Combinatorial regulation of endothelial gene expression by ets and forkhead transcription factors

Abstract

Vascular development begins when mesodermal cells differentiate into endothelial cells, which then form primitive vessels. It has been hypothesized that endothelial-specific gene expression may be regulated combinatorially, but the transcriptional mechanisms governing specificity in vascular gene expression remain incompletely understood. Here, we identify a 44 bp transcriptional enhancer that is sufficient to direct expression specifically and exclusively to the developing vascular endothelium. This enhancer is regulated by a composite cis-acting element, the FOX:ETS motif, which is bound and synergistically activated by Forkhead and Ets transcription factors. We demonstrate that coexpression of the Forkhead protein FoxC2 and the Ets protein Etv2 induces ectopic expression of vascular genes in Xenopus embryos, and that combinatorial knockdown of the orthologous genes in zebrafish embryos disrupts vascular development. Finally, we show that FOX:ETS motifs are present in many known endothelial-specific enhancers and that this motif is an efficient predictor of endothelial enhancers in the human genome.

Figure 1. Identification of a 44-bp Mef2c endothelial-specific enhancer

(A) A schematic representation of the mouse Mef2c locus is shown on the top line with exons depicted as vertical lines. The red boxes denotes the sizes and positions of the F7 and F10 fragments. F10 contains three evolutionarily conserved regions, denoted CR1-3. The lower portion of (A) depicts the deletion constructs of Mef2c F10. CR3 contains a neural crest specific enhancer. CR2 contains an endothelial specific enhancer, which encompasses a 44-bp deeply conserved region that is sufficient for endothelial enhancer activity in vivo. Endothelial and neural crest activity of each of the deletion constructs is denoted at the right as a + or -. The total number of transgenic embryos and the number that directed β-galactosidase expression to either the neural crest or endothelium are denoted at the far right of (A). (B-G) Representative X-gal stained transgenic embryos for each of the Mef2c F10 transgene deletion constructs depicted in (A). (H-M) Expression of the Mef2c F10-44-lacZ construct is specific to endothelial cells from blood island (bl) stage at E7.5 (H) throughout early endothelial development at E8.0 (I) and E8.5 (J, K). Transverse sections through an X-gal stained E9.5 transgenic embryo (L, M) demonstrate that transgene expression is restricted to endothelial cells throughout the vasculature, including the endocardium (end). al, allantois; BAA, branchial arch artery; CV, cardinal vein; DA, dorsal aorta; DRG, dorsal root ganglia; ec, ectoplacental cone; hrt, heart; LV, left ventricle; NC, neural crest; NT, neural tube; RV, right ventricle; SV, sinus venosus; YS, yolk sac.

Figure 2. Identification of a novel FOX:ETS motif simultaneously bound and synergistically activated by FoxC2 and Etv2

(A) Alignment of the mouse and zebrafish Mef2c F10-44 sequences. Red boxes denote core ETS binding sites, and the blue box denotes a non-consensus Forkhead binding element (FOX-NC). The novel, composite FOX:ETS motif is indicated above. Consensus Forkhead and Ets binding sites (Hollenhorst et al., 2007; Carlsson and Mahlapuu, 2002) are denoted, as is the mutant FOX:ETS sequence used in these studies. (B) Radiolabeled oligonucleotide probes encompassing the F10-44 ETS-A site were used in EMSA with recombinant Ets proteins. The Ets1 DNA binding domain (DBD) and Etv2 efficiently bound to the site (lanes 2, 11) and were competed by excess unlabeled self probe (wt, lanes 3, 12) but not by mutant self probe (mu, lanes 4, 13). Erg and Elf-1 displayed little or no detectable binding to ETS-A in this assay. (C) A radiolabeled oligonucleotide probe encompassing the Mef2c F10-44 FOX-NC site was used in EMSA with recombinant Forkhead proteins. FoxA2, FoxF1, and FoxH1 showed weak or no binding to FOX-NC. FoxO1 (lanes 8, 9) showed weak binding to FOX-NC. FoxC1 (lanes 10, 11) and FoxC2 (lanes 12, 13; also lanes 14-17) exhibited robust binding. Addition of excess, unlabeled self-probe, indicated by a + sign, inhibited binding of FoxO1, FoxC1, and FoxC2 to the FOX-NC site (lanes 9, 11, 13). Additionally, inclusion of a mutant version of FOX-NC (lane 17, mu) did not inhibit binding of FoxC2 to FOX-NC at the same concentration that the wild-type self-probe completely abolished binding (lane 16, wt). (D) Chromatin immunoprecipitation from mouse embryo fibroblasts transfected with pCDNA3.1-FoxC2-Flag (C2) or parental pCNA3.1 expression vector (ctrl). Sheared, cross-linked chromatin fragments were immunoprecipitated with anti-FLAG antibody and the region of the endogenous Mef2c locus, surrounding F10-44, was amplified by PCR. The Mef2c F10-44 region was specifically amplified in pDNA3.1-FoxC2-FLAG transfected cells (lane 5), similar to the amplification in control samples that were directly amplified without prior immunoprecipitation (input, lanes 1, 3). No amplification was detected in control transfected (lane 2) or non-specific IgG immunoprecipitated samples (lane 4). (E) FoxC2 and Etv2 synergistically trans-activate the Mef2c F10E enhancer. FoxC2 and Etv2 each weakly activated the reporter (lanes 2, 3) compared to parental expression plasmid control transfections (lane 1). Cotransfection of the reporter with expression plasmids for FoxC2 and Etv2 together resulted in potent synergistic activation (lane 4). Mutation of the FOX:ETS motif (mutFEM) ablated activation by FoxC2 and Etv2 (lanes 5-8). Data are presented as the mean plus SEM for four independent sets of transfections and analyses. (F) FoxC2 and Etv2 simultaneously bind the FOX:ETS motif. A radiolabeled oligonucleotide probe (Mef2c-F10 FOX:ETS) encompassing only the F10E FOX:ETS motif was used in EMSA with recombinant FoxC2 and Etv2. The labeled probe included the FOX:ETS motif plus short adjacent sequences and did not include additional potential ETS binding sites. Increasing amounts of FoxC2 in the absence of Etv2 resulted in the formation of an increasing amount of FoxC2-DNA complex (lanes 2-4). Addition of Etv2 alone resulted in the formation of an Etv2-DNA complex (lane 5). Addition of increasing amounts of FoxC2 in the presence of a constant amount of Etv2 resulted in formation of each individual protein-DNA complex as well as a slower mobility band, suggesting a FoxC2-Etv2-DNA ternary complex (lanes 6-8). Relative levels of FoxC2 and Etv2 protein and binding activity are indicated at the top of the panel. In all samples, the total amount of total protein was held constant by the addition of the appropriate amount of unprogrammed reticulocyte lysate. (G) A 3-bp mutation (CATAACAGGAA to CATAtCtaGAA) of the FOX:ETS motif (mutFEM) or mutation of the ETS-B site in the context of Mef2c F10, which contains both neural crest and endothelial enhancers, results in loss of transgene expression in the endothelium but not the neural crest. The resultant transgenic embryos show expression patterns similar to those in which the entire 44-bp element was deleted from F10 (Mef2c F10Δ44). Representative transgenic embryos from each construct are shown.

Figure 3. Misexpression of FoxC2 and Etv2 in Xenopus embryos induces ectopic endothelial gene expression

Xenopus embryos were injected with mRNAs encoding FoxC2 and Etv2 or EGFP control mRNA at the 4-cell stage and then collected at stage 36. After collection, embryos were either assayed by in situ hybridization using flk1 probe, followed by sectioning (A, B) or RNA was extracted for qPCR analysis of flk1 (C) or Pecam (D) transcripts. (A, B) flk1 expression was observed in the cardinal veins (CVs) in control (A) and FoxC2 + Etv2-injected (B) embryos. In addition, ectopic expression of flk1 was readily observed in the endoderm of the caudal region of FoxC2 + Etv2-injected embryos (B) but not in EGFP control injected embryos (A). (C, D) Quantitative, real-time PCR shows that neither FoxC2 nor Etv2 significantly activated flk1 or Pecam expression on their own, but the combination of the two factors strongly induced expression of both endothelial-specific markers. Data are shown as the mean relative expression of flk1 or Pecam transcripts plus the SEM for three independent sets of injections and analyses.

Figure 4. Cooperative regulation of vascular development in zebrafish by FoxC and Ets proteins

(A-D) The mouse Mef2c F10E enhancer directs expression of the GFP reporter gene in the vascular endothelium of transgenic zebrafish (A, B) in a nearly identical pattern to the endothelial-specific Tg(flk1:GFP)^s843 reporter (C, D). (E-H) In situ hybridization shows that the zebrafish foxc genes foxc1a (E, F) and foxc1b (G, H) are expressed in the developing vasculature at 24 hpf. (I-L) Knockdown of foxc1a and foxc1b by morpholino injection alone (J, K) and in combination (L) resulted in loss of vascular structure, as detected by reduced expression of Tg(flk1:GFP)^s843 (green) and the pooling of blood, as indicated by Tg(gata1:DsRed)^sd2 expression (red). The combined foxc1a/foxc1b knockdown (L) resulted in a more severe perturbation of vascular development than either single knockdown. Note the normal expression of Tg(flk1:GFP)^s843 and Tg(gata1:DsRed)^sd2 in the control morpholino injected embryo (I). (M-P) Injection of sub-phenotypic doses of foxc1a (N) and etsrp (O) morpholinos resulted in normal vascular development and normal expression of Tg(flk1:GFP)^s843 and Tg(gata1:DsRed)^sd2 in patterns identical to control injected embryos (M). Co-injection of the lower doses of foxc1a and etsrp morpholinos resulted in a nearly complete loss of vascular development (P), indicating cooperative regulation of vascular development by the two transcription factors. Asterisks mark the pooling of blood. Arrowheads point to the developing axial vessels, and arrows indicate the developing intersomitic vessels.

Figure 5. The FOX:ETS motif is present in multiple endothelial enhancers

(A) Sequence and genomic location of FOX:ETS motifs in MEF2C and five other known endothelial-specific regulatory elements. The ETS sites are highlighted in red, and the FOX-NC sites are highlighted in blue. Chromosome locations refer to the May 2004 assembly of the human genome. (B) ChIP from primary mouse embryo fibroblasts transfected with pCDNA3.1-FoxC2-FLAG (C2) or parental pCDNA3.1 expression vector (ctrl) shows that FoxC2 binds to the FOX:ETS motif in each of the previously described endothelial enhancers. In each case, the enhancer regions were specifically amplified in pDNA3.1-FoxC2-FLAG transfected cells (lane 5), similar to the amplification in control samples that were directly amplified without prior immunoprecipitation (input, lanes 1, 3). No amplification was detected in control transfected (lane 2) and non-specific IgG immunoprecipitated samples (lane 4). Note that these reactions were performed in conjunction with the ChIP for the myogenin promoter region, shown in Fig. 2D, which also serves as a non-specific control for these endogenous genes. (C) EMSA demonstrates that FoxC2 (lanes 2, 8, 14, 20, 26) and Ets1 DBD (lanes 5, 11, 17, 23, 29) bind directly to the FOX:ETS motifs present in FLK1, TEK (Tie2), TAL1, NOTCH4, and CDH5 (VE-CADHERIN). In each case, an excess of unlabeled FOX:ETS motif self-probe (wt) efficiently competed for binding of FoxC2 and Ets1 DBD. Small mutations within the FOX-NC site (mu) disrupted competition by unlabeled probes even when added in 50× excess (lanes 4, 10, 16, 22, 28).

Figure 6. FoxC2 and Etv2 synergistically activate multiple endothelial enhancers

(A-E) FoxC2 and Etv2 synergistically trans-activate the Flk1 (A), Tie2 (B), Tal1 (C), NOTCH4 (D) and VE-CADHERIN/CDH5 (E) enhancers. Data are presented as the mean plus SEM for three to six independent sets of transfections and analyses. Note that in (E), a 4-bp mutation in the FOX:ETS motif (mutFEM) completely abolished activation of the VE-CADHERIN promoter/enhancer by FoxC2 and Etv2. (F, G) Mutation of the FOX:ETS motif within the 3.5-kb VE-CADHERIN promoter/enhancer completely disrupts VE-CADHERIN-lacZ transgene expression at E9.5 (G) when compared to the strong, vascular-specific expression of the wild-type transgene (F). All five embryos transgenic for the wild-type 3.5-kb VE-CADHERIN enhancer expressed lacZ robustly in the endothelium, while none of the three embryos transgenic for the mutated enhancer showed detectable expression.

Figure 7. Prediction of endothelial-specific enhancers based on the presence of a FOX:ETS motif

(A) Sequence logo representing the position weight matrix of the consensus FOX:ETS motif used in a genome-wide scan. (B) The FOX:ETS motif is overrepresented in endothelial genes when compared to housekeeping and skeletal-muscle expressed genes. (C) Identification of five novel endothelial-specific enhancers from the whole-genome screen based on the presence of a FOX:ETS motif. The upper row of photos shows representative whole-mount X-gal stained transient transgenic embryos at E9.5 from the ECE1, FLT4, PDGFRβ, NRP1, and FOXP1 genes. Each directed strong lacZ expression specifically to the endothelium, which can be clearly seen in transverse sections taken from each of the transient transgenic analyses at E9.5 (lower row of photos). CV, cardinal vein; DA, dorsal aorta; hrt, heart; LV, left ventricle; NT, neural tube; RV, right ventricle; SV, sinus venosus.