Evolutionary Analyses Suggest A Function Of MxB Immunity Proteins ...

Dynamism of Mx genes in eutherian mammals

Many antiviral gene families have undergone dynamic, lineage-specific changes in copy number, presumably as a mechanism for gaining new antiviral specificities without losing existing functions [37]. In contrast to other antiviral gene families, previous studies have found that Mx gene copy number is relatively static [25,38]. However, this apparent stasis may be misleading. For instance, a recent report found that Mx genes have been lost in Odontoceti cetaceans (toothed whales) [39]. To more comprehensively determine the evolutionary dynamics of Mx genes in mammals, we performed phylogenetic analyses of mammalian Mx paralogs from at least one representative of all sequenced mammalian orders. We found shared synteny of the Mx locus throughout terrestrial vertebrates (Fig 2A). Both human and mouse genomes encode two Mx genes, but as previously described [38], rodents have lost the MxB-like gene and instead encode two MxA orthologs (Mx1 and Mx2). Platypus genomes encode two Mx genes that both appear ancestral to the eutherian MxA and MxB lineages; we term them here MxAB1 and MxAB2 (Fig 2). We were unable to identify any Mx genes in any of three marsupial genome sequences (tammar wallaby, opossum, Tasmanian devil) and suggest that Mx gene(s) were lost from the common marsupial ancestor. We therefore infer that MxA and MxB genes diverged at or just prior to the origin of the eutherian mammal lineage.

Download:

• PPTPowerPoint slide
• PNGlarger image
• TIFForiginal image

Fig 2. The complement and phylogenetic relationship of Mx genes in eutherian mammals.

(A) The orientation and relative position of genes in the Mx locus from representative mammalian orders and selected outgroups (marsupials, platypus, Reptilia and Aves) are represented by black (MxB-like), white (MxA-like), hatched (Mx ancestral), dark gray (FAM3B) or light gray (BACE2/TMPRSS2) pentagons. Slashes indicate instances in which the Mx locus is present on distinct genomic scaffolds. Cases in which inferences are based on one representative species are indicated by parentheses. The inferred ancestral Mx locus is present in species from the orders Carnivora, Cetartiodactyla, Primate and Tubulidentata (aardvark), with instances of Mx gain and loss (grey “X” represent pseudogenes, "?" symbols represent complete loss) indicated. See also S2 Fig. Altered gene order and orientation in lagomorphs and rodents suggests locus reorganization. The cartoon phylogeny (left) shows the species tree [40], with the inferred origin of the eutherian mammal Mx duplication shown as a black circle. (B) A phylogeny of Mx genes from representative mammals and selected outgroups reveals that MxA-like and MxB-like genes form two major clades. Bootstrap support for nodes greater than 50 are shown. Rodent Mx paralogs are indicated with bold branches to highlight that rodents have two MxA-like genes, rather than a single gene in each clade. Please refer to S3 Dataset.

https://doi.org/10.1371/journal.ppat.1005304.g002

Within eutherian mammals, our phylogenetic analyses revealed surprisingly poor resolution; many nodes have less than 50% bootstrap support and some discordance at well-supported nodes of the mammalian phylogeny (Fig 2B) [40]. These discrepancies are not entirely unexpected for rapidly evolving antiviral genes, and likely reflect complex evolutionary histories of Mx genes in multiple mammalian lineages (see below). Nevertheless, we were able to conclude that most eutherian mammalian genomes encode both MxA-like (orthologous to human MxA) and MxB-like (orthologous to human MxB) genes. Further, elephants encode two closely related intact MxA genes in addition to MxB; this MxA duplication appears to have occurred since divergence from the sister order Macroscelidea (e.g., elephant shrew) (Fig 2). With the exception of the loss of both MxA and MxB genes in toothed whales [39], we found evidence for an intact MxA gene in all surveyed eutherian mammal species. In contrast, we found that MxB has been lost at least three additional, independent times in Rodentia, Felidae and Xenarthra (Fig 2 and S2 Fig). Therefore, MxB loss has been tolerated on multiple occasions during eutherian mammal evolution despite the fact that its importance as an antiviral factor has been experimentally demonstrated.

Phylogenetic and synteny analyses allowed us to propose a hypothetical scenario for the loss of the MxB gene in mouse. We found that the ancestral configuration of the mammalian Mx locus was Bace2(+);Fam3B(+);MxB(+);MxA(+);Tmprss2(-) (Fig 2A). In the rabbit genome (Lagomorpha, an outgroup to Rodentia), an inversion occurred in the Mx locus such that MxB and Fam3B have opposite orientations relative to the ancestral locus (Fig 2A). Distinct rearrangements appear to have taken place in Rodentia, represented by squirrel and mouse genomes, leading to a Bace2(+);Mx1(-);Fam3B(-);Mx2(+);Tmprss2(-) configuration (Fig 2A). Intriguingly, squirrel Mx2 is an MxB-derived pseudogene, whereas mouse Mx2 is an MxA-derived intact gene (S2 Fig). Based on this, we propose that mouse Mx2 originated as a result of complete gene conversion by Mx1 (MxA-like) (Fig 2A), possibly preceded by loss of the ancestral MxB-like gene in rodents.