The conserved +15.2prox1a enhancer drives expression in a subset of the facial lymphatics. (A) Conservation analysis of the 380 kbp region surrounding the zebrafish prox1a locus compared with seven vertebrate species. Blue peaks, exons; red peaks, conserved non-coding DNA; black arrowheads, conserved peaks; blue arrowheads, tested peaks; green arrowheads, identified −2.1prox1a and +15.2prox1a lymphatic enhancers. In the 5′ the first 100 kbp of the alignment contain no conservation peak and has been omitted from the graph. (B) prox1a locus showing the position of the identified conserved lymphatic enhancers. Green boxes, −2.1prox1a and +15.2prox1a enhancers; black boxes, exons. (C) Confocal projections of facial lymphatics labelled with Tg(+15.2prox1a:EGFP; XCA:DsRed2)^uu7kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 60 hpf, 5 dpf and 7 dpf. Arrowheads show expression in the facial LECs (60 hpf) and FCLV (5 and 7 dpf). Asterisks show expression in facial lymphatic endothelium. (D) Predicted endothelial TF binding sites in +15.2prox1a. Blue, binding sites identified in zebrafish (P<1e-04); red, conserved binding sites within vertebrates. (E) Quantification of +15.2prox1a activity in mafba/b mutants. Left: confocal projections of facial lymphatics labelled with Tg(+15.2prox1a:EGFP; XCA:DsRed2)^uu7kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf showing enhancer activity (arrowhead) or lack thereof (asterisk) in the FCLV. Right: quantification of signal intensity in the FCLV normalised to the ganglia in 5 dpf embryos. Sibling (n=16) versus mafbb^−/− (n=13), mafba^−/− (n=12) and mafba^−/−;mafbb^−/− (n=7). Mean±s.d. Sibling versus mafbb^−/−, not significant (ns) (P>0.999); sibling versus mafba^−/−, P=0.017; sibling versus mafba^−/−;mafbb^−/−, P<0.001; mafbb^−/− versus mafba^−/−, ns (P=0.146); mafbb^−/− versus mafba^−/−;mafbb^−/−, P=0.008; mafba^−/− versus mafba^−/−;mafbb^−/−, ns (P>0.999) (Kruskal–Wallis test with Dunn's multiple comparison test). Four technical replicates, biological replicates correspond to the number of data points per condition. Scale bars: 50 μm.

The conserved −2.1prox1a enhancer drives expression in the lymphatic valve. (A) Confocal projections of facial lymphatics labelled with Tg(−2.1prox1a:EGFP; XCA:DsRed2)^uu3kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 3, 5 and 7 dpf. Arrowheads show expression in the developing lymphatic valve. Asterisks show expression in the facial lymphatic endothelium. (B) Confocal projections of the facial lymphatics labelled with Tg(−2.1prox1a:basEGFP;ACry:GFP)^uu10kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf. Left: expression in the lymphatic valve (arrowhead) and facial lymphatic endothelium (asterisk). Right: magnification of the boxed area in the left panel. (C) Confocal projection of the whole embryo at 5 dpf labelled with Tg(−2.1prox1a:basEGFP;ACry:GFP)^uu10kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta). Arrowhead shows expression in the lymphatic valve. Asterisks show additional expression in the skin. (D) Predicted endothelial TF binding sites in −2.1prox1a. Blue, binding sites identified in zebrafish (P<1e-04); red, conserved binding sites within vertebrates; yellow, binding sites conserved in vertebrates but absent in zebrafish. (E)  Representative images and quantification of Nfatc1 binding-dependent −2.1prox1a activity. Left: confocal projections of lymphatic valve labelled with WT, Nfatc1 binding site-mutated or scrambled −2.1prox1a:basEGFP;ACry:GFP constructs. Arrowheads show signal in the valve. Asterisk shows missing signal in the valve. Right: quantification of signal intensity in the valve cells expressing GFP in 5 dpf injected embryos. Mean±s.d. WT (n=12) versus Nfatc1 binding site-mutated (n=12) injected embryos at 5 dpf; P=0.002 (two-tailed Mann–Whitney test). Four technical replicates, biological replicates correspond to the number of data points per condition. Scale bars: 50μm (A,B,E); 500 μm (C).

snATAC-seq identifies four lymphatic prox1a enhancers. (A) Chromatin state surrounding the prox1a locus in lymphatic endothelial cells (LECs), venous endothelial cells (VECs) and arterial endothelial cells (AECs) at 4 dpf, showing the region between 32,815,787−32,975,788 base pairs of chromosome (chr) 17. Orange, tested enhancers; purple, identified accessible chromatin sequences in LECs. (B) Confocal projections of the facial and trunk lymphatics labelled with Tg(−87prox1a:EGFP; XCA:DsRed2)^uom122 (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf. Arrowheads show expression in the face and trunk lymphatics. (C) Confocal projections of the facial and trunk lymphatics labelled with Tg(−71prox1a:EGFP; XCA:DsRed2)^uom121 (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf. Arrowheads show expression in the face and trunk lymphatics. Asterisks show expression in PCV. (D) Confocal projections of the facial and trunk lymphatics labelled with Tg(−14prox1a:EGFP; XCA:DsRed2)^uom120 (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf. Arrowheads show expression in the facial lymphatics. (E) Predicted endothelial TF binding sites in −87prox1a. (P<1e-04), −71prox1a (P<1e-04) and −14prox1a (P<1e-04). Blue, binding sites identified in zebrafish; red, conserved binding sites within Actinopterygii. Scale bars: 50 μm.

−2.1prox1a_1 is the core element driving valve expression. (A) Confocal projections of the facial lymphatics labelled with Tg(−2.1prox1a_1:EGFP; XCA:DsRed2)^uu5kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 3, 5 and 7 dpf, and lack of expression in the trunk lymphatics (bottom panel) at 5 dpf. Arrowheads show expression in the developing lymphatic valve. (B) Schematic of the −2.1prox1a zebrafish and the −11Prox1 mouse enhancer. The identified sequence overlap between the two enhancers is referred to as −2.1prox1a_1 and the zebrafish unique enhancer part is referred to as −2.1prox1a_2. The TF binding site location in −2.1prox1a_1 is illustrated in the box. (C) Confocal projections of the facial and trunk lymphatics labelled with Tg(−2.1prox1a_2:EGFP;XCA:DsRed2)^uu6kk (cyan) and Tg(prox1a:RFP)^nim5 (magenta) at 5 dpf. Arrowheads show expression in the facial lymphatics. (D) Confocal projections of the immunostaining against Prox1 (magenta) and Tg(−5.2lyve1b:DsRed2)^nz101 (cyan). Prox1 expression is detected in the valve leaflets at 5 dpf (arrowheads). Scale bars: 50 μm (A,C); 20 μm (D).

−2.1prox1a is necessary for correct valve development and function. (A) Top and middle: confocal projections of immunostaining against Prox1 (cyan) and Tg(−5.2lyve1b:DsRed2)^nz101 (magenta) in sibling and Δ−2.1prox1a mutant embryos at 5 dpf. Bottom: heatmap visualisation of Prox1 protein levels in sibling and Δ−2.1prox1a mutant embryos. Left: quantification of Prox1 protein levels in the valves of 5 dpf embryos. Mean±s.d. Sibling (n=8) versus mutants (n=7). P=0.0249 (unpaired two-tailed Student's t-test). Three technical replicates, biological replicates correspond to the number of data points per condition. (B) Right: confocal projections of Tg(fli1:nEGFP)^y7 (cyan) and Tg(−5.2lyve1b:DsRed2)^nz101 (magenta) in sibling and Δ−2.1prox1a mutant embryos at 5 dpf. Left: quantification of vessel section at the valve (arrowheads) in Δ−2.1prox1a embryos at 5 dpf. Mean±s.d. Sibling (n=22) versus mutants (n=24). Not significant (ns) (P=0.627; two-tailed Mann–Whitney test). Three technical replicates, biological replicates correspond to the number of data points per condition. (C) Quantification of leaflet morphology in the valves of 5, 7 and 14 dpf sibling and Δ−2.1prox1a embryos. Mean±s.d. 5 dpf siblings (n=19) versus Δ−2.1prox1a (n=26), ns (P>0.999; Fisher's test). 7 dpf siblings (n=27) versus Δ−2.1prox1a (n=29), ns (P=0.103; Fisher's test). 14 dpf siblings (n=12) versus Δ−2.1prox1a (n=21), ns (P=0.160; Fisher's test). Three technical replicates, biological replicates correspond to the number of data points per condition. (D) Left: brightfield and TEM imaging of sibling (n=3) and Δ−2.1prox1a (n=3) valves at 7 dpf. The leaflets in the TEM images are highlighted in red. Right: quantification of cell length: mean±s.d. Siblings (n=6) versus Δ−2.1prox1a (n=6). P=0.024 (unpaired two-tailed Student's t-test). Quantification of leaflet length: mean±s.d. Siblings (n=6) versus Δ−2.1prox1a (n=6). P=0.0411 (unpaired two-tailed Student's t-test). Two technical replicates, biological replicates correspond to the number of data points per condition. (E) Left: schematic and confocal projections of Tg(−5.2lyve1b:DsRed2)^nz101 (cyan) and Qtracker (magenta) in sibling and Δ−2.1prox1a mutant embryos at 7 dpf, visualising flow through the vessel. Right: quantification of Qtracker leakage through the valve in sibling and Δ−2.1prox1a embryos at 7 dpf. Mean±s.d. Siblings (n=21) versus mutants (n=12). P=0.0332 (unpaired two-tailed Student's t-test). Two technical replicates, biological replicates correspond to the number of data points per condition. Scale bars: 50 μm (A); 20 μm (B,E); 10 μm (D).

Schematic representation of the lymphatic prox1a enhancers identified in this study. The more distal enhancers, not conserved in mammals, drive expression in wide lymphatic domains. The proximal enhancers, conserved in mammals, show instead restricted activity in specific subsets of the lymphatic vasculature.