Publications

The evolution of genomic imprinting is viewed as a problem of economic optimization that is analyzed using the tools of evolutionary game theory. We specifically consider genetic conflicts over the allocation of maternal resources between present and future offspring. Five sets of genes, with the same interests within sets but distinct interests between sets, are considered as agents: maternal alleles (Mater), paternal alleles (Pater), unimprinted offspring alleles (Filius), and imprinted offspring alleles of maternal and paternal origin (Matris and Patris). Fitness functions are derived for each agent and the parameter space in which there is conflict defined. Three potential conflicts are considered: between mother and offspring (Mater v.s. Filius); between alleles of maternal and paternal origin within offspring (Matris v.s. Patris) and between mothers and the paternally derived alleles of offspring (Mater v.s. Patris).

Inactivation of expression of the paternal allele at two maternally silent imprinted loci has recently been reported to diminish the quality of care that female mice lavish on their offspring. This suggests that there can be disagreement between the maternally and paternally derived genomes of mothers over how much care for offspring is appropriate, with the paternally derived genome favoring greater care. The reason for such disagreement is not obvious because the maternally and paternally derived alleles at a locus have equal probabilities of being transmitted to each of the mother's ova and, therefore, would appear to have equal interests in a mother's offspring. However, if a female mates with a related male, her two alleles may have different probabilities of being present in the sperm that fertilize her ova. Natural selection can favor silencing of the maternally derived allele at a locus that enhances the quality of maternal care if the average patrilineal relatedness between a female and her mates decreases more rapidly than the average matrilineal relatedness. Just such an asymmetrical decrease in relatedness over time would be expected in a structured population in which patrilineal inbreeding is more common than matrilineal inbreeding.

A conceptual model is presented here in which the birth weight distribution is decomposed into a distribution of target weights and a distribution of perturbations from the target. The target weight is the adaptive goal of fetal development. In the simplest model, perinatal mortality is independent of variation in target weight and determined solely by the magnitude of the perturbation of birth weight from the target. In this model, mortality risk is concentrated in the tails of the birth weight distribution. A difference between populations in their distributions of target weights will be associated with a corresponding shift in their curves of weight-specific risk, without any difference between the populations in overall risk. In this model, risk would be reduced by decreasing the variance of the distribution of perturbations. The model is discussed in the context of the so-called "paradoxes of low birth weight."

The kinship theory of genomic imprinting predicts that imprinted genes have effects on asymmetric kin (relatives with different degrees of matrilineal and patrilineal relatedness). The most important interaction with such a relative is a child's interaction with its mother. Therefore, the study of imprinted genes and their phenotypic effects promises to provide insights into the evolution of mother-child relations. Prader-Willi syndrome (PWS) is caused by the absence of expression of genes at 15q11-q13 that are normally expressed only when paternally derived. The kinship theory predicts that children with PWS will fail to express behaviors that have increased mothers' costs of child-rearing. Our analysis focuses on aspects of the PWS phenotype that affect appetite and feeding. Immediately after birth, children with PWS have little appetite and are usually unable to suckle, but at some stage (usually within the first 2 years) they develop a voracious appetite and an obsession with food. We conjecture that this change in appetite reflects evolutionary forces associated with weaning. Immediately after birth, when a child is completely dependent on the breast, poor appetite reduced maternal costs. However, once a child was able to consume supplemental foods, maternal costs would have been reduced by children with increased, nonfastidious appetites.

Parent-specific gene expression (genomic imprinting) is an evolutionary puzzle because it forgoes an important advantage of diploidy--protection against the effects of deleterious recessive mutations. Three hypotheses claim to have found a countervailing selective advantage of parent-specific expression. Imprinting is proposed to have evolved because it enhances evolvability in a changing environment, protects females against the ravages of invasive trophoblast, or because natural selection acts differently on genes of maternal and paternal origin in interactions among kin. The last hypothesis has received the most extensive theoretical development and seems the best supported by the properties of known imprinted genes. However, the hypothesis is yet to provide a compelling explanation for many examples of imprinting.

The kinship theory of genomic imprinting has explained parent-specific gene expression as the outcome of an evolutionary conflict between the two alleles at a diploid locus of an offspring over how much to demand from parents. Previous models have predicted that maternally derived (madumnal) alleles will be silent at demand-enhancing loci, while paternally derived (padumnal) alleles will be silent at demand-suppressing loci, but these models have not considered the evolution of trans-acting modifiers that are expressed in parents and influence imprinted expression in offspring. We show that such modifiers will sometimes be selected to reactivate the silent padumnal allele at a demand-suppressing locus but will not be selected to reactivate the silent madumnal allele at a demand-enhancing locus. Therefore, imprinting of demand-suppressing loci is predicted to be less evolutionarily stable than imprinting of demand-enhancing loci.

We present a model that considers the coevolution of genomic imprinting at a growth factor locus and an antagonistic growth suppressor locus. With respect to the two loci considered independently, our model makes the familiar predictions that an imprinted growth factor locus will only be expressed from the paternally derived allele and an imprinted growth suppressor locus only from the maternally derived allele. In addition, our coevolutionary model allows us to make predictions regarding the sequence of evolutionary events necessary for generating such a system. We conclude that imprinting at the growth factor locus preceded the evolution of growth suppressor function at the second locus, which in turn preceded imprinting at that locus. We then discuss the consistency of these predictions with currently available comparative data on the insulin-like growth factor 2 insulin-like growth factor 2 receptor system of mammals.

Some genes carry a record of the sex of the gene's carrier in the previous generation that influences the gene's expression in this generation. This additional information can result in intragenomic conflicts between an individual's maternally and paternally derived alleles over behaviors that affect relatives with whom the individual has different degrees of maternal and paternal relatedness. Asymmetries of relatedness can arise because of sex-biased dispersal. For example, if females remain in their natal group and males disperse, female members of a group will all be matrilineal relatives, but may have unrelated fathers. Sex-linked inheritance creates an evolutionary bias in favor of social groups that trace descent through the homogametic sex. This bias has a positive and negative aspect. The positive aspect is increased relatedness among siblings of the homogametic sex. The negative aspect is the lack of sex-linked relatedness between parents and offspring of the heterogametic sex.

Genomic imprinting has been proposed to evolve when a gene's expression has fitness consequences for individuals with different coefficients of matrilineal and patrilineal relatedness, especially in the context of competition between offspring for maternal resources. Previous models have focused on pre-emptive hierarchies, where conflict arises with respect to resource allocation between present and future offspring. Here we present a model in which imprinting arises from scramble competition within litters. The model predicts paternal-specific expression of a gene that increases an offspring's fractional share of resources but reduces the size of the resource pool, and maternal-specific expression of a gene with opposite effects. These predictions parallel the observation in economic models that individuals tend to underprovide public goods, and that the magnitude of this shortfall increases with the number of individuals in the group. Maternally derived alleles are more willing than their paternally derived counterparts to contribute to public goods because they have a smaller effective group size.

Comparative gene mapping and chromosome painting permit the tentative reconstruction of ancestral karyotypes. The modern human karyotype is proposed to differ from that of the most recent common ancestor of catarrhine primates by two major rearrangements. The first was the fission of an ancestral chromosome to produce the homologues of human chromosomes 14 and 15. This fission occurred before the divergence of gibbons from humans and other apes. The second was the fusion of two ancestral chromosomes to form human chromosome 2. This fusion occurred after the divergence of humans and chimpanzees. Moving further back in time, homologues of human chromosomes 3 and 21 were formed by the fission of an ancestral linkage group that combined loci of both human chromosomes, whereas homologues of human chromosomes 12 and 22 were formed by a reciprocal translocation between two ancestral chromosomes. Both events occurred at some time after our most recent common ancestor with lemurs. Less direct evidence suggests that the short and long arms of human chromosomes 8, 16 and 19 were unlinked in this ancestor. Finally, the most recent common ancestor of primates and artiodactyls is proposed to have possessed a chromosome that combined loci from human chromosomes 4 and 8p, a chromosome that combined loci from human chromosomes 16q and 19q, and a chromosome that combined loci from human chromosomes 2p and 20.