The leading theoretical explanation for the evolution of genomic imprinting is the Kinship Theory, which was proposed by David Haig here at Harvard in 1989. The Kinship theory is founded on work by Bob Trivers and Bill Hamilton, who first introduced the highly influential concepts of ‘parent-offspring conflict’ and ‘inclusive fitness’ into evolutionary biology, respectively. Here I post another fine excerpt from Brady Weissbourd’s undergraduate thesis that gives a nice introduction to the Kinship Theory. Thanks again Brady!!
An Introduction to the Kinship Theory of Genomic Imprinting
Haig’s Kinship Theory requires an understanding of modern evolutionary theory, which at the level of the gene has yielded the surprising finding that, within an individual, genes can be in conflict with each other (Haig, 1989; Haig, 2000). Richard Dawkins (1976) argues that evolution should be viewed from the level of the gene: that individuals are vessels created by genes to serve the agenda of the genes. That agenda is propagation. Indeed, from the original “replicators” to complex genomes, throughout evolution these gene-carrying machines have become increasingly complex and efficient at conveying a gene – or set of genes – to the next generation. In this sense, each gene acts in its own best interest to increase its chances of making it to the next round (Dawkins, 1976).
Despite the language of intent attributed to these genes, it is the blind power of natural selection that causes “selfish” genes to propagate in a population. For a gene to be positively selected for and therefore be successful a population, it needs to confer a benefit to itself and not necessarily to the individual, though the interests of the individual and of the genes tend to align. There are numerous of examples of “selfish genes” that survive at the expense of the individual in both plants and animals. For example, so-called segregation distortion genes (also called meiotic drive) increase their representation in the next generation often at a cost to the individual, such as lower sperm viability (Taylor and Ingvarsson, 2003).
Gene selection may also explain complex cooperative, and even so-called “altruistic” behavior. A gene that promotes the fitness of other individuals who have high probabilities of carrying that gene will also proliferate in the population, so long as the cost to self is less than the benefit to the relative, weighted by the degree of relatedness (rB>C) (Trivers, 1971; Trivers, 1974). That is, a gene acting “selfishly” can incur a cost to its reproductive opportunities so long as it promotes the passing on of that gene to the next generation via relatives. The higher the probability that a relative carries the gene, the more likely social and cooperative behaviors are to be enriched at that locus (Burt and Trivers, 2006).
However, as David Haig said, “my mothers kin are not my father’s kin” (Haig, 1997). Thus, just as relatedness coefficients can promote “altruism”, intragenomic conflict can arise in populations with asymmetric relatedness between the genes of mothers and fathers in an offspring (Burt and Trivers, 2006; Haig, 2000). Trivers’ understanding of relatedness coefficients and his equation for kin selection (rB>C) in asymmetrically related kin groups yields a framework for understanding the evolution of intragenomic conflict, and thus, genomic imprinting (Haig and Westoby, 1989/91; Haig and Trivers, 1995; Trivers, 1971; Trivers, 1974). However, asymmetric relatedness between maternal and paternal lineages exists outside of the taxa that display genomic imprinting, which highlights the special circumstances necessary for imprinting to evolve.
The fundamental origin of conflict that results in imprinting is that the father is not related to the mother (assuming the absence of inbreeding). In therian mating systems, particularly in those mammals with an invasive placenta, the process of carrying, birthing, and nursing young bears heavy costs to the mother. For each unit the mother invests in a current offspring (e.g. in the form of time or energy), there is a significant opportunity cost to the mother’s future reproductive options. For example, this opportunity cost can manifest as reduced energy available for the current or next litter, or a lower chance of survival until the next breeding season. Mothers will therefore favor some optimal investment in the current generation that leaves her ample energy to survive and invest in future litters (Haig, 2000; Haig, 2002; Haig, 2003).
The optimal maternal investment from the point of view of the father is quite different. It is often the case that females will mate with multiple males, both within and across breeding seasons. Indeed, litters of pups (or even human twins (Ambach et al, 2000)) may have multiple fathers. Therefore, given that fathers and mothers are not genetically related, the future reproductive interests of the mother are of little to no consequence to the father after the birth of their offspring. The maternal optimum takes into account the survival of each member of the current generation as well as future reproductive fitness, while fathers care only for offspring in the current generation that carry a paternally derived allele. This results in fathers preferring a higher level of investment in the current generation, specifically in any one offspring carrying a paternal allele (Haig, 2000; Haig, 2002; Haig and Wilkins, 2003).
It is then expected that paternally inherited (padumnal) genes will upregulate offspring behaviors and physiological processes that tax the mother beyond her optimum, while maternally inherited (madumnal) genes will evolve to counteract this strategy in an evolutionary “arms race”. An offspring’s madumnal genes are related to the offspring by 1, and related to the mother by 1 (retrospectively, they are maternally inherited), and therefore are selected to support maternal interests. Similarly, padumnal genes are related to offspring and father by 1, which creates an immediate conflict in interest between madumnal and padumnal genes in their interactions with kin. As these madumnal and padumnal genes come together to form a temporary vessel, they have a common interest in the reproductive potential of the organism they cohabitate. However, if a gene carries an imprint of which lineage it represents, that gene may be selected to serve the marginally different interests of the paternal versus maternal faction (Haig, 2000; Haig and Wharton, 2003).
A gene will therefore become imprinted when madumnal and padumnal genes have different optima that serve their different interests. The evolutionarily stable strategy is then to silence the gene that prefers the lower value, and express the other at its optimum. As genes in the offspring are selected to demand more resources, the corresponding maternal genes are silenced. Conversely, maternal genes suppressing resource acquisition will approach maternal optimums, accompanied by paternal silencing (Haig and Wharton, 2003).
This inherent difference in relatedness between maternally and paternally inherited genes may also apply to sibling interactions. Madumnal genes in offspring are confident of their relatedness to siblings, and mothers are confident of their equal relatedness to each pup, so selection for cooperative behaviors and equal provisioning of pups is predicted. Conversely, paternity certainty is low, and padumnal genes that enhance the intake of maternal resources, even at a cost to siblings, are predicted to spread through the population. It is also generally the case that mammals live in systems where matrilineal groups tend to stay together, with male dispersal (Haig, 1997; Haig, 2001). Therefore, conflict between “cooperative” behaviors and “selfish” behaviors will be represented by maternal and paternal lineages respectively. It is then expected that the degree of conflict will increase with increased promiscuity in the population, and decrease with increased monogamy (Burt and Trivers, 2006; Haig, 2000; Haig and Wharton, 2003).
This theory comprehensively explains the evolution of imprinting in placental mammals and provides a compelling framework for the study of the physiological and behavioral effects of imprinted genes. Further, the formulation of kinship theory that takes into account how imprinting can arise from the asymmetries of relatedness inherent in mammalian social groups – not just in the relationship between mothers and their offspring – suggests that imprinted genes likely have evolved to control complex social interactions and behaviors (Haig, 2000).
Evidence from known imprinted phenomena strongly supports the kinship theory of imprinting. The preliminary studies of failed development in parthenotes demonstrate that in androgenotes, there is enhanced development of extraembryonic tissue – those responsible for the acquisition of resources from the mother – with impaired growth of the embryo itself. Conversely, gynogenotes have severe deficits in extraembryonic tissue, with relatively normal embryonic growth (Surani et al, 1984). Further, insulin-like growth factor 2 (IGF2) is known to promote fetal growth while IGF2 receptor (IGF2R) reduces concentrations of IGF2, reducing growth. As predicted, IGF2 is a paternally expressed gene (PEG) and IGF2R is a maternally expressed gene (MEG). Other PEGs such as Kcnq1ot1 have been shown to increase growth while a number of MEGs, such as cdkn1c, H19, PHLDA2 are known to suppress it (Haig, 2004).
Brandon Weissbourd, Harvard Undergraduate Honors Thesis, 2009