler said.
The International HapMap project aims to map 1 million variants in 270 individuals of different ethnic backgrounds. “There is tremendous progress in genotyping technology -- prices are falling,” Altshuler said. Nevertheless, Aravinda Chakravarti (Johns Hopkins University) noted that we possess only 1 percent of the tide of genomic data that can be expected over the next five years. “Comparative genomic sequencing will predict specific experiments in the wet-dry cycle,” Chakravarti said. Experimental data will be intrinsically intertwined with mathematical models, which in turn will drive further experimentation, such as why 15 to 20 percent of simple genetic disorders have no identifiable mutations in the relevant coding or regulatory regions.
Pfizer’s Andrew Hopkins argued for improved ROI on new drug targets. Thirty percent of targets are abandoned during high-throughput screening, 60 percent at the “hit”-to-candidate stage. Together, discovery accounts for 50 percent of R&D costs. Only 10 percent of marketed drugs break one of Lipinski’s classic “Rule of Five” parameters for small molecules. “Promiscuous drugs [including most blockbusters] tend to be smaller,” Hopkins said. “Druggability can be assessed a priori, but we cannot hit every target with a small, orally available drug.” Currently, drug discovery is a $50-billion industry aimed at fewer than 300 targets. “Even doubling the number of current targets is very exciting,” Hopkins said.
Catherine Burgess (Curagen) pointed out that only four new drugs approved last year addressed novel drug targets: Velcade (multiple myeloma), Iressa (lung cancer), Raptiva (psoriasis), and Zavesca (Gaucher’s disease). Curagen has aggressively mined the human genome for “druggable genes” of disease relevance. The first patent from this endeavor -- on hepatoma-derived growth factor-like proteins -- was issued earlier this year.
Several speakers addressed the challenges of dissecting polygenic traits and developing drugs. Ten years ago, the obese gene (leptin) was discovered in mice by Jeff Friedman’s group at Rockefeller University. Nevertheless, Friedman, a co-founder of Millennium Pharmaceuticals, noted that conventional medical wisdom in treating obesity has changed little since the advice of Hippocrates 2,000 years ago: “Eat once a day, take no baths, sleep on a hard bed, and walk naked all day!”
Leptin mutations in humans are vanishingly rare, but in such cases, the effect of leptin injections is as miraculous as injecting insulin for diabetes, dramatically reducing body weight and insulin levels. But with most obese individuals insensitive to leptin, Friedman said, “there are therapeutic opportunities by understanding the neural circuit.”
Friedman has been working with authorities on the Micronesian island of Kosrae, where obesity affects half the population, recording medical and family histories of more than 2,000 inhabitants. Friedman’s team has tracked the entire population into a single complex family tree. Candidate genes should emerge soon.
Calorie restriction is a key to longevity, which as Lenny Guarente (MIT) has demonstrated in model organisms, involves a gene called Sir2. SIRT1, the mammalian homologue, is a repressor of the p53 tumor suppressor gene, the virtues of reducing cell growth presumably offsetting the inhibition of the so-called guardian of the genome.
Studies in mouse models and inbred families have enabled Angela Christiano (Columbia University) to identify a handful of genes that regulate hair growth. The early drug candidates are for hair removal -- one drug from Bristol-Myers Squibb effectively slows cell growth at the base of the hair follicle. Christiano hopes that microarray studies will shed light on a particularly debilitating complex trait -- that of male-pattern baldness.
The grandmaster of complex genetics is Kari Stefansson (deCODE), who traces his own personal heritage back a millennium to Egil Skallagrimson -- “a great poet, great warrior, and the ugliest man ever to live,” his distant descendant said. His ancestors “may have been poor, belligerent drunks, but they knew how to write a genealogy.” DeCODE has isolated 15 disease-related genes, 13 of which encode “good druggable targets.” Several drug development projects are under way, including a clinical trial for stroke, using DG031, the former Bayer asthma drug Stefansson called a “good target, wrong disease.” (See DeCODE’s Report on Cardiac Gene Study Causes a FLAP, March 2004 Bio-IT World, page 14.)
Good Targets, Right Disease
Art Sands (Lexicon Genetics) pointed out that the top 100 selling drugs address only 43 drug targets, and all drugs currently in Phase II or III trials provide only an additional 23 novel targets. Lexicon has “fully phenotyped 1,500 knockout mice of its ‘Genome5000’” initiative, yielding more than 40 drug discovery programs. Bristol-Myers Squibb is Lexicon’s neuroscience partner, but Sands spotlighted several other promising knockouts in areas of obesity, Alzheimer’s disease, and cancer.
Alexander “Sasha” Kamb, the new head of oncology at Novartis, addressed the issue of “nontraditional targets,” advocating an approach of “guided empiricism” to survey crystal structures to optimize small-molecule binding. Noting that the terminal fragment of a protein called SMAC inhibits IAP, which is involved in cell death, Novartis developed a 3-amino-acid peptide that retains inhibitory activity. First-generation small-molecule mimetics are currently under investigation.
Eidogen, a Pasadena, Calif., company, stresses “structural informatics” as the key to modeling drug-target interactions. Founder Derek Debe said that the “PDB data deluge” is “a $4-billion resource” for the company (25,000 proteins at $150,000 per crystal structure). Debe says Eidogen’s programs identified alternative “drugging” sites for the Gleevec target, sites that were independently confirmed in a Cell paper last year.
With so much progress at the genome level, Fred Ledley (MyGenome Inc.) questioned why predictive genetic tests reach only 5 percent of the population, despite the availability of tests for heart disease, Alzheimer’s disease, osteoporosis, cancer, eclampsia, and so on. Implementation problems include demographics, unreliable predictability, clinical utility, and privacy/confidentiality, but much wider testing is justified. More than 3 million people take warfarin (rat poison), but only 5,000 tests for the CYP2C9 genotype, which governs clearance of warfarin, are performed annually. More than 4 million patients have Alzheimer’s disease, but the excellent APOE4 test is not commercially available. With only 5,000 counselors worldwide, the solution is in informatics systems to help record patient histories and disseminate information.
The track closed with a fascinating discussion between Beth Arnold (Foley Hoag) and Richard Gold (McGill University) on “What to Patent?”, focusing on issues such as European opposition to the Myriad breast cancer patent. Surprisingly, the top gene patent holder in the United States is not a Big Pharma or Incyte or Celera, but the University of California at San Francisco, with almost 900 patents, closely followed by the U.S. government.
SIDEBAR: Land of the Rising SNP
Arguably Japan’s leading genome researcher, Yusuke Nakamura updated delegates with the latest progress in SNP studies. More than 2,300 megabases of DNA have been resequenced, and the corporation Riken contributes 25 percent to the International HapMap Project, with 300 million SNP genotypes per year. The secret is the clever “Invader” colorimetric PCR assay to type individual genotypes, and an automated ultrasound sealing method to prevent reagent loss. Several SNPs have been associated with disease states, including myocardial infarction, rheumatoid arthritis, and diabetic nephropathy.
Nakamura also described the Biobank Project, a national “personalized medicine” program to discover disease genes and gather information for evidence-based drug development. 300,000 individuals have provided DNA and sera for SNP and proteomic analysis to identify medically important genes. Samples are stripped of identification data. Eight hospital and university groups are contributing to the Biobank, while 89 percent of Japanese subjects (about 40,000) approached so far have granted informed consent.
One million sample tubes will be stored in a fully automated system in 50 liquid nitrogen tanks.
SIDEBAR: The Single Life
With the current cost of resequencing a single human genome estimated at $30 million to $50 million, the search for viable alternatives is attracting great interest. Two companies presented contrasting strategies.
The approach at Solexa, based in the United Kingdom, is to sequence billions of short, randomly arranged oligonucleotide fragments in parallel (without PCR) on a centimeter chip. Using four fluors corresponding to each nucleotide, the elongating strands are digitized, said Simon Bennett, the company’s new head of business development. Bennett foresees myriad applications, in areas from comparative genomics and disease genetics, to measuring drug response and prognostic screening.
In Madison, Wis., OpGen is also looking at single molecules, but on a rather larger scale. Using a process called optical mapping, OpGen takes cells from microbes, humans, or any origin, lyses them, and extracts and arrays the DNA molecules. A sophisticated algorithm identifies the DNA fragments. “One thing all these [current] systems have in common is they basically amplify very small sections of the genome,” said OpGen chief scientific officer Colin Dykes. “The problem is that it ignores a great deal of information that is present in the genetic structure.”
Early studies have focused on microbes such as aspergillus, but the company is now looking at human applications, including mapping genomic rearrangements in tumor cells. ”We should be able to isolate and study human DNA as a series of long single molecules,” Dykes said. “The single molecules give you more information.”
* “The Druggable Genome,” co-organized by Kevin Davies and Jim Golden (Life Science Insights). Sponsor: Foley Hoag. Media Sponsor: Nature