Technical Comments

Response to Comment on “Parasite Selection for Immunogenetic Optimality”

Science  13 Feb 2004:
Vol. 303, Issue 5660, pp. 957b
DOI: 10.1126/science.1093355

The superior parasite resistance of stickleback major histocompatibility complex (MHC) genotypes with an intermediate number of about six alleles has been demonstrated under both natural (1) and experimental conditions (2, 3). The current debate does not concern the validity of these findings, but rather the machinery of evolution and population genetics maintaining or eliminating the observed variation around the optimum.

Hedrick (4) used a simple model to show that optimizing selection should result in a monomorphic population with respect to allele numbers. However, a stable monomorphic population can exist only under unrealistic assumptions implicit in the model. It could consist of individuals all carrying totally distinct haplotypes (Fig. 1A), which requires an infinite number of alleles. Otherwise, as soon as two gametic haplotypes share alleles, recombination leads to offspring with a suboptimal number of alleles. Alternatively, monomorphic populations might be maintained when all genotypes are identical (i.e., six homozygous loci, each bearing a different allele; Fig. 1B). Under this scenario, the six different alleles must be adapted to the local parasite fauna by associations with pathogens (5). This would be possible, but only if parasite pressure remains constant. On the other hand, given the high number of alleles in stickleback populations, haplotypes are obviously not fixed (1).

Fig. 1.

Theoretical parental genotypes depicting conditions that would have to be met to guarantee a stable, monomorphic population of optimal genotypes. (A) All haplotypes consist of three loci (A, B, and C), which are completely distinct with respect to the alleles they bear. This requires an infinite number of alleles. (B) All individuals are identical and homozygous for their six loci (A through F).

An explicit assumption in Hedrick's model (4) is that all alleles confer similar resistance toward infection. Because in reality pathogens vary spatially and temporally (in influenza pandemics, for example), the quality of alleles is essential for predicting individual fitness. Such “Red Queen” dynamics are thought to maintain the typical polymorphism of MHC genes found in sticklebacks (3, 6, 7) and other vertebrates (8). The time lag generated by host-parasite coevolution is likely to create an array of suboptimal genotypes. Furthermore, in multi-locus systems such as stickleback MHC, polymorphism in allele number will be maintained by linkage disequilibrium (LD) for extended periods of time, as pointed out by Lewontin in (9). LD will be constantly renewed by parasite selection because different parasites will select for haplotypes carrying specific MHC alleles that confer resistance (5)—even if the total number of alleles they bear is deviating from the optimum (i.e., trade-off between allele quality and optimal number of alleles). LD will also create a reservoir of alleles associated with resistance against diseases that temporarily disappeared from the population. Reoccurring parasites will therefore not require new mutations (10).

As soon as one accepts that multiple MHC alleles can be selectively maintained within populations, one also buys its consequences for the individuals. Under random mating, the degree of polymorphism at each locus dictates an individual's number of alleles. In order to escape the stochastic nature of Mendelian inheritance (i.e., segregational load), there is selection on females to optimize offspring MHC diversity with reference to their own MHC genotype (7, 11). Since females can potentially perceive only genotypes, and not haplotypes, their mate choice will be imperfect. For example, the mating between two optimal genotypes with six alleles each might result in suboptimal genotypes when the haplotypes of each parent consist of packages (gametes) of two and four different alleles. The expected distribution of filial genotypes would then result in only 50% of genotypes with the optimal number of six alleles. In this sense, the Mendelian mixing of haplotypes can be regarded as a constraint of sexual recombination in diploids. It will inevitably create a population of individuals that differ in their number of MHC alleles. LD and Mendelian inheritance only maintain individual variation in the number of MHC alleles if allelic variation on the population level is maintained by ever changing parasites. We have thus shown that a number of basic population genetic processes will always generate variance in allele numbers. These processes should be considered in future models.


Related Content

Navigate This Article