Evolution of X chromosome inactivation in therian mammals

Abstract

In therian (eutherian and marsupial) mammals, X gene dosage between XY males and XX females is equalized by transcriptional silencing of one X chromosome in the somatic cells of females; a process called X chromosome inactivation (XCI). There are fundamental differences between the random, stable XCI in eutherians, and the paternal and partial system in marsupials, which can be exploited to dissect the mechanisms and evolution of XCI. A striking molecular difference is the involvement of DNA methylation in eutherian XCI, which seems, from a few previous studies, not to be mirrored by marsupials. I therefore searched for sex-specific DNA methylation of X-borne genes in marsupials using mass array bisulfate sequencing (Chapter 2). I tested methylation differences of six X-borne genes in four different tissues of male and female tammar wallaby and identified sex-specific methylation differences in 5' CpG islands of two genes. However, I observed no correlation between methylation status and inactivation of genes. I conclude that DNA methylation plays no role in maintaining inactivation (at least of the loci I examined) in marsupials. To gain new insights into the mechanisms and evolution of XCI I studied gene transcription at the level of individual nuclei (Chapters 3 and 4). I used RNA in situ hybridization to assess whether one or both alleles of X-borne loci were transcribed in fibroblasts from three marsupial model species (tammar wallaby, Tasmanian devil and South American opossum) and three eutherian species (human, mouse and elephant). In marsupials I observed that different X loci have a characteristic frequency of expression from one or both alleles, indicating that it is the probability of expression (either 1X-active or 2X-active), rather than the rate of transcription that contributes to the partial expression of the paternal allele. I observed no polarity that might reveal an X inactivation centre. Most orthologous genes on the conserved region of the eutherian X (XCR) were completely inactivated. However, genes that escape XCI on the evolutionarily distinct added region of the eutherian X (XAR, autosomal in marsupials) were 1X- or 2X-active in reproducible frequencies, analogous to genes on the marsupial X. I therefore proposed that marsupial XCI, and the evolutionarily distinct XAR of the eutherian X, retain features of an ancient incomplete silencing mechanism, which was progressively stabilized (perhaps by the evolution of XIST) on the eutherian XCR. I then tested the hypothesis that incomplete stochastic monoallelic expression represents an ancient mechanism to control gene expression, which was exapted into genomic imprinting, as well as XCI. Genomic imprinting results in parent-specific monoallelic expression, and involves epigenetic modifications similar to genes on the inactive X. I therefore assessed transcription from nine autosomal-imprinted loci in mouse, and their tammar wallaby orthologues. Surprisingly, I observed biallelic expression of imprinted genes in a proportion of mouse and tammar nuclei. As for genes on the marsupial X and the eutherian XAR (as well as genes on the independently evolved platypus sex chromosomes), transcription appeared to be stochastic.