Ancient Pbx-Hox signatures define hundreds of vertebrate developmental enhancers

Abstract

Background: Gene regulation through cis-regulatory elements plays a crucial role in development and disease. A major aim of the post-genomic era is to be able to read the function of cis-regulatory elements through scrutiny of their DNA sequence. Whilst comparative genomics approaches have identified thousands of putative regulatory elements, our knowledge of their mechanism of action is poor and very little progress has been made in systematically de-coding them.

Results: Here, we identify ancient functional signatures within vertebrate conserved non-coding elements (CNEs) through a combination of phylogenetic footprinting and functional assay, using genomic sequence from the sea lamprey as a reference. We uncover a striking enrichment within vertebrate CNEs for conserved binding-site motifs of the Pbx-Hox hetero-dimer. We further show that these predict reporter gene expression in a segment specific manner in the hindbrain and pharyngeal arches during zebrafish development.

Conclusions: These findings evoke an evolutionary scenario in which many CNEs evolved early in the vertebrate lineage to co-ordinate Hox-dependent gene-regulatory interactions that pattern the vertebrate head. In a broader context, our evolutionary analyses reveal that CNEs are composed of tightly linked transcription-factor binding-sites (TFBSs), which can be systematically identified through phylogenetic footprinting approaches. By placing a large number of ancient vertebrate CNEs into a developmental context, our findings promise to have a significant impact on efforts toward de-coding gene-regulatory elements that underlie vertebrate development, and will facilitate building general models of regulatory element evolution.

Figure 1

Meis2 CNEs from zebrafish and lamprey drive equivalent expression patterns in zebrafish and lamprey embryos. A, multiple alignment of orthologous genomic regions containing the gene c15orf41 (blue peak), downstream of meis2, revealing CNEs (red peaks). Human, zebrafish and lamprey sequences are aligned with the fugu sequence as a baseline. Zebrafish CNE 329X is translocated in the current zebrafish genome assembly so does not appear in this alignment. B-M, orthologous elements from lamprey (B-G) and zebrafish (H-M) drive similar GFP expression patterns in the nervous system of zebrafish embryos at 54hpf: element 3285 in the cranial ganglia (arrows) and primary neurons of the hindbrain and spinal cord (arrowhead) (B, H); 3288 in neurons of the hindbrain posterior to rhombomere (r) 4 (C, I), as determined by comparison with r3r5 RFP expression (D, J); 3299 in the anterior hindbrain - r2-4 for the lamprey homolog (E, F) and r3-4 plus the corresponding neural crest for the zebrafish homolog (K, L); 329X in the hindbrain and neurons of the midbrain (G, M). N-O , embryonic day 14-15 lamprey embryos transgenic for lamprey elements 3285 (N) and 3299 (O) show GFP expression in the cranial ganglia (arrowheads) and anterior hindbrain respectively, consistent with their expression in zebrafish (3285: B, 3299: E). P-Q, dorsal views of the head of lamprey (P) and zebrafish (Q) embryos transgenic for lamprey element 3299. cg: cranial ganglia; hb: hindbrain; mb: midbrain; nc: neural crest; sc: spinal cord.

Figure 2

CNE 3299 harbours essential conserved Pbx-Hox and Meis binding-sites. A, multiple sequence alignment of a region of CNE 3299 from human, zebrafish and lamprey highlighting conserved Pbx-Hox (blue box) and Meis (green box) binding-site motifs. The specific sites mutated in elements sub1 and sub2 are indicated below the alignment. B-C, compared to wild-type 3299 expression (B), mutating the first Pbx-Hox and Meis motif cluster (sub1) results in the loss of reporter expression in the neural crest (arrow) and broader expression in the hindbrain (arrowhead) (C). nc: neural crest.

Figure 3

Pbx-hox motifs in CNEs strongly resemble verified PBX-HOX binding-sites. Position frequency logos generated from Gnathostome alignments (based on 712 conserved human TGATNNAT motifs in 4529 CNE alignments), Human CNEs (generated from the CONDOR CNE set using Cis-finder [41]) and from previous studies [36] (Literature). The relative base frequencies at positions 5 and 6, and 9 and 10, in CNEs, are in good agreement with known functional Pbx-Hox binding sites, supporting a strong KR consensus.

Figure 4

Pbx-Hox motifs correlate with segment-specific hindbrain and pharyngeal arch reporter expression. A-R, zebrafish elements from the lamprey (A-J, M, O, Q) and jawed vertebrate (K, L, N, P, R) CNE sets drive GFP expression in the hindbrain and pharyngeal arches. Elements: Evi1_40224 (A, B), Tshz3_43509 (C, D), NR2F2_27254 (E, F), Pax2_217 (G, dorsal view: H), ZNF503_32799 (I, J), Nkx6-1_4281 (K, L), Tshz3_24804 (M), Pax9_2099 (N), TshZ3_24805-6 (O), FoxP1_886 (P), Tshz3_24807 (Q), BCL11A_2554 (R). Expression in the hindbrain is often restricted to certain rhombomeres, as shown by comparison with r3r5 RFP expression (B, D, F, H, J, L). Tshz3_24807 drives expression in the trunk musculature (Q). Elements show temporal variation in reporter expression, expressing most strongly at 24-30hpf (C, D), 48-54hpf (A, B, E, F, I, J, Q) or 72-78hpf (G, H, K, L, M, N, O, P, R). hb: hindbrain; pa: pharyngeal arches; m: muscle.