Skip Navigation


Human Molecular Genetics Advance Access originally published online on October 27, 2004
Human Molecular Genetics 2004 13(24):3089-3102; doi:10.1093/hmg/ddh337
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
13/24/3089    most recent
ddh337v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Rana, N. A.
Right arrow Articles by Hardcastle, A. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rana, N. A.
Right arrow Articles by Hardcastle, A. J.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, Vol. 13, No. 24 © Oxford University Press 2004; all rights reserved

Recombination hotspots and block structure of linkage disequilibrium in the human genome exemplified by detailed analysis of PGM1 on 1p31

Naheed A. Rana1,2,*, Neil D. Ebenezer1, Andrew R. Webster1, Andres R. Linares2, David B. Whitehouse2, Sue Povey2 and Alison J. Hardcastle1

1Department of Molecular Genetics, Institute of Ophthalmology, University College London, 11–43 Bath Street, London EC1V 9EL, UK and 2MRC Human Biochemical Genetics Unit, Galton Laboratory, University College London, 4 Stephenson Way, London NW1 2HE, UK

Received July 20, 2004; Revised October 6, 2004; Accepted October 13, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
The distribution of linkage disequilibrium (LD) in the human genome has important consequences for the design of experiments that infer susceptibility genes for complex disease using association studies. Recent studies have shown a non-random distribution of human meiotic recombination associated with intervening tracts of LD. Little is known about the processes, patterns and frequency of reciprocal meiotic recombination in humans. However, this phenomenon can be better understood by the fine structure analysis of several genomic regions by mapping hotspots and characterizing regions with variable LD. Here, we report clustered hotspot activity with intervening blocks of LD within the human PGM1 gene (1p31) using data derived from meiotic and population studies. Earlier work has suggested a high recombination rate in two regions within the PGM1 gene, site A (exons 4–8) and site B (exons 1A–4). Sequencing of eight individuals across 6 kb of targeted regions in site B identified 18 informative SNPs. Individuals from three distinct populations, Caucasian (n=264), Chinese (n=222) and Vietnamese (n=187), were genotyped, and haplotypes were determined using estimate of haplotypes, ldmax and Arlequin. Allelic association and haplotype analysis in these samples revealed variable recombination rates across PGM1, demonstrating the presence of: (i) three hotspots and (ii) three haplotype blocks. The spatial arrangement of haplotype blocks was identical in all populations studied. The pattern of association within PGM1 represents a region decomposed into small blocks of LD, where increased recombination activity has disrupted the ancestral chromosome. Additionally, crossovers in phased data mapped preferentially to regions where LD collapses, which also overlap with sequence motifs.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
There is considerable interest in understanding the patterns of linkage disequilibrium (LD) in the human genome to examine the dynamics of meiotic recombination, investigate human evolution and facilitate association studies in complex disease. Recent studies have demonstrated that the human genome is organized in discrete blocks of low haplotype diversity, where constituent markers within blocks are in high LD. The degree of haplotypic diversity and the extent to which these haplotype blocks extend have been investigated in different human populations and for different regions of the genome (1–4). These studies have revealed great variability in haplotype structure which differs considerably from one genomic region to another and are frequently interspersed by regions of low LD (5). In some cases, haplotype blocks have been found to extend only a few kilobases, whereas others extend >100 kb (6,7). Consistency in the spatial arrangement of haplotype blocks in different populations has been reported for several regions (8,9) implicating the possibility of a common mechanism in the formation of these blocks. An African/non-African dichotomy has also been described in several segments of the human genome (2,4,10,11), where haplotype blocks extend further in non-African populations (>44 kb) than in African populations (>22 kb) (3).

The emerging pattern of meiotic recombination suggests that regions of low LD (corresponding to regions with high recombination rates) delimit haplotype blocks (6). The pattern of LD in a 216 kb segment in the MHC class II region showed extended domains of strong association (60–90 kb) punctuated by areas of linkage equilibrium, which correspond to recombination hotspots (12).

There is growing evidence that recombination hotspots are a major contributor to the formation of haplotype blocks and disruption to blocks are due to chromosome regions of relatively high recombination (9,13). This is especially important when patterns of LD are consistent in different population groups, implying that uneven levels of LD may be controlled by recombination hotspots rather than population-specific factors. However, one cannot disregard the interplay between various population-specific factors and genome-specific mechanisms in shaping LD structure.

With the heightening exploration of the genome and its recombination activities several studies have also investigated the association of meiosis and crossing over with interactions between chromatin and various structural and functional components of the nucleus. These analyses show that a variety of potential factors, both local (attributable to sequence, position and chromatin structure) and unlinked (having influence in trans elsewhere or implicated as participants in the process) can affect recombination rates over a specific chromosomal segment. Hence, candidate motifs associated with recombination activity have been proposed from investigations in: bacteria (14), yeast (15–19), rodents (16,20,21) and humans (22–27).

This article examines the recombination activity and LD structure within one of the most highly recombining regions in the human genome, the human phosphoglucomutase-1 (PGM1) locus. The enzyme is highly polymorphic with 10 common phenotypes identified in most human populations (28). The PGM1 gene spans 67 kb and contains 11 exons with two alternatively spliced first exons. Exon 1A is transcribed in a variety of cell types and exon 1B transcribed in fast muscle (29).

The original suggestion for recombination within PGM1 came from phylogenetic studies, where three point mutations and multiple intragenic recombination events between these sites generate allelic diversity (30,31). Direct evidence for a high recombination rate (0.5%, ~28 cM/Mb) within PGM1 was observed with isozyme studies of mother–child pairs where incompatible haplotypes were likely to be the result of intragenic recombination during oogenesis (32). This estimate of the overall rate of recombination between two sites (2/1 and +/–) is 28-fold the expected recombination rate for sites only 17 kb apart (29). Subsequent work showed the intragenic recombination frequency within human PGM1 to be very high (1.7%, 29 cM/Mb), indicative of a recombination hotspot (33). In one region of the gene, site A (spanning exons 4–8) pairwise allelic association in three different populations also revealed LD breakdown (33). This study also implicated another region to harbour hotspot(s) within the PGM1 gene, referred to as site B (extending from exon 1A–4). This region has recently been investigated by comparisons between genetic and physical maps and the analysis of molecularly phased data in family studies (N.A. Rana et al., manuscript in preparation).

Here, we investigate the putative intragenic recombination hotspot(s) in site B within the PGM1 locus, using detailed LD mapping in three different populations; Caucasian, Chinese and Vietnamese. We demonstrate the presence of haplotype blocks delimited by hotspots for recombination across the PGM1 gene and investigate the presence of sequence motifs associated with enhanced recombination.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Identifying SNPs within site B of the PGM1 gene
Novel markers were developed within site B (38.5 kb) by sequencing targeted regions in eight independent chromosomes. Fragments ranging from 480 to 1350 bp were sequenced in 3–4.8 kb intervals, screening a total of ~6 kb (based on sequence data from BAC/PAC clone AL109925) of site B. Eighteen SNPs were characterized, and nine were selected on the basis of correct segregation in CEPH families and no significant deviation from Hardy Weinberg equilibrium (HWE). Three additional SNPs taken from The SNP Consortium (TSC) database (http://snp.cshl@org/news/) were also validated; rs 855301, 860606 and 865115 referred to hereinafter as N2F, N5F and N7F, respectively. Details of the location of SNPs and corresponding variation are described in Table 1 and their locations in relation to the classic PGM1 2/1 and +/– isozyme sites and other known PGM1 markers (34) are displayed in Figure 1.


View this table:
[in this window]
[in a new window]
  		Table 1. Location and molecular basis of novel and previously identified SNPs within PGM1
 


View larger version (12K):
[in this window]
[in a new window]
  		Figure 1. The location of markers within human PGM1 (sites A and B). The illustration shows the relative location of markers in site B (38.5 kb) which begins at exon 1A and extends to the classic 2/1 site in exon 4. Novel SNPs identified within site B, together with those taken from TSC database are shown as well as the previously characterized site A markers (34). The classic +/– site residing in exon 8, in site A is also indicated. The distances shown are current physical distances.

 
Pairwise association analysis within the PGM1 locus
Association analysis in site B.
Allelic association was carried out within site B for 14 markers (Table 1). The validity of any interpretations made from the analysis depends on assumptions such as limited effect from population-specific factors (i.e. genetic drift, admixture and inbreeding) and genome-specific factors (e.g. age of mutation, mutation rate, gene conversion and natural selection). One way of ensuring this is to use various homogeneous populations whereby consistent interpopulation results are likely to reflect genuine patterns of LD rather than artifactual properties influencing associations (35). Multiple populations confer greater benefits to association studies than using a single population (36) in addition to increasing the sample size, any shortcomings due to hidden demographic factors may be overcome.

Comprehensive pairwise association analysis of polymorphic markers within site B were investigated in three distinct population samples; Caucasian (n=264), Chinese (n=222) and Vietnamese (n=187). Allele frequency for 12 SNPs (nine novel and three from TSC) were calculated in the three population groups after genotyping, so that any markers with minor allele frequency (m.a.f.) <0.1 could be omitted (Table 2). Using a SNP where the m.a.f. is <0.1 would have an impact on the association with other markers, since its low polymorphic nature would claim LD almost always with neighbouring markers. The impact of varying marker allele frequencies on association has recently been investigated (37,38). BunC was the only marker removed from this analysis since its m.a.f. was <0.1 and it was monomorphic in the Chinese group. The population data were also tested for deviation from HWE, a test of the data against the prevalence of genotypes at each of the markers (data not shown).


View this table:
[in this window]
[in a new window]
  		Table 2. Summary of allele frequency data in the three population samples
 
Haplotypes were estimated using the Estimate of Haplotypes (EH) program which employs the Estimation Maximization (EM) algorithm to resolve the issue of double heterozygotes. The strength of association was measured between pairs of markers using Lewontins |D'| measurement (39,40).

All polymorphic markers examined for association in this study are located within the PGM1 locus. Any allelic association detected between pairs of markers must therefore be due to the close linkage of markers, and the lack of association must indicate frequent recombination between particular markers. Since the aim of this study was not only to assess recombinationally inert regions but also to characterize domains of high recombination, the departure from association is also of interest. This is likely to be reflected in the instance where the data does not deviate significantly from the null hypothesis (P>0.05), implying that the alleles at these two loci have been re-shuffled as a result of recombination to generate a state of equilibrium. In other words, the site of recombination exists between two loci, located in close proximity.

Figures 2A–C show the pairwise allelic association (|D'|) and corresponding P-values of the 78 pairwise comparisons in the three population samples. The pairwise association data are indicative of uneven levels of LD across site B within the PGM1 gene, where recombinationally inert regions are flanked by two distinct domains where LD collapses. The analysis shows a considerable collapse in LD for markers associated with N7F, BunA, BunR and BunH to varying degrees in all three populations (Figs 2A–C). Another more prominent breakdown of LD is observed between BunR and 2/1 in all three population samples, and a further substantial collapse in LD between markers BunH and BunJ is observed in the Caucasian population (Fig. 2A).



View larger version (35K):
[in this window]
[in a new window]
  		Figure 2. Pairwise allelic association within site B. |D'| and corresponding P-values are shown for 78 pairwise allelic associations (markers in site B) in (A) Caucasian, (B) Chinese and (C) Vietnamese samples. A block of significantly high LD (P<0.05), involving markers N2F, N5F and BunF is flanked by regions of equilibrium. Collapse in LD is observed for markers associated with N7F, BunA, BunH, BunR and 2/1 to different degrees in all three population samples.

 
Conversely, the analysis revealed regions of consistently high LD (P<0.05), i.e. blocks of LD where limited or no recombination has taken place encompassed by the markers N2F and BunF extending ~8.7 kb (Figs 2A–C). The Caucasian (Fig. 2A) and Vietnamese samples (Fig. 2C) also show significant LD with a block made up of the markers Bts, Hpy and BunR located in close proximity to each other (spanning <700 bp). This pattern of association holds true for all three populations, thereby reflecting the genuine nature of the association. One may also infer confidently the homogenous nature of the samples used.

Association analysis across PGM1 (sites A and B).
Having assessed pairwise association within site B, patterns of LD in the three population samples across the entire PGM1 gene were examined by incorporating genotype data from both sites A and B (Fig. 1). In this way, one could deduce if some regions appeared to be clustered and whether there are indications that this genomic region may actually be part of a wider array of variable association. LD between all polymorphic sites was assessed (210 associations) using the ldmax resource to generate D'-values and corresponding P-values for the Caucasian, Chinese and Vietnamese populations. Figures 3A–C illustrates the global patterns of association across 55.5 kb within the human PGM1 gene revealing an overall alternating pattern of LD. Three pronounced blocks of LD emerge from the extended analysis in all three populations (Figs 3A–C), which appears to be stronger in the Caucasian sample (Fig. 3A). Two LD blocks, defined as a contiguous set of markers in which the average D' is greater than a predetermined threshold (in this case |D'|≥0.9) and where the markers within each block are highly associated (P<0.05), were identified in site B. The first LD block is bounded by the markers N2F and BunF, and the second by spanning markers BunJ to BunR. A further LD block was identified in site A, encompassed by markers M1 and M4 (Figs 3A–C). These LD blocks appear in consecutive blocks (Figs 3A–C) and correlate with significant P-values (P<0.05). All three LD blocks are interrupted by regions of persistently low LD: (i) between markers N7F and BunA, (ii) around markers BunR and 2/1 and (iii) in the previously described region M2 and M6 (site A) (33).



View larger version (43K):
[in this window]
[in a new window]
  		Figure 3. Global patterns of association across the human PGM1 gene. |D'| and corresponding P-values are shown for all pairwise associations (all markers in site A and B) in the (A) Caucasian, (B) Chinese and (C) Vietnamese population samples. Three pronounced blocks of high LD are evident (P<0.05); these blocks are shown as purple on the illustration. The LD blocks are interrupted by regions of low LD (indicated in blue). The pairwise adjacent association plot (separating the P-value and D' illustration) highlights the breakdown of LD at three points (indicated by arrows). The three population samples exhibit very similar patterns of marker association across PGM1.

 
There is good concordance in the association data between the three population samples examined, with only slight variation (probably due to population-specific factors). This result is in agreement with opinions that the same sign and magnitude of disequilibrium would be expected in different populations, unlike the effect of random drift (41). This suggests that the clustering of hotspots within the PGM1 gene is not likely to be artifactual, again liberating these observations from any stochastic influences. An unusual association (P<0.05) between marker N1 and the +/– site emerged in the Caucasian (Fig. 3A) and Vietnamese (Fig. 3C) samples. The two sites are located at opposite ends of the gene separated by 50 kb and span the recombination hotspots. There may be several explanations for this observation; it is possible that this is a statistical artefact, it may be a result of one marker with low m.a.f. (N1 is a haplotype marker) claiming LD with +/–, a recent mutation on a haplotype which sustained high population frequency and remained unaffected by recombination, or perhaps a result of selection. An example of such an anomaly has been reported in the human TAP2 region, where markers spanning the hotspot are in absolute LD (42).

The identification of intense hotspots within PGM1 also allows the assignment of major LD blocks, moreover global LD (association with markers outside these blocks) with these markers is also high. In summary, the association data presented here implicate the presence of three domains within 55.5 kb of the PGM1 gene, where LD is significantly high. Two domains of high LD extend 658 bp and 8.7 kb in site B, and a third region in site A which is 3 kb in size. Furthermore, these blocks of high LD are separated by recombinationally active regions extending 1, 2.7 and 3.3 kb. Hence, the pattern of association within PGM1 appears to portray traits of a region decomposed into smaller blocks of LD, where increased recombination activity has disrupted the ancestral chromosome structure.

PGM1 haplotype analysis
The patterns of LD observed in the pairwise association analysis were then subjected to further examination in order to determine whether common haplotypes are represented within LD blocks (which could then be deemed ‘haplotype blocks’) and to attempt to precisely define the boundaries of LD blocks and recombination hotspots.

The initial haplotype analysis used molecularly phased data to construct chromosomes with limited ambiguity, in order to obtain an overview of the common haplotypes that exist between N1 and +/– sites. Common haplotypes (in 38 independent chromosomes) were deduced from family data on the basis of correct segregation observed at all informative loci, thereby placing confidence in the assignment of haplotypes. These haplotypes were then compared with the common haplotypes deduced by Arlequin in the three different populations. Since the population data are unphased and inferred from the Arlequin program, it was important to verify the common haplotypes by means of comparison with limited ambiguity haplotypes from CEPH families and elucidate common haplotypes.

The five most common haplotypes in the three populations and their frequencies, derived from Arlequin, are shown in Table 3. The results are promising in that the common haplotypes deduced from the family data are among the five most common population sample haplotypes (data not shown). The fact that the five common haplotypes represent <30% of the chromosomes sampled is in contrast to results from other studies where 3–5 common haplotypes represent 80–95% of the chromosomes in regions ranging from 10–100 kb (3,6,43). Thus, the initial overview of haplotypes across the PGM1 gene is consistent with the LD results, depicting a region with great variability.


View this table:
[in this window]
[in a new window]
  		Table 3. Five common haplotypes across PGM1 in the three population samples
 
However, embedded within this region of variable LD also lie apparently conserved blocks of genomic DNA, found in the three population groups. It was of interest to see if these LD blocks are in fact haplotype blocks. Several definitions have been proposed to identify haplotype blocks from LD data (44). Morris et al. (7) examined blocks of LD across chromosome 19, where a block was defined to consist of at least three SNPs, with a stringent threshold of |D'|≥0.9 for each pair. Another study examined 40 human genomes, using a block definition of |D'|>0.8 among all pairs (45). Others have defined haplotype blocks on the basis of chromosome coverage, where a minimum number of contiguous SNPs account for the majority of common haplotypes (43) or a reduced level of haplotype diversity (6).

Looking closely at the haplotypes in the proposed LD block N2F–BunF, 80–90% of the chromosomes in all three populations are represented by two haplotypes (CGG and TAG). A similar observation is made for the proposed second haplotype block spanning across the markers BunJ, Tsp, Hpy, Bts and BunR, where ~80% of the chromosomes are represented by two common haplotypes, GGGCC and GTTTT (Table 4). This is good evidence for the presence of two haplotype blocks, one extending 8.7 kb and the other 658 bp in site B. Moreover, closer inspection of the haplotypes in the site A region revealed yet another haplotype block encompassing the markers M1, M3 and M4, where again 80–90% of the chromosomes are accounted for by one of two haplotypes (AAA and BBB; Table 4). Thus, three haplotype blocks, limited in diversity and almost always represented by a common haplotype, exist within the PGM1 gene. It is also immediately apparent that the regions between N1 and BunA do not display such common haplotypes. In fact, all possible two marker haplotypes between N7F and BunA are observed in the Caucasians and are amongst the most common haplotypes, implying that considerable recombination has taken place in this region. LD between proposed blocks and individual markers was also examined (data not shown), indicating the consistent lack of association with markers BunA, 2/1 and M5. This strongly suggests the presence of three recombination hotspots around these markers and supports the notion that this region on chromosome 1p contains segments with limited recombination (haplotype blocks) interspersed by recombination hotspots. A consistent haplotypic profile is observed in all populations, thereby strengthening this concept.


View this table:
[in this window]
[in a new window]
  		Table 4. Haplotype blocks within human PGM1
 
Since PGM1 is erratic in its association, one would expect a higher density of markers to identify regions of significant association. However, detecting three haplotype blocks with identical spatial arrangement (representing 80–90% of the chromosomes) in the different population groups indicates that a smaller number of SNPs could detect the common haplotype structure across the region.

Candidate sequences associated with recombination
Having explored association between markers and discovering common haplotypes, it was of interest to examine the distribution and presence of sequences potentially involved in recombination (46). Other investigations (N.A. Rana et al., manuscript in preparation) based on comparisons between genetic and physical maps and family studies (where most recombinant chromosomes fine mapped to site B) urged the examination of this region. With the recent availability of sequence information, it was possible to explore this region for candidate motifs that have previously been associated with recombination activity.

A 47 kb genomic segment containing the site B sequence was examined for various candidate motifs including the bacterial {chi} sequence, translin binding sites, consensus origin of replication (OR) sequences and consensus PUR elements together with other motifs associated with the recombination machinery in other organisms (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 5. Site B search for candidate sequence signals proposed to be involved in recombination
 
Motifs within site B
The motifs identified within site B are illustrated in Figure 4. Several consensus sequences associated with origins of replication have been proposed to be involved in some way in mediating or enhancing recombination, namely through its effect on the duplex DNA molecule by causing instability. These include the 21 bp OR consensus sequence (24), matrix/scaffold attachment regions (MARs/SARs) (47,48) and a 16 bp consensus sequence for the purine-rich PUR element (27). The 21 bp consensus OR sequence (5'-AAT TTT TAT TGA TGA TAA ATT-3') was found at the 5' end (~5 kb) of the gene. It is interesting to note that this consensus OR sequence lies within the region where a collapse of LD is observed, given the probability of finding the sequence in 40 kb of random DNA is 0.0011 (24). Other signals thought to be involved in conferring helical instability, such as MARs/SARs and ARS, were found in the vicinity of this 21 bp consensus sequence (Fig. 4). This fits well with the criteria for potential initiation of replication regions suggested by Dobbs et al. (24). The 21 bp consensus OR sequence has previously been identified upstream of the hotspot in site A (34) and together with the PUR element has also been reported in the vicinity of the ß-globin gene cluster recombination hotspot (27).



View larger version (21K):
[in this window]
[in a new window]
  		Figure 4. An amalgamation of approaches characterizing recombination activity within PGM1. The main plot depicts adjacent allelic association across sites A and B within human PGM1, highlighting regions of high LD and, conversely, equilibrium. Polymorphic markers within the gene are displayed below the plot. Several sequence motifs identified within site B are also indicated, as are recombinant chromosomes derived from family studies. The circles represent markers between which the crossover has taken place. It is interesting to note the overlap between crossovers, low levels of LD and the presence of sequence motifs.

 
Specific sequence motifs shown to be related to recombination in other organisms have recently been reported in humans, within the ß-globin gene cluster (27) and the MHC cluster (46). In humans, various sequence parameters such as interspersed repetitive elements (49), the sequence d(GT)n (50), GC content and CpG islands have been correlated with variation in recombination rates. These studies have been performed in specific regions (46), on a chromosome-specific basis (50) and on a genome-wide basis (49,51). The sequence (GT)21 and (GT)26 dinucleotide repeats were found, within site B (~11 kb) and one just outside (40 kb). Again these lie in close proximity to crossover regions in recombinant families and also coincide with regions of low LD (Fig. 4). (GT)>12 have also been found to be significantly associated with recombination hotspots within the MHC cluster (46). Such repeats have been shown to enhance homologous recombination (52).

Of the repetitive elements, 36 SINEs (comprising 15 Alus and 21 MIRs), 13 LINEs and 12 LTR elements were found. Four perfect {chi} sequences (expected to occur by chance once every 28.5 kb) were also identified in site B (38.5 kb), and may be involved in rendering chromatin accessible to the recombination machinery, together with a combination of complex factors (53; Fig. 4).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 REFERENCES
 
The aspects explored within the PGM1 gene in this study include (i) pairwise allelic association in sites A and B, (ii) the overall haplotype diversity across the gene, (iii) intra haplotype diversity levels, (iv) the subsequent summation of LD and haplotype analysis to direct the inter association of haplotype blocks with markers and (v) the exploration of sequences associated with enhanced recombination. The various collated approaches facilitate the description of the diversity and association patterns between distinct genomic segments within the recombinationally variable PGM1 gene. Figure 4 illustrates the amalgamated approaches in characterizing PGM1.

The region is clearly highly diverse, implicated initially by LD and confirmed by long-range haplotype analysis, in that <30% of the chromosomes are represented by the five most common haplotypes. This is in contrast to the results of other studies that have indicated limited haplotype diversity where three or four haplotypes account for 80–95% of all observed chromosomes (3,54). The definition for haplotype blocks adopted in this study combines proposals from other investigators (44), namely regions displaying limited diversity, where pairwise disequilibrium values are greater than a certain threshold (|D'|>0.8) and also the suggestion that at least 80% of the chromosomes are accounted for by two or more haplotypes (43). The size of the smallest haplotype block (658 bp) between BunJ and BunR, where the haplotype diversity is low and LD between constituent markers making up the block is high, may be dictated by the fact that it lies between two recombinogenic regions (Fig. 4). A haplotype block defined by Wang et al. (55) suggests no recombination within blocks, but allows for recombination between blocks and predicts block size to decrease with increasing recombination. This tempts one to speculate on the extent of haplotype blocks observed within PGM1, and taken together with the association analyses allows a more defined picture of the haplotypic variation underlying this region. This region may be considered to comprise clusters of hotspots (i) around markers N7F and BunA (2.7 kb), (ii) between markers BunR and 2/1 (3.3 kb) and (iii) in the region encompassed by M2 and M6 (1 kb). The size of the hotspots is in agreement with the range identified within the MHC (12) and those shown by LD maps (56). These hotspots surround three haplotype blocks which extend from (i) markers N2F and BunF (8.7 kb), (ii) between BunJ and BunR (658 bp) and (iii) span markers M1–M4 (3 kb). The extent of haplotype blocks correlate with block sizes that have been reported in other studies (6,57) (Fig. 4).

The association observed between the N1 and +/– sites in the Caucasian and Vietnamese samples, despite intervening markers showing variable block-like patterns, has been described for other regions. In some non-African studies, where long-range LD blocks have been predicted, the blocks may actually break down into smaller ones. This is exemplified in a recent study where long-range LD was initially observed through the haplotypic association of two genes whose alleles are significantly associated with hypersensitivity reaction (HSR) in patients undergoing treatment for HIV infection (58,59). The two markers, TNF and HLA-B (both in the HLA region), located 200 kb apart showed considerable allelic association (|D'|>0.95). Closer analysis revealed that despite being in high LD with each other, additional SNPs within the intervening region divided the 200 kb segment into three haplotype blocks (60). It may be possible that the PGM1 region is part of a much larger cluster composed of many hotspots and coldspots. Long-range LD also highlights the limited power of SNPs and/or their density when used in such studies. The limited power of sparsely selected markers indicating long-range LD can be overcome by dense coverage of the region of interest, as highlighted in this study.

The HSR study also highlights that extended LD (beyond the confines of a block) can be detected readily in pharmacogenetic association experiments, sometimes across a large region and several haplotype blocks (60). However, the frequency of most adverse events such as genetic association with disease are low, and so the use of common haplotypes may not detect these associations.

Many studies promote the idea that genome-wide patterns of LD are primarily dictated by the presence of recombination hotspots (9,57). This notion is supported by the consistency observed in the spatial arrangement of haplotype blocks within the PGM1 locus between different populations. Besides the hotspot model, there have been proposals that genetic drift may act as an alternative mechanism for the genesis of block-like patterns of LD, thereby necessitating multiple ethnic group studies. Arguably though, if genetic drift alone created blocks then, one would not expect the same region in different populations to show block-like patterns. We have demonstrated through the study of three populations that genetic drift is not the primary cause of LD variation within the PGM1 gene.

Several sequence motifs that may promote preferential recombination have been investigated, and are of greater interest if recombination hotspots do in fact influence the formation of haplotype blocks. Identifying such sequences could promote creation of a useful hotspot map, and simultaneously contribute to a haplotype map (HapMap).

The site B region within PGM1 was investigated for such sequences, revealing the presence of a 21 bp consensus OR sequence and signals conferring helical instability, e.g. MARs/SARs and ARS (Table 5). In addition to the dinucleotide repeats (GT)21 and (GT)26, four perfect chi sequences were also observed. An interesting feature to note is the overlap between some sequence motifs within site B and regions of low LD (Fig. 4). Crossovers identified in molecularly phased data (580 informative meioses) have also mapped preferentially to regions where LD collapses, strengthening the meiotic recombination hotspot notion (Rana et al., manuscript in preparation). Preferential crossing over to hotspot regions has also been described in the ß-globin gene cluster (27). The extent of meiotic crossover regions are shown in Figure 4, further fine mapping of these crossovers has been limited by intervening marker homozygosity (Rana et al., manuscript in preparation). Sophisticated experimental techniques such as sperm typing can enhance the resolution of such meiotic crossovers, thereby allowing a more precise assessment of crossover correlation to regions of low LD (12,61).

It is intriguing that putative signals for recombination have been found in such active regions, but it remains unclear as to what the overall relationship of such sequences is to recombination. Strict concordance between some of these sequences/motifs to hotspots and discordance in coldspots is an initial means to acknowledge the complexities in our genome, but is not sufficient to resolve the true nature of such discrepancies within genomic segments. In order to circumvent this limitation, a larger number of hotspots need to be discovered and examined in parallel. Further analysis of genomic regions is required to evaluate these findings and to reveal exactly how particular motifs may influence recombination. Availability of more completed sequence, higher resolution of genetic maps and the use of multiple regression analyses will certainly advance our understanding of sequences that are associated with regional recombination rates. With further effort to characterize these regions genome-wide, one would expect a clearer understanding of relevant sequences predisposing chromatin instability. In addition to the efforts of the human genome project, genomes of many other species are being sequenced and it will be important to compare sequence motifs and rates of recombination for PGM1 across species.

The human PGM1 gene highlights the fundamental processes of genome instability, genetic polymorphism and sequence evolution. Little is known about the processes of reciprocal meiotic recombination in humans, and so PGM1 serves as a model system to explore this important biological phenomenon. Knowledge of the patterns and frequency of recombination are important for many purposes such as understanding the factors underlying LD, mapping quantitative trait loci and eventually unravelling the intricacy of recombination pathways. The mapping of hotspots and characterization of regions with variable LD has many implications in the eventual delineation of complex disorders, predominantly in tailoring the search for candidate gene/s involved in such disorders. One of the applications for generating an LD map of the genome is to identify haplotype blocks where a small fraction of SNPs decipher a large proportion of haplotypes. However, when creating LD maps one must consider SNP ascertainment and any bias this may create (62). Consequently, the human HapMap project has been initiated in an attempt to provide better understanding and application of association studies of common diseases in different populations, motivated by the suggestion that blocks exist ubiquitously throughout the genome. Thus, there is a necessity to identify and explore in detail, hot- and coldspots in several genomic regions before success is achieved in resolving the basis of common complex disease.

Putative hotspots for meiotic recombination within the human genome include the ß-globin gene cluster (27,63) and the MHC region (64,65). Recent proposals based on studies in the MHC region suggest that there is at least one hotspot every 0.8 Mb of DNA, or potentially 3000–4000 intense hotspots throughout the human genome (46). Despite such predictions, it remains unclear as to what dictates local mutation rates and why hotspots for recombination occur, but this mechanism is extremely important in the strategies adopted to find association. At present there appear to be no unifying theories to predict the occurrence of recombination hotspots, a greater understanding of this can only be achieved by the continual dissection and fine structure analysis of several genomic regions, such as those presented in this report.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
REFERENCES
 
Samples
Samples from three population groups were investigated: Caucasian (n=169), Chinese (n=222) and Vietnamese (n=187). The samples used in this study are the same set used by Yip et al. (33). In addition, 95 DNA samples from ECACC were included in the Caucasian samples (total n=264).

PCR
PCR was performed with Taq polymerase (Bioline) using the manufacturers 10x buffer containing 10 mM Tris–HCl, 50 mM KCL, 15 mM MgCl2 and 0.1% (w/v) non-ionic detergent. Standard PCR amplification was carried out in a total reaction volume of 25 µl (50 µl for restriction digest) containing: 1x manufacturers reaction buffer, 0.2 mM of each dNTP, 0.1–0.2 µm of each primer, 50–100 ng of template DNA and 0.5 unit Taq DNA polymerase. Typical cycling conditions were carried out: denaturation of 95°C for 4 min followed by cycling parameters of denaturing at 94°C for 25 s, annealing at the appropriate temperature for 25 s (available on request) and extending for 30 s at 72°C. The cycle number varied between 30 and 40, depending on the template and particular methodology. A final extension step at 72°C for 5 m followed. Amplification was carried out using DNA Thermal cycler TCI (Perkin Elmer) and Techne Genius PCR thermocyclers.

Direct automated sequencing of PCR products
Automated DNA sequencing was performed on an ABI 3100 DNA sequencer (Applied Biosystems) using the ABI PRISM® AmpliTaq® DNA polymerase FS, Dye Terminator Cycle sequencing Ready Reaction Kit or the BigDyeTM terminator kit.

Mutation detection
Genotyping: SNPs were initially genotyped and validated in CEPH families, using genotyping methods outlined subsequently. Data were reviewed independently for each SNP. Those showing clear segregation, consistent with Mendelian inheritance patterns and having no significant (P<0.01) departure from HWE were used for subsequent analysis. Of the 18 markers, 30% were excluded from further analysis, as they were uninformative in the CEPH samples tested (i.e. monomorphic), failing to yield valid genotyping scores. A final set of 12 SNPs were informative in CEPH samples and were developed further into functional genotyping assays.

Restriction enzyme digestion of DNA
Restriction digests were performed using 2–4 units of enzyme per µg of DNA template, typically in a 50 µl reaction including 5 µl of 10x restriction buffer (NEB) and made up to volume with distilled water. Reactions were incubated for a minimum of 2 h at the recommended temperature (usually 37°C) and the products of digestion resolved by agarose gel electrophoresis. The restriction endonucleases/enzymes used in this study were: BtsI, HpyII and Tsp509I.

The TaqmanTM assay
The TaqmanTM assay (P.E. Applied Biosystems) is an automated method for SNP typing and was employed to analyse a selection of SNPs in this study. TaqmanTM PCR amplification of 100–200 bp specific genomic segments; 5 µl PCR reaction mixture consisted of 5–10 ng/µl DNA, a ‘probe and primer mix’ (Ppmix, PE ABI) and Taqman 2x Universal PCR mastermix (UMM, PE ABI). The UMM mixture is optimized for Taqman reactions and contains: AmpliTaq Gold® DNA polymerase, AmpErase UNG, passive Reference I and optimized buffer components polymerase. Primers and probes were designed and supplied by ABI-using Assay By Design (Applied Biosystems). PCR thermal cycling conditions were: 50°C for 2 min followed by an initial denaturation of 95°C for 10 min, cycling parameters; 92°C for 15 seconds and annealing temperature of 60°C for 1 min for 40 cycles. This was performed on a Perkin Elmer thermal cycler. Subsequent analysis was carried out according to manufacturer's instructions (PE Applied Biosystems, facility at HGMP Cambridge).

The Wave® DNA fragment analysis system
Standard PCR was used to amplify marker N7F, samples were then analysed on the Wave system (Transgenomic) with buffers A (0.1 M TEAA) and B (0.1 M TEAA in 25% acetonitrile) at a flow rate of 0.9 ml/min (TEAA supplied by Tansgenomic, Inc., San Jose), following the manufacturer's instructions. PCR optimization was facilitated by using proof reading Pfu polymerase to minimize PCR induced mutations (Pfu Instruction manual, Stratagene, LA Jolla, CA, USA). The melting profile and gradient selection for a fragment are determined by a Wave® utility software package (Transgenomic), and optimum temperature is obtained by an incremental temperature scan.

Computer aided and statistical analyses
DNA analysis.
The software program DNASTAR (DNASTAR, Inc. Wisconsin, USA), which includes a range of utilities, was used as a tool to analyse DNA sequences. This analysis included editing of sequences, alignments, restriction maps and motif search. NIX analysis (http://www.rfcgr.mrc.ac.uk/) was used to explore the region for sequence motifs. References are listed in Table 5.

Haplotype analysis.
Several utility programs employ the EM algorithm (65) to estimate haplotype frequencies, and a variety were used in this study for different purposes. The EH program (http://linkage.rockefeller.edu/ott/eh.htm) was used to estimate haplotype frequencies (in site B) from population data, both with allelic association (Hypothesis H1) and independent association (Hypothesis H0). The output files were then manipulated for the calculation of LD parameters (information available on request). The Arlequin program (http://lgb.unige.ch/arlequin/) was predominantly used to examine haplotype data across PGM1 in the three populations.

Allelic association analysis.
LD between SNP pairs was measured using the absolute value of Lewontin's D' (|D'|), where in a large sample the absolute value of D'=1 is reflective of complete LD and 0 corresponds to a state of complete equilibrium (66). The ldmax program (http:www.sph.umich.edu/csg/abecasis/ldmax.html) was used to derive LD parameter (|D'|-values) for pairwise association for all markers in sites A and B.

SNP pairs were classified according to the estimated |D'|-value and corresponding P-value, i.e. SNP pairs were classified as being in high LD where the |D'|-value was ≥0.9 (P<0.05), whereas those showing significant collapse in LD were categorized by a |D'|-value <0.3.

Haplotype block.
A haplotype block was defined operationally as any series of three or more markers, where |D'|-values for constituent markers ≥0.9 (P<0.05) and where ≥80% of chromosomes sampled were represented by at most three haplotypes.

Repeat sequence analysis.
Repeat Locater was used to screen DNA sequences for tandem repeat blocks (kindly provided by A. Webster). RepeatMasker (http://www.hgmp.mrc.ac.uk) was used to analyse DNA sequences for interspersed repeats.

Visual tools.
The graphical online display of disequilibrium (GOLD) is a visualization tool enabling depiction of a continuous global view of LD for the sites A and B regions within the PGM1 gene. Input files were derived from ldmax and formatted for GOLD display (http://www.sph.umich.edu/csg/abecasis/GOLD/).


   ACKNOWLEDGEMENTS
 
We gratefully acknowledge Shea Ping Yip for access to previous data. We thank Andrew Dearlove, Justin Brooking and Frank Dudbridge at the HGMP Resource Centre for assistance with Taqman and statistical advice. Deepest gratitude to Reshma Patel and Saima Rana for their invaluable support in preparation of the manuscript, and many thanks to Suba Poopalsundaram and Rhian Gwilliam for proofreading. We also thank Shomi Bhattacharya and David Hopkinson for providing resources to complete the study. This project was funded by the Medical Research Council.


   FOOTNOTES
 
* To whom correspondence should be addressed at: Department of Molecular Genetics, Institute of Ophthalmology, University College London, 11–43 Bath Street, London EC1V 9EL, UK. Tel: +44 2076086971; Fax: +44 2076086863; Email: naheed_rana{at}yahoo.co.uk


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 

  1. Abecasis, G.R., Noguchi, E., Heinzmann, A., Traherne, J.A., Bhattacharyya, S., Leaves, N.I., Anderson, G.G., Zhang, Y., Lench, N.J., Carey, A. et al. (2001) Extent and distribution of linkage disequilibrium in three genomic regions. Am. J. Hum. Genet., 68, 191–197.[CrossRef][ISI][Medline]

  2. Reich, D.E., Cargill, M., Bolk, S., Ireland, J., Sabeti, P.C., Richter, D.J., Lavery, T., Kouyoumjian, R., Farhadian, S.F., Ward, R. et al. (2001) Linkage disequilibrium in the human genome. Nature, 411, 199–204.[CrossRef][Medline]

  3. Gabriel, S.B., Schaffner, S.F., Nguyen, H., Moore, J.M., Roy, J., Blumenstiel, B., Higgins, J., DeFelice, M., Lochner, A., Faggart, M. et al. (2002) The structure of haplotype blocks in the human genome. Science, 296, 2225–2229.[Abstract/Free Full Text]

  4. Shifman, S., Kuypers, J., Kokoris, M., Yakir, B. and Darvasi, A. (2003) Linkage disequilibrium patterns of the human genome across populations. Hum. Mol. Genet., 12, 771–776.[Abstract/Free Full Text]

  5. Cardon, L.R. and Abecasis, G.R. (2003) Using haplotype blocks to map human complex trait loci. Trends Genet., 19, 135–140.[CrossRef][ISI][Medline]

  6. Daly, M.J., Rioux, J.D., Schaffner, S.F., Hudson, T.J. and Lander, E.S. (2001) High-resolution haplotype structure in the human genome. Nat. Genet., 29, 229–232.[CrossRef][ISI][Medline]

  7. Morris, A., Lawrence, R. and Cardon, L. (2002) SNP ascertainment is a crucial factor in determining LD block structure. Am. J. Hum. Genet., 71 (suppl.), 220.

  8. Perola, M., Koivisto, M., Varilo, T., Hennah, W., Ekeluna, J., Lugg, M., Peltonen, L., Ukkonen, E. and Mannila, H. (2002) A method to find and compare the strength of haplotype block boundaries and its application in populations with different settlement histories. Am. J. Hum. Genet., 71 (suppl.), 219.

  9. Kauppi, L., Sajantila, A. and Jeffreys, A.J. (2003) Recombination hotspots rather than population history dominate linkage disequilibrium in the MHC class II region. Hum. Mol. Genet., 12, 33–40.[Abstract/Free Full Text]

  10. Menashe, I., Man, O., Lancet, D. and Gilad, Y. (2002) Population differences in haplotype structure within a human olfactory receptor gene cluster. Hum. Mol. Genet., 11, 1381–1390.[Abstract/Free Full Text]

  11. Stead, J.D., Hurles, M.E. and Jeffreys, A.J. (2003) Global haplotype diversity in the human insulin gene region. Genome Res., 13, 2101–2111.[Abstract/Free Full Text]

  12. Jeffreys, A.J., Kauppi, L. and Neumann, R. (2001) Intensely punctate meiotic recombination in the class II region of the major histocompatibility complex. Nat. Genet., 29, 217–222.[CrossRef][ISI][Medline]

  13. Greenwood, T.A., Rana, B.K. and Schork, N.J. (2004) Human haplotype block sizes are negatively correlated with recombination rates. Genome Res., 14, 1358–1361.[Abstract/Free Full Text]

  14. Smith, G.R. (1988) Homologous recombination in procaryotes. Microbiol. Rev., 52, 1–28.[Free Full Text]

  15. Schuchert, P., Langsford, M., Kaslin, E. and Kohli, J. (1991) A specific DNA sequence is required for high frequency of recombination in the ade6 gene of fission yeast. EMBO J., 10, 2157–2163.[ISI][Medline]

  16. Gale, J.M., Tobey, R.A. and D'Anna, J.A. (1992) Localization and DNA sequence of a replication origin in the rhodopsin gene locus of Chinese hamster cells. J. Mol. Biol., 224, 343–358.[CrossRef][ISI][Medline]

  17. Fox, M.E., Yamada, T., Ohta, K. and Smith, G.R. (2000) A family of cAMP-response-element-related DNA sequences with meiotic recombination hotspot activity in Schizosaccharomyces pombe. Genetics, 156, 59–68.[Abstract/Free Full Text]

  18. Gerton, J.L., DeRisi, J., Shroff, R., Lichten, M., Brown, P.O. and Petes, T.D. (2000) Inaugural article: global mapping of meiotic recombination hotspots and coldspots in the yeast Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA, 97, 11383–11390.[Abstract/Free Full Text]

  19. Nasar, F., Jankowski, C. and Nag, D.K. (2000) Long palindromic sequences induce double-strand breaks during meiosis in yeast. Mol. Cell. Biol., 20, 3449–3458.[Abstract/Free Full Text]

  20. Edelmann, W., Kroger, B., Goller, M. and Horak, I. (1989) A recombination hotspot in the LTR of a mouse retrotransposon identified in an in vitro system. Cell, 57, 937–946.[CrossRef][ISI][Medline]

  21. Zimmerer, E.J. and Passmore, H.C. (1991) Structural and genetic properties of the Eb recombinational hotspot in the mouse. Immunogenetics, 33, 132–140.[ISI][Medline]

  22. Jeffreys, A.J., Wilson, V. and Thein, S.L. (1985) Hypervariable ‘minisatellite’ regions in human DNA. Nature, 314, 67–73.[CrossRef][Medline]

  23. Bergemann, A.D. and Johnson, E.M. (1992) The HeLa Pur factor binds single-stranded DNA at a specific element conserved in gene flanking regions and origins of DNA replication. Mol. Cell. Biol., 12, 1257–1265.[Abstract/Free Full Text]

  24. Dobbs, D.L., Shaiu, W.L. and Benbow, R.M. (1994) Modular sequence elements associated with origin regions in eukaryotic chromosomal DNA. Nucl. Acids Res., 22, 2479–2489.[Abstract/Free Full Text]

  25. Aoki, K., Suzuki, K., Sugano, T., Tasaka, T., Nakahara, K., Kuge, O., Omori, A. and Kasai, M. (1995) A novel gene, Translin, encodes a recombination hotspot binding protein associated with chromosomal translocations. Nat. Genet., 10, 167–174.[ISI][Medline]

  26. Rooney, S.M. and Moore, P.D. (1995) Antiparallel, intramolecular triplex DNA stimulates homologous recombination in human cells. Proc. Natl Acad. Sci. USA, 92, 2141–2144.[Abstract/Free Full Text]

  27. Smith, R.A., Ho, P.J., Clegg, J.B., Kidd, J.R. and Thein, S.L. (1998) Recombination breakpoints in the human beta-globin gene cluster. Blood, 92, 4415–4421.[Abstract/Free Full Text]

  28. Roychoudry, A.K. and Nei, M. (1988) Human Polymorphic Genes. Oxford University Press, New York.

  29. Putt, W., Ives, J.H., Hollyoake, M., Hopkinson, D.A., Whitehouse, D.B. and Edwards, Y.H. (1993) Phosphoglucomutase 1: a gene with two promoters and a duplicated first exon. Biochem. J., 296 (Pt 2), 417–422.[Medline]

  30. Carter, N.D., West, C.M., Emes, E., Parkin, B. and Marshall, W.H. (1979) Phosphoglucomutase polymorphism detected by isoelectric focusing: gene frequencies, evolution and linkage. Ann. Hum. Biol., 6, 221–230.[CrossRef][ISI][Medline]

  31. Takahashi, N., Neel, J.V., Satoh, C., Nishizaki, J. and Masunari, N. (1982) A phylogeny for the principal alleles of the human phosphoglucomutase-1 locus. Proc. Natl Acad. Sci. USA, 79, 6636–6640.[Abstract/Free Full Text]

  32. Wetterling, G. (1990) Intragenic recombination within the PGM1 system. In Polesky, H.F. and Mayr, W.R. (eds), Advances in Forensic Haemogenetics. Springer-Verlag, Berlin, Vol. 3, pp. 218–221.

  33. Yip, S.P., Putt, W., Hopkinson, D.A. and Whitehouse, D.B. (1999) Identification and characterisation of polymorphisms in human phosphoglucomutase (PGM1). Ann. Hum. Genet., 63 (Pt 2), 129–140.[CrossRef][ISI][Medline]

  34. Yip, S.P., Lovegrove, J.U., Rana, N.A., Hopkinson, D.A. and Whitehouse, D.B. (1999) Mapping recombination hotspots in human phosphoglucomutase (PGM1). Hum. Mol. Genet., 8, 1699–1706.[Abstract/Free Full Text]

  35. Lander, E.S. and Schork, N.J. (1994) Genetic dissection of complex traits. Science, 265, 2037–2048.[Abstract/Free Full Text]

  36. Zavattari, P., Deidda, E., Whalen, M., Lampis, R., Mulargia, A., Loddo, M., Eaves, I., Mastio, G., Todd, J.A. and Cucca, F. (2000) Major factors influencing linkage disequilibrium by analysis of different chromosome regions in distinct populations: demography, chromosome recombination frequency and selection. Hum. Mol. Genet., 9, 2947–2957.[Abstract/Free Full Text]

  37. Ke, X., Hunt, S., Tapper, W., Lawrence, R., Stavrides, G., Ghori, J., Whittaker, P., Collins, A., Morris, A.P., Bentley, D. et al. (2004) The impact of SNP density on fine-scale patterns of linkage disequilibrium. Hum. Mol. Genet., 13, 577–588.[Abstract/Free Full Text]

  38. Schulze, T.G., Zhang, K., Chen, Y.S., Akula, N., Sun, F. and McMahon, F.J. (2004) Defining haplotype blocks and tag single-nucleotide polymorphisms in the human genome. Hum. Mol. Genet., 13, 335–342.[Abstract/Free Full Text]

  39. Lewontin, R.C. and Kojima, K. (1960) The evolutionary dynamics of complex polymorphisms. Evolution, 14, 458–472.[Medline]

  40. Lewontin, R.C. (1964) The interaction of selection and linkage. I. General considerations; heterotic models. Genetics, 49, 49–67.[Free Full Text]

  41. Hill, W.G. and Robertson, A. (1968) The effects of inbreeding at loci with heterozygote advantage. Genetics, 60, 615–628.[Free Full Text]

  42. Jeffreys, A.J., Ritchie, A. and Neumann, R. (2000) High resolution analysis of haplotype diversity and meiotic crossover in the human TAP2 recombination hotspot. Hum. Mol. Genet., 9, 725–733.[Abstract/Free Full Text]

  43. Patil, N., Berno, A.J., Hinds, D.A., Barrett, W.A., Doshi, J.M., Hacker, C.R., Kautzer, C.R., Lee, D.H., Marjoribanks, C., McDonough, D.P. et al. (2001) Blocks of limited haplotype diversity revealed by high-resolution scanning of human chromosome 21. Science, 294, 1719–1723.[Abstract/Free Full Text]

  44. Wall, J.D. and Pritchard, J.K. (2003) Haplotype blocks and linkage disequilibrium in the human genome. Nat. Rev. Genet., 4, 587–597.[CrossRef][ISI][Medline]

  45. Zwick, M.E., Cutler, D.J., Lin., S. and Chakravarti, A. (2002) An empirical estimate of the number of SNPs required for a whole genome association study. Am. J. Hum. Genet, 71 (suppl.), 219.

  46. Cullen, M., Perfetto, S.P., Klitz, W., Nelson, G. and Carrington, M. (2002) High-resolution patterns of meiotic recombination across the human major histocompatibility complex. Am. J. Hum. Genet., 71, 759–776.[CrossRef][ISI][Medline]

  47. van Drunen, C.M., Sewalt, R.G., Oosterling, R.W., Weisbeek, P.J., Smeekens, S.C. and van Driel, R. (1999) A bipartite sequence element associated with matrix/scaffold attachment regions. Nucl. Acids Res., 27, 2924–2930.[Abstract/Free Full Text]

  48. Santucci-Darmanin, S., Neyton, S., Lespinasse, F., Saunieres, A., Gaudray, P. and Paquis-Flucklinger, V. (2002) The DNA mismatch-repair MLH3 protein interacts with MSH4 in meiotic cells, supporting a role for this MutL homolog in mammalian meiotic recombination. Hum. Mol. Genet., 11, 1697–1706.[Abstract/Free Full Text]

  49. Yu, A., Zhao, C., Fan, Y., Jang, W., Mungall, A.J., Deloukas, P., Olsen, A., Doggett, N.A., Ghebranious, N., Broman, K.W. et al. (2001) Comparison of human genetic and sequence-based physical maps. Nature, 409, 951–953.[CrossRef][Medline]

  50. Majewski, J. and Ott, J. (2000) GT repeats are associated with recombination on human chromosome 22. Genome Res., 10, 1108–1114.[Abstract/Free Full Text]

  51. Kong, A., Gudbjartsson, D.F., Sainz, J., Jonsdottir, G.M., Gudjonsson, S.A., Richardsson, B., Sigurdardottir, S., Barnard, J., Hallbeck, B., Masson, G. et al. (2002) A high-resolution recombination map of the human genome. Nat. Genet., 31, 241–247.[CrossRef][ISI][Medline]

  52. Wahls, W.P., Wallace, L.J. and Moore, P.D. (1990) The Z-DNA motif d(TG)30 promotes reception of information during gene conversion events while stimulating homologous recombination in human cells in culture. Mol. Cell. Biol., 10, 785–793.[Abstract/Free Full Text]

  53. Smith, G.R., Kunes, S.M., Schultz, D.W., Taylor, A. and Triman, K.L. (1981) Structure of chi hotspots of generalised recombination. Cell, 24, 429–436.[CrossRef][ISI][Medline]

  54. Drysdale, C.M., McGraw, D.W., Stack, C.B., Stephens, J.C., Judson, R.S., Nandabalan, K., Arnold, K., Ruano, G. and Liggett, S.B. (2000) Complex promoter and coding region beta 2-adrenergic receptor haplotypes alter receptor expression and predict in vivo responsiveness. Proc. Natl Acad. Sci. USA, 97, 10483–10488.[Abstract/Free Full Text]

  55. Wang, N., Akey, J.M., Zhang., K., Chakraborty, R. and Lin, J. (2002) Haplotype block definitions and their relationship with population history, recombination rate and, SNP density. Am. J. Hum. Genet., 71 (suppl.), 219.

  56. Collins, A. (2002) Linkage disequilibrium maps: ‘blocks’ without haplotyping and application to disease mapping. Am. J. Hum. Genet., 71 (suppl.), 220.

  57. Zhang, K., Akey, J.M., Wang, N., Xiong, M., Chakraborty, R. and Lin, J. (2002) Randomly distributed recombination may generate block-like patterns of linkage disequilibrium: an act of genetic drift. Am. J. Hum. Genet., 71 (suppl.), 220.

  58. Hetherington, S., Hughes, A.R., Mosteller, M., Shortino, D., Baker, K.L., Spreen, W., Lai, E., Davies, K., Handley, A., Dow, D.J. et al. (2002) Genetic variations in HLA-B region and hypersensitivity reactions to abacavir. Lancet, 359, 1121–1122.[CrossRef]