Variable levels of drift in tunicate cardiopharyngeal gene regulatory elements

Abstract

Background: Mutations in gene regulatory networks often lead to genetic divergence without impacting gene expression or developmental patterning. The rules governing this process of developmental systems drift, including the variable impact of selective constraints on different nodes in a gene regulatory network, remain poorly delineated.

Results: Here we examine developmental systems drift within the cardiopharyngeal gene regulatory networks of two tunicate species, Corella inflata and Ciona robusta. Cross-species analysis of regulatory elements suggests that trans-regulatory architecture is largely conserved between these highly divergent species. In contrast, cis-regulatory elements within this network exhibit distinct levels of conservation. In particular, while most of the regulatory elements we analyzed showed extensive rearrangements of functional binding sites, the enhancer for the cardiopharyngeal transcription factor FoxF is remarkably well-conserved. Even minor alterations in spacing between binding sites lead to loss of FoxF enhancer function, suggesting that bound trans-factors form position-dependent complexes.

Conclusions: Our findings reveal heterogeneous levels of divergence across cardiopharyngeal cis-regulatory elements. These distinct levels of divergence presumably reflect constraints that are not clearly associated with gene function or position within the regulatory network. Thus, levels of cis-regulatory divergence or drift appear to be governed by distinct structural constraints that will be difficult to predict based on network architecture.

Keywords: Developmental systems drift; Gene regulatory networks; Heart development; Selective constraints; Tunicates.

Fig. 1

Ciona robusta cardiopharyngeal gene regulatory network and tunicate phylogeny. a–c Regulatory network diagrams for cardiopharyngeal founder lineage cells during three embryonic stages. Schematics on the left indicate stage and cell lineage. Background colors delineate discrete regulatory modules. Solid lines indicate regulatory connections supported by functional enhancer analysis, while dashed lines indicate regulatory connections supported by expression data. Circles represent signal dependent activation and double slanted lines represent signal transduction. a Initial specification of the cardiopharyngeal founder cells (pink) through exclusive up-regulation of Mesp and subsequent expression of Ets1/2. b Signal-dependent regulation of early trunk ventral cell genes by Ets1/2 and an unknown ATTA-binding co-factor. Ets1/2 activation in the TVCs is dependent on FGF9/16/20 signaling transduced by the MapK pathway. c Presumptive modules differentially regulated by FoxF, Hand-like, or GATAa. FoxF is portrayed as the primary regulator of TVC migration, while GATAa regulates a highly conserved heart “kernel” in conjunction with BMP2/4 signaling. F1, H1, H2, and K1 represent hypothetical target genes. Figure is based on Woznica et al. and Cota et al. [41, 42]. d Simplified tunicate phylogeny based on DeBiasse et al. (in prep), that is congruent with Delsuc et al. [43]. Background colors represent sub-clades, Phlebobranchia (yellow), Stolidobranchia (red), or Appendicularia (blue)

Fig. 2

Conserved founder cell lineage behavior and TVC induction in Corella embryos. a–c Representative Corella embryos expressing Cirobu.Mesp −1916:Esconsin-3XGFP in presumptive founder lineage cells. Note labeling of mitotic spindle in 8H embryo (a). Hours post-fertilization indicated at the lower right of each panel. d Representative Corella embryo expressing Cirobu.Mesp −1916:H2B:GFP to track founder lineage cell divisions in later stages. e, f Transgenic Cirobu.Mesp 1916:GFP Corella embryos treated with the Map Kinase inhibitor U0126 at 7.5 HPF, e immediately prior to founder cell division or f ~ 2 h after division at 10-11 HPF. Arrow points to migrated TVCs. g Summary of results for U0126 treatments. Data spans 6 trials, N > 70 for each condition, Student’s T test, p value < 0.0005. Note that the levels of migration defects in the 10-11HPF treatment samples were similar to basal levels seen in untreated, transgenic embryos (data not shown). h, i Representative embryos illustrating TVC expression for the Cirobu.FoxF-3052:GFP and Cirobu.Hand-like-2954/− 445:− 296:lacZ reporters

Fig. 3

Characterization of the C. inflata FoxF TVC enhancer. a mVISTA alignments depict sequence conservation between C. robusta and C. savignyi and between C. robusta and C. inflata for the FoxF gene and 5′ intergenic region (LAGAN alignment, conservation across 100 bp window, conservation > 70% highlighted). There is increased conservation associated with the FoxF coding region (orange) and conserved TVC enhancer (purple). b ClustalW alignment of the 183 bp conserved TVC enhancer with Ets1/2 (red), ATTA (blue), and TGTT (orange)-binding motifs highlighted. Dark-shaded-binding motifs were required for reporter expression and boxed-binding motifs exhibited no functionality. C. robusta FoxF-binding motif knockout data come from Beh et al. and Woznica et al. [41, 58]. c–e Representative embryos showing the activity of Coinfl.FoxF −2622 GFP reporter constructs in C. inflata and C. robusta (arrows indicate expression in TVCs, and scale bar is 50 μm). f Representative C. inflata mid-tailbud stage embryo displaying expression of Coinf.FoxF in TVCs (arrow) and epidermis. g Effect of Ets1/2 and ATTA-binding motif knockouts (Δ) on reporter expression driven by the C. inflata 146 bp minimal TVC enhancer fused to a 255 bp basal promoter (Coinfl.FoxF −547/−401::−255). Names of binding motifs correspond to the names in b. LacZ reporter constructs are diagramed on the left with X indicating a binding motif knockout. The graph depicts %TVC expression in C. inflata (number of trials ≥ 2, total N ≥ 150, and error bars indicate standard deviation). Significance relative to Coinfl.FoxF −547/−401::−255 was determined with a Student’s t test, p < 0.05 indicated by *

Fig. 4

Characterization of the C. inflata Hand-like TVC enhancer. a Minimization of the C. inflata Hand-like (HL) upstream genomic fragment to test two predicted enhancers. LacZ reporter constructs are diagramed on the left. The graph depicts %TVC expression in C. robusta (number of trials ≥ 2, total N ≥ 75, and error bars indicate standard deviation). Significance relative to Coinfl.HL −1737 was determined with a Student t test (p < 0.001 indicated by ***). The second predicted enhancer is both necessary and sufficient for reporter expression in the TVCs. b–e Representative embryos showing the expression of LacZ reporter constructs that contain the second predicted enhancer (Coinfl.HL −1048) or lack the second predicted enhancer (Coinfl.HL −899) in both C. robusta and C. inflata (arrows indicate expression in TVCs, and scale bar is 50 μm). f Effect of Ets1/2 and ATTA-binding motif knockouts (Δ) on the expression of a C. inflata Hand-like::LacZ reporter construct containing a 1048 bp upstream genomic fragment (Coinfl.HL −1048). Names of binding motifs correspond to the names in panel B. LacZ reporter constructs are diagramed on the left with X indicating a binding motif knockout. The graph depicts %TVC expression in C. robusta (number of trials ≥ 2, total N ≥ 25, and error bars indicate standard deviation). Significance relative to Coinfl.HL −1048 was determined with a Student t test (p < 0.01 indicated by ** and p < .001 indicated by ***). g Comparison of Hand-like TVC enhancer structure in C. robusta and C. inflata. Darkly shaded binding motifs were required for reporter expression. Lightly shaded binding motifs exhibited ‘limited” functionality as assessed by mutagenesis of multiple sites in the minimal Cirobu.FoxF enhancer [41] or by a non-significant reduction in reporter expression following mutagenesis (this study). Boxed binding motifs exhibited no functionality. C. robusta Hand-like binding motif knockout data comes from Woznica et al. [41]

Fig. 5

Characterization of the C. inflata Mesp founder lineage enhancer. a Minimization of the C. inflata Mesp 5′ intergenic region to identify the B7.5 founder lineage enhancer. LacZ reporter constructs are diagramed on the left. The graphs depict % founder lineage (TVC + ATM) expression or % founder lineage + primary tail muscle lineage expression in C. robusta and C. inflata (number of trials ≥ 2, total N ≥ 25, and error bars indicate standard deviation). Significance relative to Coinfl.Mesp −651 was determined with a Student t test (p < 0.01 indicated by ** and p < 0.001 indicated by ***). b Representative C. robusta embryo showing founder lineage-specific expression of Coinfl.Mesp −866 (arrows indicate TVCs and ATMs, and scale bar is 50 μm). c Representative C. robusta embryo showing the founder lineage and primary tail muscle lineage expression for Coinfl.Mesp −576. d Representative C. robusta embryo showing the lack of founder lineage expression for Coinfl.Mesp −421. e Representative C. inflata embryo showing the founder lineage-specific expression for Coinfl.Mesp −866. f Effect of TBX6 and LHX3 binding motif knockouts (Δ) on the expression of the C. inflata Mesp founder cell enhancer. Binding motifs designated as shown in g. LacZ reporter constructs are diagramed on the left with an X indicating a binding motif knockout. The graphs depict % founder lineage expression in C. robusta and C. inflata (number of trials ≥ 2, total N ≥ 75, and error bars indicate standard deviation). Significance relative to Coinfl.Mesp −576 or the minimal −576/−421 enhancer was determined with a Student t test (p < 0.05 indicated by * and p < 0.01 indicated by **). g Structure of the C. inflata Mesp founder cell enhancer. Darkly shaded green TBX6 motifs were required for reporter expression, and lightly shaded pink LHX binding motifs exhibited some functionality, as determined by mutagenesis of multiple motifs. There is no conservation of functional binding motifs in the aligned upstream genomic region of C. robusta. h Representative C. robusta embryo showing the founder lineage and primary tail muscle lineage expression for Coinfl.Mesp −576/−421::−138. i–k Representative C. robusta embryos showing lack of reporter expression for i Coinfl.Mesp ΔT1, and j Coinfl.Mesp ΔL1,2,3,4 and k reporter expression in the primary tail muscle lineage, but not the founder lineage for Coinfl.Mesp ΔL4,5,6

Fig. 6

Functional constraint on binding site spacing in the C. robusta FoxF TVC enhancer. a The first Ets1/2-binding site was moved by knocking out the endogenous binding site (GGAT ⟶ GCTT) and introducing a new binding site using site-directed mutagenesis. Reporter constructs contained the 245 bp minimal C. robusta FoxF TVC enhancer fused to the C. robusta Fkh basal promoter (Cirobu.FoxF −1072/−827::pFkh). The sequence of the enhancer region containing this first Ets1/2-binding site is shown on the left with Ets1/2 (red) and ATTA (blue)-binding sites highlighted. The graph depicts %TVC expression in C. robusta (number of trials ≥ 2, total N ≥ 75, and error bars indicate standard deviation). b–d Representative C. robusta embryos showing reporter expression for b Corobu.FoxF −1072/−827::pFkh, c Corobu.FoxF −1072/−827::pFkh ∆E1, or d Corobu.FoxF −1072/−827::pFkh Move 1. Arrows point to normal position of TVCs in the trunk region. Note substantial ectopic expression in the anterior tail muscle lineage (ATM) and in other muscle and mesenchyme lineage cells

Fig. 7

Model for the differential constraint on FoxF vs. other early TVC enhancers. a Before FGF induction, the chromatin around early TVC gene enhancers is condensed preventing aberrant expression. In contrast, chromatin is decondensed at the FoxF TVC enhancer locus, suggesting that a repressor (purple) is required to prevent precocious expression. b FGF/MapK-signaling phosphorylates Ets1/2 in the TVCs, permitting recruitment of a co-factor (green) that serves to lift repression. The cooperative recruitment of this co-factor constrains binding site position and orientation. FoxF (orange) then accumulates in the TVC nuclei, where it acts as a pioneer factor opening the chromatin around other TVC enhancers. c Once early TVC gene enhancers are open, the binding of Ets1/2, ATTA, and FoxF activates transcription in a non-cooperative fashion, as reflected by a lack of constraint on binding site position. FoxF also binds the FoxF TVC enhancer helping to maintain its own expression