August 1, 2015;
Cdx ParaHox genes acquired distinct developmental roles after gene duplication in vertebrate evolution.
The functional consequences of whole genome duplications in vertebrate evolution are not fully understood. It remains unclear, for instance, why paralogues were retained in some gene families but extensively lost in others. Cdx homeobox genes encode conserved transcription factors controlling posterior
development across diverse bilaterians. These genes are part of the ParaHox gene cluster. Multiple Cdx copies were retained after genome duplication, raising questions about how functional divergence, overlap, and redundancy respectively contributed to their retention and evolutionary fate. We examined the degree of regulatory and functional overlap between the three vertebrate Cdx genes using single and triple morpholino knock-down in Xenopus tropicalis followed by RNA-seq. We found that one paralogue, Cdx4
, has a much stronger effect on gene expression than the others, including a strong regulatory effect on FGF and Wnt genes. Functional annotation revealed distinct and overlapping roles and subtly different temporal windows of action for each gene. The data also reveal a colinear-like effect of Cdx genes on Hox genes, with repression of Hox paralogy groups 1 and 2, and activation increasing from Hox group 5 to 11. We also highlight cases in which duplicated genes regulate distinct paralogous targets revealing pathway elaboration after whole genome duplication. Despite shared core pathways, Cdx paralogues have acquired distinct regulatory roles during development. This implies that the degree of functional overlap between paralogues is relatively low and that gene expression pattern alone should be used with caution when investigating the functional evolution of duplicated genes. We therefore suggest that developmental programmes were extensively rewired after whole genome duplication in the early evolution of vertebrates.
[+] show captions
Transcriptomic assessment of Cdx paralogue function during Xenopus tropicalis development. a Duplication of the ancestral ParaHox cluster and subsequent gene loss in vertebrates resulting in three Cdx genes. b Experimental procedure: injection of eggs with morpholino oligonucleotides (MO) targeting specific Cdx copy, collection of stage 14 (early neurula) embryos for RNA extraction, and transcriptomic characterisation by RNA-seq. c Gene set overlap of the three Cdx MO targets inferred by differential expression analysis of RNA-seq data (see Methods). Venn diagram categories are specified for all genes (black), upregulated genes (red), and down-regulated genes (blue). d Gene set overlap for three Cdx MOs and co-injection of all three MOs (triple). e Gene expression and fold-change intensities across differential expression for all three Cdx MOs. Above scale fold-change values are noted as triangles. This representation reveals a strong quantitative extent of Cdx4 MO effect compared with Cdx1 and Cdx2 MOs
Cdx expression, cross-regulation, and pair-wise comparison. a Normalized expression of three Cdx paralogues at stage 14 derived from the control uninjected embryos. b Fold-change effect of alternative Cdx MOs on Cdx expression. Error bars indicate standard error; significance level noted as (**) and (*) for Benjamini-Hochberg-adjusted P <0.005 and <0.05, respectively. c Temporal expression profile of Cdx paralogues recovered from data of . d–g Pairwise comparison of gene-specific fold-change triggered by distinct Cdx MOs. Genes with expression affected by both Cdx MOs are plotted as red dots, other genes plotted as grey dots
Fig 3. K-mean clustering of multiple Cdx MOs effects. a Heatmap representation of gene expression fold-changes triggered by the three Cdx MOs and the co-injection triple MOs. Genes are arranged according to k-mean clustering and 12 clusters (left) are delineated to capture the diversity of responses to treatments. b Detail of gene expression response to Cdx MOs in the 12 selected clusters. The average response is plotted as a bold line while response of each member gene of the cluster is plotted in a thin grey line
Fig. 4. Gene set enrichment analysis of Cdx MO effects. Functional annotation derived from Panther pathways and the GO biological process version of Panther database were employed for term enrichment analysis using distinct tests accounting for direction of expression change: distinct directional (DD), mixed directional (MD), non-directional (ND), as well as UP or DOWN regulation. The scheme yielding the best enrichment score was retained as the one providing the best description of the enrichment for the term (bubble fill colour). Displayed terms were retained as showing an enrichment >5 in at least one condition
Distinct Cdx effect on known targets and 2R paralogues. Fold-change in expression induced by different Cdx MOs, or by triple Cdx MO injection, is indicated using a blue (down) to red (up) colour scale for each gene. a, b Schematized Hox gene clusters in X. tropicalis showing effects of Cdx gene knockdown on expression levels. Effects of Cdx4 (a) and triple injection (b) shown. Genes in grey are below expression cut-off (FPKM<2). c Fold-change in Hox gene expression caused by triple MO injection plotted against paralogy group assignment; each data point represents one Hox gene. Only genes with statistically significant change in expression are included. Colours denote anterior (red), middle (green), and posterior (blue) paralogy group assignments, assigning group 3 to anterior. d, e Heatmap representation of MO effect on Cdx target gene pathways: Fgf genes, Wnt genes, Retinoic acid pathway. Only genes expressed at stage 14 (FPKM>2) are included. g Effect of Cdx genes on pairs or triplets of genes originated through 2R vertebrate genome duplication (see Methods) which show differential regulation by distinct Cdx paralogues. Significance of differential expression is denoted by one (P <0.05) or two (P <0.005) asterisks (a, b, d–g)