Regulatory links between imprinted genes: evolutionary predictions and consequences

Abstract

Genomic imprinting is essential for development and growth and plays diverse roles in physiology and behaviour. Imprinted genes have traditionally been studied in isolation or in clusters with respect to cis-acting modes of gene regulation, both from a mechanistic and evolutionary point of view. Recent studies in mammals, however, reveal that imprinted genes are often co-regulated and are part of a gene network involved in the control of cellular proliferation and differentiation. Moreover, a subset of imprinted genes acts in trans on the expression of other imprinted genes. Numerous studies have modulated levels of imprinted gene expression to explore phenotypic and gene regulatory consequences. Increasingly, the applied genome-wide approaches highlight how perturbation of one imprinted gene may affect other maternally or paternally expressed genes. Here, we discuss these novel findings and consider evolutionary theories that offer a rationale for such intricate interactions among imprinted genes. An evolutionary view of these trans-regulatory effects provides a novel interpretation of the logic of gene networks within species and has implications for the origin of reproductive isolation between species.

Keywords: evolution; gene network; genomic imprinting; trans regulation.

Figure 1.

Predicted cis and trans effects in light of the kinship and co-adaptation theories. (a) Under the kinship theory, MEGs and PEGs have different phenotypic optima (shown in light blue ovals). Shown here for ease of presentation are the growth optima, but other phenotypes—most notably phenotypes associated with social behaviour—are subject to similar conflicts between the two parental genomes. As predicted by the theory [22], MEGs typically inhibit growth and PEGs enhance it. Shown here is an MEG that achieves growth inhibition via an ncRNA both in cis (within the defined region) and in trans (outside of the defined locus). In cis, the ncRNA is involved in silencing the neighbouring gene, a PEG, thereby reducing the total level of expression of a growth enhancing gene. In trans, the ncRNA reduces the total expression level of a different PEG, further reducing the total gene expression level of growth enhancing genes. Similar effects on gene expression are predicted for MEGs that encode proteins. (b) Under the co-adaptation theory [–25], imprinting is favoured by natural selection to coordinate the expression of genes that interact epistatically. Epistatic interactions between alleles within a parental haplotype are fitter, on average, than epistatic interactions that would be produced between parental haplotypes. This results from selection in the prior generation, which produces an excess of haplotypes with good epistatic combinations. Imprinting can, therefore, spread from one imprinted gene to its epistatic partners so as to increase the chance of a high-fitness interaction. This spread is more likely for linked genes, leading to a prediction of imprinted genes with cis-regulatory effects on other imprinted genes. Additionally, unlinked genes are subject to the same selective pressure, giving rise to the predicted parental-origin-specific pattern of trans-regulatory effects.

Figure 2.

(a) Effects of IPW lncRNA in trans. IPW, a PEG in the PWS imprinted domain of human chromosome 15, produces an lncRNA that has inhibitory effects on the expression of several MEGs in the imprinted DLK1-DIO3 domain on chromosome 14. IPW appears to achieve its trans regulation by recruiting G9A, a lysine methyltransferase that modifies chromatin, to the DLK1-DIO3 region [41]. (b) The Dlk1-Dio3 locus on mouse chromosome 12 is host to imprinted genes with trans effects. The expression of Rtl1, a PEG, is regulated by miRNAs processed from the Rtl1as, a maternally expressed antisense RNA. The expression of these miRNAs reduces the expression of Rtl1 [42]. Similarly, in postnatal muscle in the mouse, a cluster of maternally expressed miRNAs located further downstream appears to negatively affect Dlk1 expression [43].

Figure 3.

Roles of Zac1 and H19 in the imprinted gene network. (a) The imprinted gene network, as defined by Gabory et al. [52] and Varrault et al. [19]. MEGs are shown in pink, PEGs in blue. Grey circles indicate that the network also consists of non-imprinted genes; three are shown for illustrative purposes only. The network also consists of other imprinted genes, but for clarity only those with the strongest interactions are shown. (b) The deletion of the paternally inherited allele of Zac1 in fetal mouse liver affects the expression levels of several genes. Circle sizes indicate relative expression levels. Only genes with significant expression changes are labelled. (c) Deletion of the maternally inherited allele of H19 in fetal mouse muscle also influences the expression of several genes in the IGN.