NATURAL SELECTION AND THE EVOLUTIONARY HISTORY OF MAJOR HISTOCOMPATIBILITY COMPLEX LOCI

Austin L. Hughes and Meredith Yeager

Department of Biology and Institute of Molecular Evolutionary Genetics, The Pennsylvania State University, University Park PA 16802 USA

Received 5/11/98 Accepted 5/22/98 5. THE ROLE OF RECOMBINATION

5.1 Intralocus Recombination in the Class I MHC

Alignment of DNA sequences of MHC genes has suggested the possibility that recombination has played a role in the evolution of this gene family. Both recombination among alleles at a locus (intralocus or interallelic recomnination) and recombination between loci (interlocus recombination) have been proposed as mechanisms in the diversification of these genes. However, these hypotheses have been controversial. First, it has been noted that many of the cases that have been attributed to recombination might be just as easily explained by convergent or parallel evolution (38). Natural selection can greatly increase the probability of convergent or parallel evolution over the neutral case. Because of the positive selection acting on PBR codons of polymorphic MHC loci, convergent or parallel evolution would not be surprising in this case, and indeed there is evidence of convergent evolution of similar sequence motifs in the class II PBR of mammals of different orders (39).

In addition, most proposed cases of recombination have not been subjected to any statistical test but have only involved visual inspection of sequences. It is desirable to have some way of discriminating statistically between the amount of similarity between two sequences that might be observed by chance and the degree of similarity that indicates a past recombination event. Unfortunately, although a number of methods have been proposed to test for recombination in molecular sequence data, there may be problems with applying them to the MHC because the balancing selection at these loci often causes violations of the methods’ underlying assumptions.

We can distinguish two types of inter-allelic recombination: (i) large-scale recombination involves the exchange of a long stretch of sequence between two alleles, such as may occur by an ordinary crossing-over mechanism; and (ii) small-scale recombination involves the exchange of a short sequence motif or "mini-cassette" and may involve a "gene conversion" mechanism. A number of fairly reliable methods can be used to test for large-scale recombination, including construction of phylogenetic trees for different gene regions and estimation of numbers of nucleotide substitutions per site in different gene regions (40). This approach cannot be used in the case of small-scale recombination. Available methods, such as that of Stephens (41), identify clusters of sites shared between sequences. If these sites are nonsynonymous, such a cluster might result from convergent evolution at the amino acid level, rather than recombination. Thus, this method cannot always discriminate between the hypotheses of recombination and gene conversion. Nonetheless, in some cases, recombination "mini-cassettes" involving synonymous sites have been identified, giving strong support to the hypothesis of interallelic recombination (40).

Although it may be difficult to decide in some specific cases whether or not interallelic recombination has been involved, studies employing a variety of methods have been able to detect major trends. One type of trend that has been observed is a difference between loci. In the case of the class I MHC of humans, it is clear that interallelic recombination has played a far more prominent role in the history of the HLA-B locus than of either HLA-A or HLA-C (40,42,43). It is so far uncertain whether this difference has occurred because the actual rate of recombination is higher at the HLA-B locus than at the other two loci or whether more recombinants at the HLA-B locus have been selectively favored and thus have increased in the population. Some evidence favors the latter hypothesis. First, putative recombination events that have occurred at human class I loci disproportionately involve the PBR codons, as is consistent with their being selectively favored. Second, although HLA-C is closely related to HLA-C, it has a low level of recombination like HLA-A. This might not be expected if the high observed rate of recombination at HLA-B is simply a consequence of some aspect of its sequence, since HLA-B and HLA-C are similar in sequence (43).

Immunologists may have a difficult time understanding the fact that interallelic recombination leads to homogenization within loci in class I introns (23), but can enhanced diversity in exons. However, this is exactly what population genetics theory predicts in the case of balancing selection; it certainly would not be true in the absence of such selection. In introns, recombination breaks up linkage of intron sites with nonsynonymous PBR sites that are under balancing selection, thus allowing genetic drift to reduce sequence diversity in introns. In exons, polymorphism has been selectively favored, evidently because of the wide immune surveillance of heterozygotes, and over time this selection has given rise to numerous quite distinct alleles. A recombinant between two very different alleles may create a new form of the PBR, which may confer an immediate selective advantage; and such a favored recombinant will increase in frequency. Thus, recombination alone does not enhance diversity; only recombination combined with balancing selection as seen in the case of the MHC PBR.

5.2 Intralocus Recombination the the Class II MHC

So far, the levels of recombination at MHC class II loci have not been studied as extensively as those at class I loci. We developed a simple method to address this problem. For a given pair of alleles, we computed the proportion of nucleotide difference (p) within a sliding window of 10 nucleotide positions in exon 2 of the gene, which includes the PBR codons. The coefficient of variation (C.V.) of p was used as an index of the non-homogeneity of the difference between the two sequences. This in turn provided an indication of the rate of past recombination. For example, two sequences very similar in one region but very different in another would have a high C.V. of p; and such a pattern might have resulted from recombination. However, non-homogeneity of p could conceivably be caused by factors other than recombination. In the case of the MHC, the main such factor is natural selection enhancing the rate of non-synonymous substitution in the PBR codons. Therefore, we removed the PBR codons prior to this analysis.

The results (figure 5) show marked differences among the class II b chain loci of human and mouse with respect to the C.V. of p and thus the putative level of past recombination. In humans, these results suggest that there has been a much higher level of recombination at HLA-DRB1 and HLA-DPB1 than at HLA-DQB1 (figure 5). In the mouse, they suggest that there has been a much greater rate of recombination at H2-Ab than at H2-Eb (figure 5). These conclusions are consistent with those of previous studies using other methods. For example, phylogenetic analysis of different portions of exon 2 suggested that recombination has occurred at high levels in the human HLA-DRB1 locus (44) and the mouse H2-Ab locus (45).

Figure 5. Coefficient of variation (C.V.) in proportion of nucleotide difference in a sliding window of 10 in exon 2 (excluding PBR codons) in pairwise comparisons among human (HLA-) and mouse (H2-) class II b chain alleles. For each locus, the data are represented as the percentage of comparisons with a C.V. greater than 1.5 and the proportion of comparisons with a C.V < 1.5. Loci with greater percentages of comparisons with C.V. > 1.5 show greater incidence of interallelic recombination.

5.3 Interlocus Recombination

There is much less evidence of interlocus recombination in the MHC than of intralocus recombination. However, as in most multi-gene families, sequence analyses have suggested that such events have occurred in the past. For example the human class Ia HLA-A locus arose from a recombination between (1) a 5’ region (exon 1 through exon 3) of a gene related to the other human class Ia loci HLA-B and –C; and (2) a 3’ region (from intron 3 to the end of the gene) from a gene closely related to the human class I pseudogene HLA-70 (46).

A persistent theme in the history of MHC biology has been the search for amazing or unusual molecular mechanisms to explain the high polymorphism of these loci. For example, it was proposed that MHC loci have a high mutation rate (47). Now it is known that in fact the mutation rate at these loci is lower than the mammalian average (9). Similarly, it was argued that some unusual mechanism of interlocus recombination must exist to create diversity at these loci (48), but sequence analyses have failed to support this hypothesis (49). Rather, the MHC seems to represent a rather ordinary multi-gene family in most respects. The only really dramatic difference is that the MHC includes some loci that are subject to balancing (most likely overdominant) selection of a type that has persisted since the origin of the MHC itself early in vertebrate history and most likely will persist as long as vertebrates are exposed to parasites of any sort. All unique or unusual features of the MHC genes are best understood as consequences of that selection.