A cis-regulatory signature for chordate anterior neuroectodermal genes

Abstract

One of the striking findings of comparative developmental genetics was that expression patterns of core transcription factors are extraordinarily conserved in bilaterians. However, it remains unclear whether cis-regulatory elements of their target genes also exhibit common signatures associated with conserved embryonic fields. To address this question, we focused on genes that are active in the anterior neuroectoderm and non-neural ectoderm of the ascidian Ciona intestinalis. Following the dissection of a prototypic anterior placodal enhancer, we searched all genomic conserved non-coding elements for duplicated motifs around genes showing anterior neuroectodermal expression. Strikingly, we identified an over-represented pentamer motif corresponding to the binding site of the homeodomain protein OTX, which plays a pivotal role in the anterior development of all bilaterian species. Using an in vivo reporter gene assay, we observed that 10 of 23 candidate cis-regulatory elements containing duplicated OTX motifs are active in the anterior neuroectoderm, thus showing that this cis-regulatory signature is predictive of neuroectodermal enhancers. These results show that a common cis-regulatory signature corresponding to K50-Paired homeodomain transcription factors is found in non-coding sequences flanking anterior neuroectodermal genes in chordate embryos. Thus, field-specific selector genes impose architectural constraints in the form of combinations of short tags on their target enhancers. This could account for the strong evolutionary conservation of the regulatory elements controlling field-specific selector genes responsible for body plan formation.

Figure 1. Mutational analysis of D1(bcde) reveals at least three binding sites necessary for its activity.

(A) Alignment of D1(bcde) (195pb) between Ci/Cs. Conserved blocks in D1(bcde) with putative transcription factor binding sites (TFBS). Four classes of putative TFBS within ten blocks of conserved sequence: O1 and O2 display the characteristic GATTA/TAATC core sequence that is known to bind K50 Paired-related homeodomain proteins, including otx/otd, pitx and goosecoid. Putative Fox protein binding sites were identified in the blocks called T1 and T2 (T/A-rich). Blocks M1 and M2 show similarity to known binding sites of the TALE-class Meis/TGIF homeodomain proteins. Finally, putative Smad1/5/8 binding sites were observed in blocks labelled G1, G2, G3 (G/C-rich). Conserved blocks of nucleotides are highlighted in grey. Putative binding sites are represented by colored boxes. Mutations designed to disrupt the binding specificity of each sites are indicated in grey above the sequence (m0 to m9). (B) Schematic representation of D1(bcde) and series of mutations introduced into the pD1bcde:CES2 construct. On the right is shown for each construct the rate of mid-tailbud embryos expressing the lacZ reporter in the anterior neural boundary (anb), the anterior epidermis (ae), the ventro-anterior sensory vesicle (vasv) and the mesenchyme (mes) (wt n = 444; m0 n = 198; m1 n = 56; m2 n = 348; m3 n = 40; m4 n = 58; m5/6 n = 182; m7/8/9 n = 29). Most mutations led to a general reduction of the expression level in all of aforementioned domains, while mesenchyme expression remained high. Only construct m3 (block M1 and M2) retained the activity in the anterior neural boundary.

Figure 2. Artificial enhancer constructs reveal a tandem-like structure.

(A) D1(abcde) Ci-Pitx enhancer. Stars show conserved positions with Ciona savignyi. The element has been divided into five parts (a, b, c, d, e). (ab) fragment is in dark blue, (d) in light blue with blue outline, (e) in light blue. Conserved nucleotides stretches contain putative transcription factor binding sites (TF-BS). O1 and O2 sites (white box) correspond to the BS for K-50 Paired homeodomain proteins and P (yellow), T1, T2 (green) and G1, G2, G3 (red) resemble Pax, Forkhead and Smad protein consensus BS respectively. (B) Side view of an early-tailbud embryo electroporated with pD1abcde:CES2:lacZ: expression in the anterior neural boundary (ANB). In some cases, ectopic expression occurs in the mesenchyme and the tail muscles (not shown). (C) Expression of artificial enhancers in the anterior neuroectoderm of mid-tailbud embryos. LacZ expression was observed after two hours of staining. The D1(abde) construct drives lacZ expression in the anterior neuroectoderm (ANB), ventro-anterior sensory vesicle (vasv) and epineural epidermis (ene) in 77.9% of developed embryos (n = 57). Deletions of (e) or (de), but not (c), abolish LacZ expression (D1(abd), D1(ab)). (D) Two (D1(ab)(ab)) or five (D1(ab)×5) copies of the 54bp D1(ab) drive expression in most of the embryos (88% (n = 167) and 77% (n = 72), respectively). Only 17% of the embryos express lacZ following the deletion of the second P site (D1(ab)(ab-P_del), n = 90). Mutations of O, T, and G sites in the second copy of (ab) strongly decrease lacZ expression. (D1(ab)(ab-O^mut): 0% (n = 137), D1(ab)(ab-T^mut): 7% (n = 84), D1(ab)(ab-G^mut): 0% (n = 118)). (E) Mid-tailbud embryo electroporated with pD1(ab)(ab):CES2:lacZ. Expression is visible in the ANB, the vasv, the ene and less frequently in the endodermal strand. (F) Introduction of spacer regions of 25/50/75/150 bp between the two D1(ab) fragments strongly decreased the activity of the tandem constructs. From 88% (D1(ab)(ab)) to 31.6% (n = 76), 16.9% (n = 154), 2.5% (n = 79) and 1.9% (n = 106), respectively. Scores were obtained after one week of LacZ revelation.

Figure 3. OTX fusions influence the activity of the Ci-pitx cis-regulatory element.

Co-electroporation of Pitx full length distal region (pDistalPitx:lacZ, 5.3kb, containing D1), respectively with pSix3:Venus (control), with pSix3>OTX_HD::enR (dominant negative OTX) and pSix3>OTX_HD::VP16 (hyper-active OTX). (A) Side view of an embryo co-electroporated with pDistalPitx:lacZ and pSix3:Venus. Three positive cells can be detected in the ANB. (B) Co-electroporation of pDistalPitx:lacZ and pSix3>OTX_HD::VP16. In addition to the expression in the ANB, ectopic expression is detected in the ASV cells (bracket) where OTX:VP16 is produced under the control of pSix3. (C) Numbers of lacZ expressing cells decrease with the OTX_HD::enR protein (2.78 to 0.74 cells) and increase with the OTX_HD::VP16 protein (2.78 to 4.49 cells) The distributions differ significantly from the control in both groups according to two Wilcoxon–Mann–Whitney two-sample rank-sum tests: Control/OTXenR (U_OTXenR = 3891, n(emb)_OTXenR = 139, n(emb)_ctrl = 131, P = 2.536e-07 two-tailed) and control/OTXVP16 (U_OTXVP16 = 15582.5, n(emb)_OTXVP16 = 98, n_ctrl = 131, P<2.2e-16 two-tailed). (D) Distributions of cell numbers in the ANB and ASV after co-electroporation of DistalPitx:lacZ and OTX fusions (yellow: control, red: enR fusion, blue: VP16 fusion; X-axis: cell numbers, Y-axis: proportions of embryos). Each experiment has been performed twice.

Figure 4. Motif-tissue scores for the motif 2×GATTA/125bp against genes expressed in various tissues.

These territories are also visualized on schematic representation of an ascidian tailbud embryo. The number of genes is indicated for each category. To illustrate that changes in gene annotation are very unlikely to affect the overall ranking, we shuffled 10% of the gene-tissue assignments, repeated the procedure 100 times and plotted 95%-confidence intervals with error bars.

Figure 5. Enhancers with duplicated GATTA are active in the anterior region of the ascidian embryo.

(A) Schematic representation of the main regions of gene expression in a mid-tailbud Ciona intestinalis embryo. Cell cortices are stained with Alexa-phalloidin (Christiaen et al. 2007). (B–D): expression domains of three enhancers, respectively from Ci-Six3/6, Ci-Eya1 and Ci-Tbx3 after electroporation and X-Gal staining at mid-tailbud stage. Lower panels show a schematic representation of the LacZ expression driven by the enhancer (left) and endogenous gene expression as assayed by in situ hybridization (ISH) (right). Enhancers can be subdivided into different classes following their expression domains: very restricted expression only in the ANB while the gene expression domain is slightly larger (Six3, (B)); broad anterior expression recapitulating more or less the endogenous expression pattern (Eya, (C)); only the most anterior expression domains are driven by the enhancer (Tbx3 (D)). anb: anterior neural boundary, asv: anterior sensory vesicle, psv: posterior sensory vesicle, vg: visceral ganglion, nt: neural tube, ae: anterior epidermis (or epineural epidermis), p: palps (precursors), m: mesenchyme, n: notochord. Lateral views, anterior to the left.