A murine specific expansion of the Rhox cluster involved in embryonic stem cell biology is under natural selection

Abstract

Background: The rodent specific reproductive homeobox (Rhox) gene cluster on the X chromosome has been reported to contain twelve homeobox-containing genes, Rhox1-12.

Results: We have identified a 40 kb genomic region within the Rhox cluster that is duplicated eight times in tandem resulting in the presence of eight paralogues of Rhox2 and Rhox3 and seven paralogues of Rhox4. Transcripts have been identified for the majority of these paralogues and all but three are predicted to produce full-length proteins with functional potential. We predict that there are a total of thirty-two Rhox genes at this genomic location, making it the most gene-rich homoeobox cluster identified in any species. From the 95% sequence similarity between the eight duplicated genomic regions and the synonymous substitution rate of the Rhox2, 3 and 4 paralogues we predict that the duplications occurred after divergence of mouse and rat and represent the youngest homoeobox cluster identified to date. Molecular evolutionary analysis reveals that this cluster is an actively evolving region with Rhox2 and 4 paralogues under diversifying selection and Rhox3 evolving neutrally. The biological importance of this duplication is emphasised by the identification of an important role for Rhox2 and Rhox4 in regulating the initial stages of embryonic stem (ES) cell differentiation.

Conclusion: The gene rich Rhox cluster provides the mouse with significant biological novelty that we predict could provide a substrate for speciation. Moreover, this unique cluster may explain species differences in ES cell derivation and maintenance between mouse, rat and human.

Figure 1

A. Dotplot analysis comparing repeat A to repeats B to H using the NCBI m33 mouse assembly (freeze May 27, 2004, strain C57BL/6J). Black lines represent regions of identity interspersed with LINEs (horizontal pink lines). The orientation and approximate size of each repeat unit is presented as black arrows. The relative position and orientation of Rhox1 (green lines, black arrowhead) and Rhox2, 3 and 4 (blue lines, white arrowhead) and Rhox5 (which is upstream of this region, black arrowhead) are shown. The X chromosome co-ordinates are provided below the dotplot. B. BLAST analysis comparing genomic sequence of each repeat to all others. LINE repeats and rearrangements were not added. Top row represents a comparison of repeat A to B-H, second row repeat B to C-H and so on. C. Southern Blot analysis. CGR8 ES cell genomic DNA digested with Asp718 (lane 1) and HindIII (lane2) and hybridised to Rhox4 exon 1 and intron 1 probe. Multiple bands can be seen with both digests (arrowed) of approximately 20, 13, 10 and 5 kb (HindIII) and 23, 18, 13, 10, and 8 kb (Asp718). Mus domesticus (lane 3) and Mus spretus (lane 4) genomic DNA digested with HindIII and hybridised to a Rhox4 exon 2 probe. Fragment sizes common to the Mus musculus (ES cells) and Mus domesticus are arrowed. Additional bands are predicted to be due to the differences between the probes used. Positions of size standards are provided in kilobases.

Figure 2

A & C. Unique nucleotide profile and relative expression of the predicted Rhox2A-H (A) or Rhox4A-H (C) transcripts. The position of the nucleotides is relative to the published start sites. Column RT and EST refers to the number of each transcript detected by RT-PCR or in the EST database respectively that matched this profile. B & D. ClustalW alignment of predicted amino acid sequence from Rhox2A-H (B) or Rhox4A-H (D) Residues different from the consensus are highlighted in blue (conservative amino acid change) or red (non-conservative amino acid change). The homeobox domain is underlined with hydrophobic residues (+) crucial for homeodomain packing and residues involved in DNA binding (*) annotated.

Figure 3

A, B & C. Phylogeny of the Rhox2 (A), Rhox3 (B) and Rhox4 (C) paralogues. The numbers shown along each branch are the maximum likelihood estimates of the ratio of nonsynonymous and synonymous substitutions for the entire gene along that branch. Data for a "free-ratio" model is shown which allows for a different dN/dS ratio (parameter ω) for each branch in the tree. Branches are drawn in proportion to estimates of their lengths.

Figure 4

A & B. Posterior Probability of site classes along the Rhox2 (A) and Rhox4 (B) coding aligments. Model M2 is used. Only probabilities of w2 are shown. Sites with a posterior probability higher than 0.95 are considered as having dN/dS ratios significantly greater than 1. C & D. Variation in ω across the Rhox2 (C) and Rhox4 (D) genes was modelled using Omegamap 17. We averaged over the number and position of the blocks using reversible-jump MCMC to obtain the posterior distribution of the omega. The dotted lines are the top and bottom 95% of the estimates of ω from 500,000 iterations. The black line is ω per codon smoothed over blocks of undefined size. Posterior probabilities of >0.95 (blue line) likely indicate significant ω or positive selection. The location of the largest ω values coincide with similar locations as shown in Figure 4A & B obtained from PAML analyses (see text).

Figure 5

A. Total number of puromycin-resistant colonies generated after supertransfection of episomal vectors expressing Rhox2, Rhox4, anti-sense Rhox2, anti-sense Rhox4 or a control empty vector. Mann Whitney test revealed significant differences between control and Rhox2 sense (P < 0.03) and control and Rhox4 sense (P < 0.03). B. Alkaline phosphatase stained ES cell cultures after supertransfection with either an empty episomal vector (vector) or an Egfp-, Rhox4-, antisense Rhox4, Rhox2- or antisense Rhox2- expressing episomal vector. Transfected cells were selected in puromycin for 6 days and subsequently replating in the presence (+) and absence (-) of LIF. C. Western blot analysis of COS7 cells transiently co-transfected with vectors expressing sense (S) Rhox2 (lane 3 and 5) or Rhox4 (lanes 2, 4 and 6) and vectors expressing anti-sense (AS) Rhox2 (lanes 1 and 2) or anti-sense Rhox4 (lanes 3 and 4) showing efficient knockdown of RHOX4 protein product in the presence of anti-sense Rhox4 (lane 4) but not anti-sense Rhox2 (lane2). Blots were stripped and re-probed with GAPDH to ensure equivalent loading.