Motivating Hotspots

See allHide authors and affiliations

Science  14 Oct 2005:
Vol. 310, Issue 5746, pp. 247-248
DOI: 10.1126/science.1120154

An estimated 10 to 30% of fertilized human ova have more or fewer than the normal number of 46 chromosomes (1). Most cases of aneuploidy (that is, having an abnormal number of chromosomes) result from a reduced number of recombination events (less than one per chromosome arm) or an unusual placement of such events along the chromosome (2). Recombination between nonallelic copies can also lead to chromosomal rearrangements and deletions, many of which have been associated with disease (2). Despite its importance, much remains unknown about the process of recombination, especially for mammals. In particular, no sequences or chromosome structures that influence human recombination rates have been identified. On page 321 in this issue, Myers et al. (3) present a fine-scale recombination map for the human genome and use it to identify the first set of candidate motifs.

Estimates of recombination rates for the entire genome have previously come from comparisons of physical maps to genetic maps based on pedigrees. Such comparisons revealed higher recombination rates in females than in males and suppressed recombination near centromeres, among other features, but lack the resolution to be informative about genetic distances below a centimorgan (2). To estimate finer-scale rates, there are currently two approaches: sperm-typing and analyses of linkage disequilibrium. Sperm-typing experiments provide direct estimates of recombination rates in regions from one to several hundred kilobases (kb) in length (4). The dozen regions examined by this technique show extensive heterogeneity in fine-scale recombination rates, with most crossover resolutions concentrated in segments of 1 to 2 kb called “hotspots.” These findings therefore suggested that hotspots may be an extremely common feature of the human recombination landscape (5), as they are in yeast (6). Unfortunately, sperm-typing is labor intensive and not feasible on a genome-wide scale.

The alternative is to estimate fine-scale recombination rates indirect from levels of linkage disequilibrium, the nonrandom association of alleles on chromosomes. Patterns of linkage disequilibrium observed in extant individuals reflect recombination events that have occurred in the ancestors of the individuals from whom the sample was obtained. More recombination in the ancestral lineages tends to result in weaker associations between alleles (that is, less linkage disequilibrium), and vice versa. Given a model of recombination and population history, one can therefore use linkage disequilibrium data to estimate the recombination rates along a chromosome and to test for the presence of hotspots (that is, for a local increase in the rate of recombination relative to that expected by chance, under a model of uniform rates). What one obtains is a population rate of recombination, averaged over both male and female ancestors, in the time frame reflected by linkage disequilibrium—less than 1 million years for humans. When applied to the X chromosome, the approach provides estimates of fine-scale rates of recombination in females (which sperm-typing obviously cannot do). Despite the dependence on an oversimplified model, linkage disequilibrium-based estimates of recombination produce a fairly reliable characterization of rate variation, both in simulations and when compared to sperm-typing data from the region encoding the major histocompatibility complex and to genetic map-based estimates (7-9). Moreover, they can be readily applied to any chromosomal region for which there are genetic variation data.

Sequence motifs in the human genome affect hotspot activity. (Top)

Recombination refers to the exchange of DNA between chromosomes during meiosis. Shown here is a crossing-over event in which sections of homologous chromatids are exchanged. Recombination is necessary for the proper segregation of chromosomes in meiosis. (Bottom) Recombination rates vary along a chromosome, with most crossover events occurring in hotspots. Myers et al. identify sequence motifs in the hotspots that may influence their activity, with some alleles leading to greater intensity than others.

With the release of genome-wide genotyping data, it has therefore become possible to construct a fine-scale genetic map for most of the human genome. Myers et al. do so by applying their linkage disequilibrium-based estimator to publicly available genotyping data for 1.6 million single nucleotide polymorphisms distributed over most of the human genome (10). Confirming sperm-typing studies and analyses of linkage disequilibrium for subsets of the genome (7, 8, 11), they find extensive rate heterogeneity over the scale of kilobases. They further estimate that there is a hotspot every 50 kb or so in the human genome, with ~80% of the recombination events occurring in 10 to 20% of the sequence. Their analyses of the X chromosome demonstrate that this heterogeneity is a feature of female as well as male recombination landscapes.

This high-resolution genetic map of the human genome will be an important resource for the design of association studies for complex diseases (4). Beyond its descriptive value, however, the map is a fantastic tool to begin understanding the determinants of recombination rates. For example, in yeast, hotspots have been classified into three groups: α-hotspots, where the recombination machinery is recruited to transcription factor binding sites; β-hotspots, found in nucleasesensitive regions; and γ-hotspots, associated with elevated GC nucleotide content (6). As the Myers et al. study reveals, two of these cases don't seem to be widespread in humans. Recombination rates tend to be lower, not higher, near coding regions, increasing (on average) for about 30 kb from the start codon, and GC content is not a strong predictor of fine-scale rate variation. Whether hotspots tend to lie in nuclease-sensitive regions or whether a new model is warranted for humans remains to be determined.

With more than 25,000 putative hotspots at their disposal, Myers et al. also find the first set of sequence features substantially overrepresented in hotspots relative to coldspots. Although the motifs are neither necessary nor sufficient for hotspot activity, their top-scoring candidate alone may play a role in a substantial proportion (11%) of them. The motif is a seven-nucleotide oligomer (CCTCCCT) and its effect on recombination is strongest when it lies within a specific mobile element (namely, within the long terminal repeats of a retrovirus-like retrotransposon). Low-copy number repeats are thought to play an important role in the generation of recombination events between nonallelic copies. The finding of a recombinagenic motif within the repeats may therefore help to explain the observation that the breakpoints of nonallelic recombination events are often clustered (12). The overall influence of mobile DNA elements on recombination remains unclear, however, with some over- and some underrepresented within hotspots.

The seven-nucleotide motif is not among those previously associated with recombination in other species. However, its role in influencing recombination is supported by sperm-typing experiments, as is the role of another nine-nucleotide motif motif (CCCCACCCC) identified by the authors. Indeed, at a subset of hotspots in humans, mouse, and yeast, variation in hotspot intensity among individuals has been shown to depend on particular alleles, with recombination events occurring more often initiating on the background of the “hot” variant. When Myers et al. examined the sequence context of two human hotspots whose intensity has been shown to vary among alleles, they found that the “hot” alleles were their top-scoring seven and nine oligomer motifs and that in both cases, the “colder” alleles were a mutation away from that motif. This result strongly suggests that these sequences modulate hotspot activity (see the figure). Further evidence will come from sperm-typing studies of other hotspots polymorphic at the same motifs, as well as at other candidate sequences.

In light of recent reports that hotspot locations are largely discordant in humans and chimpanzees (9, 13), the discovery of human motifs that appear to influence hotspot activity raises a number of additional questions: Can changes to sequence motifs explain most of the interspecies differences, or do other genomic features, such as chromatin accessibility or transposable element activity, explain their rapid evolution? Given that most recombination events take place within hotspots, and hotspot locations appear to be rapidly evolving, is there any constraint on recombination rates below that of a chromosomal arm? For example, are the density and intensity of hotspots constrained within circumscribed regions of the genome? With more sperm-typing experiments and extensive linkage disequilibrium data collection in close evolutionary relatives of humans, answers to these questions should no longer be elusive.


View Abstract

Stay Connected to Science

Navigate This Article