The Promise that Haplotypes Hold

Sifting through the 3 billion letters of DNA that comprise the human genome to find disease-associated sequences is at best a daunting task. But recent evidence suggests that this genome is organized into highly structured blocks called haplotypes, highly conserved packages that contain SNPs (single nucleotide polymorphisms) and exhibit very little genotypic variation. Enlisting these haplotypes in the search may make this undertaking easier than previously thought. Evidence of such organization has been building over the past several years, says scientist Mark Daly, Whitehead Institute Center for Genome Research, Cambridge, Mass. But a slew of papers in last month's Nature Genetics are the first attempt to bring the data together.1-5 If the evidence is proven, scientists plan to use these haplotypes to locate disease genes, a more concise process than wading through the entire genome. Courtesy of Whitehead Institute John Rioux Single base pair differences form most human genetic variations and have been found on all 23 human chromosomes. So far, scientists have mapped more than 1.4 million of them on the human genome. But "SNPs don't genetically travel on their own," says geneticist John Rioux, Daly's co-author on two of the Nature Genetics papers.1,2 "They travel in packs," explains David Altshuler, also at the Whitehead Institute and a Harvard Medical School professor. These packs, known as haplotypes. are highly conserved ancestral chromosomal segments that have not undergone the shuffling, or recombination, that occurs in the genome during sexual reproduction. The SNPs that make up an individual haplotype are so tightly linked that genetic rearrangements have occurred within them at a much lower than expected rate. So, they have been inherited as a single unit, or block, through the generations. Each block comes in a handful of different haplotypic variations or "flavors," says Altshuler. A block made up of 10 SNPs may only have a few different flavors, even though there are 1,024 (210) different possible combinations, explains Rioux. The most recent evidence suggests that these blocks may be arranged in a highly structured pattern along the length of the chromosome.1 Daly and his colleagues show that a 500,000-base segment of human chromosome 5 can be broken up into a set of 11 discrete haplotype blocks ranging in size from 10,000 to 100,000 bases.1 The blocks show "a striking lack of diversity," the authors write. For example, in one 84,000-base block that contained eight SNPs, 95 percent of genetic variation in the population comprised two distinct haplotypic flavors. If those eight had traveled solo, there would have been 28 different possible allelic combinations. Daly's paper describes the block pattern, but, says Altshuler, it's the Alex Jeffries paper that "suggests that we might begin to understand why it looks the way it does." In the same Nature Genetics issue, geneticist Jeffries and his colleagues from the University of Leicester, UK, describe recombinational "hot spots" clustered between blocks on a 216,000-base segment of the major histocompatibility complex (MHC).3 Their data show how "biologically, recombination is not happening at random through the genome," says Daly. Instead, it is occurring in clusters, creating the block-like pattern. The fact that these new results are consistent with already published data about other parts of the genome suggest that haplotype blocks may be a genome-wide phenomenon, as Daly and his co-authors indicate. Still, the data are from a fairly limited region of the genome, says geneticist Leonid Kruglyak, Howard Hughes Medical Institute investigator at the Fred Hutchinson Cancer Research Center in Seattle. Of course "such regions should exist," he says, but working out the details is tricky. "It's still open to question as to exactly what's going on. There is a tremendous amount of variability for different portions of the genome." Rioux agrees that it may be too early in the game to say with certainty that the block pattern is genome-wide. At this point, "we're opening the door a little crack and peeping in," he comments, but all the data thus far indicate that the blocks are a general phenomenon. Altshuler and his collaborators may be opening that door a little wider. At the Cambridge Healthtech Institute's Fifth Annual Functional Genomics conference in Cambridge on Oct. 9, Altshuler described a sampling of results from an unpublished study on highly conserved haplotypic blocks at 60 different genomic regions. "We see patterns that are broadly consistent," he says, "suggesting that this is a genome-wide pattern." Mapping Made Easier Researchers use a technique known as linkage disequilibrium (LD) mapping to locate disease genes. Linkage disequilibrium occurs when alleles of different genes are inherited together, usually because they are in close proximity to one another. By measuring the LD between a disease gene and each SNP in and around that gene, it should be possible to pinpoint a disease gene's location, right down to one SNP. But this kind of single-marker LD mapping is really not very powerful, statistically speaking, according to Altshuler. And it is never clear if the implicated SNP is the true culprit or just one of its neighbors. The data derived from this kind of analysis are usually "just so scattered that there's no rhyme or reason to it," he comments. Rioux says that using haplotypes instead of SNPs to locate disease genes provides a much crisper picture, which is exactly what he and his collaborators from the Whitehead Institute, McGill University's Montreal Genome Center, and the University of Toronto did.2 First, Rioux and colleagues used family-based linkage studies to implicate a gene for Crohn's disease on chromosome 5. They identified 651 SNPs within and around a 250,000-base block at the gene's center. But instead of uncovering numerous, possible combinations that these polymorphisms could display if inherited independently, the researchers found that two to four common haplotypes comprised 90-98 percent of the chromosomes in the study. Rather than try to correlate Crohn's disease with one of hundreds of SNPs in the region--an improbable task--the researchers homed in on the block instead. Their efforts proved successful as they identified a risk haplotype that increased the possibility of Crohn's disease by two-fold in heterozygotes and six-fold in homozygotes. Thus, Rioux's team narrowed the search for a Crohn's disease-causing gene from 651 SNPs to the 11 that made up the risk haplotype. Narrowing it further "is not a small challenge," says Rioux. Indeed, haplotype-based association studies are a double-edged sword, notes Altshuler. They make localizing genes easier, but "once you find the haplotype, genetics doesn't really take you much further," says Altshuler. There may be a dozen mutations in a single haplotype block, for example, but because they are so tightly correlated, determining which one is causing the association is unknown. But at least the search is narrowed. "You're not just implicating a region," Altshuler says, "you're actually implicating specific variants." LD mapping can still be used for haplotype hunting, but the first step is to identify the SNPs that are unique to each one. In another study in the same Nature Genetics issue, researchers refer to haplotype-specific SNPs as "haplotype tag SNPs, or htSNPs."4 They warn, however, that only about 25 percent of the htSNPs identified in their study were included in the public SNP database. Many more htSNPs will need to be catalogued before more researchers can turn to haplotype-based LD mapping. A Boost for Biomedicine Such mapping "will make much more feasible some of the medical genetic work that we'd like to do in the future," Daly says. Not only will it be easier to locate disease genes, as Daly and his colleagues illustrate with Crohn's disease, but it will also make it easier to hunt genes that affect how individuals respond to disease therapy. "Let's say that there are a few mutations that influence how an individual responds to a drug, but that we don't have any idea where in the genome they are. What do we need to do to carry out a proper search for such mutations?" asks researcher David Goldstein, University College, London. "One thing we could do is fully sequence all individuals that respond to the drug and all those who do not and look at every single difference between the two groups. Unfortunately, we are not in a position to do this routinely, and we may not be for a while longer. When we understand the haplotype structure of the genome, however, we will be able to use a subset of the variable sites in the genome to represent all the others." The human genome is made up of millions of polymorphisms, says Kruglyak. With haplotype-based LD mapping, "the hope is that you could reduce that number down to hundreds of thousands, an order of magnitude smaller." If this is true, "it seems possible that we will be in a position, in the next few years, to interrogate the entire genome for variants influencing common disease and variable drug reactions." Says Goldstein, "This has the potential to finally start delivering the practical benefits of human genetics that have been long-promised." Leslie Pray (lpray@nasw.org) is a freelance writer in Leverett, Mass. References 1. M.J. Daly et al., "High-resolution haplotype structure in the human genome," Nature Genetics, 29:229-32, October 2001. 2. J.D. Rioux et al., "Genetic variation in the 5q31 cytokine gene cluster confers susceptibility to Crohn's disease," Nature Genetics, 29:223-8, October 2001. 3. A.J. Jeffreys et al., "Intensely punctate meiotic recombination in the class II region of the major histocompatibility complex," Nature Genetics, 29:217-22, October 2001. 4. G.C.L. Johnson et al., "Haplotype tagging for the identification of common disease genes," Nature Genetics, 29:233-7, October 2001. 5. D.B. Goldstein, "Islands of linkage disequilibrium," Nature Genetics, 29:109-11, October 2001.

The Promise that Haplotypes Hold

Interested in reading more?

Become a Member of