Gene Map Shortcut
Context: The Human Genome Project provided a parts list of our genes, but that alone cannot connect genetic variants with health; diseases and drug responses must be correlated with genetic markers that vary from person to person. The most common and easily assessed markers are single-nucleotide polymorphisms (SNPs), in which genetic sequences swap one DNA “letter” for another.
But an individual has many more SNPs than researchers can afford to measure, and until now, there have been no reliable tools for selecting the most representative ones. Recently, the biotech company Perlegen Sciences and its collaborators at the International Computer Science Institute and the University of California, San Diego, completed the first map of SNPs that provides these tools. This map will move us closer to an era in which patients’ genetic makeup routinely guides their medical treatment.
Methods and Results: To figure out which SNPs are most informative, researchers at Perlegen needed to know how frequently particular SNPs occur in a population and which ones are most likely to occur together. Led by David Cox, the researchers selected SNPs that represented all 46 chromosomes and whose least-frequent versions still occurred in 1 to 5 percent of individuals. Each occurred at least once in 71 volunteers identified as being of African, European, or Han Chinese ancestry; the researchers kept track of who had what SNPs and which ones were likely to occur together in each population. Overall, Perlegen studied nearly 1.6 million SNPs. The researchers compared their data with that from other studies of much smaller sections of the genome, where nearly all of the so-called common SNPs have been found. That comparison revealed that the SNPs that Perlegen studied could be used to predict the occurrence of most others that were not in its set of 1.6 million. In other words, the researchers found that studying a subset of SNPs captures most of the information about frequently occurring variations. They also found that the subset of common SNPs varied slightly among different ethnic groups.
Why it Matters: The Perlegen study shows that future genetic-variation studies might be easier and cheaper than previously thought. Because so many SNPs are correlated with each other, studies that measure fewer SNPs can potentially get results as powerful as those that measure many more, and the map the researchers created will help epidemiologists select SNPs to study.
A second map of SNPs, which surveys more people, is due out from the International HapMap Project later this year. The two maps will complement each other. For now, Perlegen has shown that these maps can be made and could provide a powerful tool to link genetic variations with disease and medicine.
Out with the Bad, in with the Good
Context: The usual goal of gene therapy is to supply missing instructions for creating a protein. A newer kind of gene therapy instead short-circuits problematic instructions. Supplying the missing protein has shown both promise and problems: it cured children of severe combined immunodeficiency, a rare genetic disease, but has also led to fatal leukemia and toxicity. The second approach improved symptoms in mice suffering from a disease similar to Huntington’s but is not yet ready to try in humans. If it proves safe, a more versatile option may be to combine both approaches, suppressing a problematic protein while creating a therapeutic one. That’s what Patrick Aebischer’s team at the Ecole Polytechnique Federale de Lausanne in Switzerland has done for a mouse with the rodent version of Lou Gehrig’s disease.
Methods and Results: Amyotrophic lateral sclerosis, or Lou Gehrig’s disease, kills muscle-controlling neurons; patients eventually become paralyzed and die. In some cases, mutations in a gene called SOD1 are responsible. Mice engineered with a mutant version of human SOD1 have trouble walking, breathing, and grooming themselves and eventually die. To silence this problem-causing gene, Aebischer’s team used a technique called RNA interference (RNAi). In RNAi, short strands of RNA match up with a gene’s messenger RNA, destroying it before it can take part in protein assembly. The researchers inserted instructions for RNA targeting SOD1 into a genetically modified virus. Then they injected the virus into mice. RNAi suppressed the mutant human SOD1 gene but ignored normal mouse SOD1. Though the treated mice still got sick and died, they remained healthy longer. However, in humans, RNAi would likely silence not just the mutant version of the SOD1 gene but also the healthy version. So Aebischer’s team genetically engineered a virus to deliver both instructions for RNAi and a gene for human SOD1 designed so that it would not be shut down by RNAi. Though the new virus has so far been tested only in cells, it is the first demonstration of a promising new technique.
Why it Matters: Aebischer’s technique could overcome a nagging limitation of RNAi. Particularly for diseases like Lou Gehrig’s, which can be caused by many different mutations in a particular gene, RNAi cannot be engineered to attack the disease-causing form of a gene without attacking its normal counterpart. By designing genes unaffected by RNAi and delivering them alongside instructions to silence the existing gene, Aebischer’s team has found a general way to use the technique against diseases in which the nontargeted gene must remain functional.
Better Hope for Brain Cancer
Context: Patients diagnosed with glioblastoma, a particularly malicious kind of brain cancer, rarely live much longer than a year, even with surgery and radiation therapy. Chemotherapy seems to help some patients but not most. Now, in a large clinical trial, teams led by Roger Stupp and Monika Hegi at the Centre Hospitalier Universitaire Vaudois in Switzerland and the European Organisation for Research and Treatment of Cancer have identified a more-effective form of chemotherapy and a way to single out which patients are likely to benefit from it.
Methods and Results: Patients were randomly assigned to receive radiation either on its own or combined with a drug called temozolomide, which damages DNA. On average, patients receiving the drug lived about 10 weeks longer than patients who did not, a benefit that has been considered significant enough to warrant the approval of other cancer drugs. (In fact, in March, the U.S. Food and Drug Administration approved temozolomide for treating glioblastoma.) When possible, the researchers also examined a DNA-repair gene in patients’ tumors to see if it had been silenced by a so-called epigenetic modification, which alters gene expression without changing a gene’s sequence. In healthy cells, such silencing is often one of the first steps toward cancer. Ironically, however, the silenced gene repairs the very damage caused by temozolomide. The results were striking: patients with silent DNA-repair genes who received temozolomide and radiation lived nine months longer than patients with active genes who received the same treatment. In fact, in cases where the DNA-repair gene was active, patients who received combination therapy did not live significantly longer than patients receiving radiation therapy alone.
Why it Matters: Chemotherapy makes people feel sick, and physicians are loath to prescribe it when it is unlikely to help a patient live longer. However, in the United States, chemotherapy is often prescribed even if it has only a slight chance of extending life. Stupp and Hegi’s results show not only an effective therapy but also a way to pinpoint those patients most likely to benefit from it. The study also illustrates the need to improve the ease of genetic tests, particularly those that look for problems with gene expression rather than changes in DNA sequences. The researchers could tell whether a DNA-repair gene was silent in only 67 percent of the patients for whom tumor material was available, because some tumor samples were of poor quality. With such a promising way to detect which patients should get chemotherapy, researchers need better ways to collect and store samples, as well as tests that work with ill-preserved samples.