a Schematic diagram illustrating vSMC coverage of arteries and lack of coverage of veins. b Schematic diagram of a zebrafish larva with the red box highlighting the area imaged in panel c. c Confocal micrograph of the anterior trunk of a 4 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic zebrafish larva expressing eGFP in vSMCs (green) and mCherry-CAAX in the endothelium (magenta)². vSMCs are associated with the dorsal aorta (DA) and not the cardinal vein (CV). d Schematic diagram of a zebrafish larva with the red box highlighting the area imaged in panels e and f. e, f Whole mount in situ hybridization of the mid-trunk of 48 hpf zebrafish larvae probed for cxcl2b ligand (e) or cxcr4a receptor (f). Expression of both genes is enriched in the dorsal aorta compared to the cardinal vein. n = 20; scale bars = 75 µm. g Schematic representation of arterial-enriched expression of cxcl12b and cxcr4a.

a, b Confocal images (left) and schematic representations (right) of the dorsal aorta (DA) in the anterior trunk of 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic sibling (a) or cxcl12b^Δ24/Δ24 mutant (b) zebrafish expressing eGFP in vSMCs (green) and mCherry-CAAX in the endothelium (magenta). c, d Quantification of the number of associated vSMC (c) and width (d) of the dorsal aorta in 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic sibling (black columns; n = 37 (c) and n = 60 (d)) or cxcl12b^Δ24/Δ24 mutant (green columns; n = 15 (c) and n = 50 (d)) larvae. Values are expressed as a percentage of control siblings and averaged from three individual experiments. e, f Confocal images (left) and schematic representations (right) of the dorsal aorta (DA) in the anterior trunk of 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic siblings (e) or cxcr4^Δ7/Δ7 mutant (f) zebrafish expressing eGFP in vSMCs (green) and mCherry-CAAX in the endothelium (magenta). The schematic representations demonstrate what were counted as cells in the adjacent panels. g, h Quantification of the number of associated vSMC (G) and width (h) of the dorsal aorta in 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic sibling (black columns; n = 25 (g) and n = 30 (h)) or cxcr4^Δ7/Δ7 mutant (purple columns; n = 19 (g) and n = 30 (h)) larvae. Values are expressed as a percentage of control siblings and averaged from three individual experiments. i, j Confocal images of immunohistochemically stained transverse sections through the dorsal aorta of E12.5 Cxcr4^+/− heterozygous sibling (i) and Cxcr4^−/− mutant (j) mice, probed for platelet endothelial cell adhesion molecule-1 (PECAM) for endothelium (blue) and alpha smooth muscle actin (SMA) for vascular smooth muscle (vSMC, red). White brackets note the thickness of the vSMC layer surrounding the DA. k, l Quantification of aortic wall thickness (k) and lumenal area (l) of E12.5 Cxcr4^+/− heterozygous sibling (black columns; n = 26 (k) and n = 6 (l)) and Cxcr4^−/− mutant (purple columns; n = 37 (k) and n = 6 (l)) mice, measured from immunohistochemically stained sections as in panels i and j. The range for wall thickness was defined by the inner and outermost SMA staining detected in the images. Values are expressed as a percentage of heterozygous siblings and averaged from three individual experiments. Scale bars = 75 µm (panels a, b, e, f), 400 µm (panels i, j). Box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5x above and below the interquartile range. All data points are shown as individual dots, with outliers shown above or below the whiskers. P-values are indicated above statistically significant datasets and were generated using student’s t-tests.

a, b Schematic diagrams illustrating the experimental design for using the mrc1a promoter to drive ectopic mosaic expression of cxcl12b in veins. a A Tol2(mrc1a:cxcl12b-2a-mCherry) DNA construct co-translationally expressing cxcl12b and mCherry under the control of the mrc1a promoter is injected into Tg(tagln:eGFP) transgenic zebrafish embryos at the 1 cell stage. b At 4 dpf tol2(mrc1a:cxcl12b-2a-mCherry)-injected zebrafish larvae are analyzed for vSMC (eGFP) association at sites of mCherry (i.e., cxcl12b) expression in the dorsal aorta and cardinal vein. c, d Representative confocal images of the mid-trunk of 4 dpf Tg(tagln:eGFP) transgenic larvae injected with either control Tol2(mrc1a) “empty vector” (c) or Tol2(mrc1a:cxcl12b-2a-mCherry) (d). eGFP-expressing vSMCs are shown in green, cxcl12b-2a-mCherry expression in dorsal aorta (DA) or cardinal vein (CV) endothelium is shown in magenta. e Quantification of eGFP-positive vSMC associated with the dorsal aorta (DA) or cardinal vein (CV) in 4 dpf Tg(tagln:eGFP) transgenic zebrafish injected with either control Tol2(mrc1a) “empty vector” (black columns; n = 3 for artery and vein) or Tol2(mrc1a:cxcl12b-2a-mCherry) (green columns; n = 5 artery, n = 6 vein), showing strongly increased association of vSMCs with the cardinal vein. f Schematic diagrams showing potential models for direct (left) versus indirect (right) mechanisms for promoting arterial recruitment of vSMC via CXCL12. g Schematic diagram illustrating the 3-D coronary artery smooth muscle cell (CASMC) motility assay. CXCL12, PDGFB, or nothing (control) is placed within the collagen gel to determine if CASMCs migrate towards these potential chemoattractants. h Representative lateral images of 3-D collagen gels showing CASMCs within the collagen matrix for each gel condition. i Quantification of the relative number of CASMCs invading the collagen gel. The control is set to 100% and the CXCL12 and PDGFB conditions normalized to this level of invasion. Scale bars = 75 µm (panels c, d), 200 µm (panel h); n = 6. Box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5x above and below the interquartile range. All data points are shown as individual dots, with outliers shown above or below the whiskers. P-values are indicated above statistically significant datasets. Statistics in panels e and i were run using two-way ANOVA to calculate P-values.

a Schematic representation of the area imaged for in situ hybridization analysis, demonstrating the location of the dorsal aorta (DA) and cardinal vein (CV) within these images. b, c Whole mount in situ hybridization (WISH) images of pdgfba transcript (purple) in 34 hpf (b) and 72 hpf (c) zebrafish. Red dashed lines outline the dorsal aorta, blue dashed lines outline the cardinal vein. An expanded time course including these images can be found in Supplemental Fig. 1. d–h Confocal images (left) and schematic representations (right) of the dorsal aorta (DA) in the anterior trunk of 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic pdgfbb^+/−, pdgfba^+/+ sibling (d); pdgfbb^+/−, pdgfba^+/− (e); pdgfbb^+/−, pdgfba^−/− (f); pdgfbb^−/−, pdgfba^+/− (g); or, pdgfbb^−/−, pdgfba^−/− double homozygous mutant (h) zebrafish expressing eGFP in vSMCs (green) and mCherry-CAAX in the endothelium (magenta). The schematic representations demonstrate what were counted as cells in the adjacent panels. Scale bars = 75 µm. i, j Quantification of the number of associated vSMCs (i) and width (j) of the dorsal aorta in 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) double-transgenic zebrafish carrying different combinations of heterozygous or homozygous pdgfba and pdgfbb mutants. Values are averaged from three individual experiments and expressed as a percentage of the pdgfbb^+/−, pdgfba^+/+ control. Total n number per condition is shown below each sample in the graphs. k Schematic diagram illustrating the proposed model for endothelial-autonomous chemokine signaling driving increased endothelial PDGFB ligand production, and thereby indirectly promoting vSMC acquisition by arteries. Box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5x above and below the interquartile range. All data points are shown as individual dots, with outliers shown above or below the whiskers. P-values are indicated above statistically significant datasets and were calculated using one-way ANOVA.

a, b PDGFB transcript (a) and protein (b) in HUVEC cells cultured in vitro in a confluent cell monolayer for up to 8 h with (“+CXCL12”) or without (“CTRL”) added recombinant CXCL12. Relative PDGFB transcript levels (a) and protein levels (b) were measured by qPCR and Western blot, respectively, showing an upregulation of both PDGFB transcript and PDGFB protein levels in response to stimulation by CXCL12. The PDGFB western blots are shown in comparison to two loading controls, tubulin and GAPDH. c–e PDGFB transcript (c) and protein levels (d, e) in HUVEC cells cultured in vitro in a confluent cell monolayer and treated with either control, CXCR4 (target #1 or target #2- all experimental quantification utilizes siRNA target #1), or CXCL12 siRNAs. Relative PDGFB transcript (c) and protein (e, f) levels were measured by qPCR and western blot, respectively, showing suppression of both PDGFB transcript and protein in response to either CXCR4 or CXCL12 knockdown. The PDGFB western blots are shown in comparison to two loading controls, tubulin (top) and GAPDH (bottom), and the quantification is normalized to tubulin expression. Values in a, c, and e are representative of three individual experiments and expressed as a percentage of control. Samples are generated by pooling 20 individual embryos per condition. (a, c). n = 3 for all in vitro experiments. f Confocal images of immunohistochemically stained transverse sections through large arteries and large veins of E12.5 Cxcr4^+/− heterozygous sibling (left) and Cxcr4^−/− mutant (right) mice, probed for platelet-derived growth factor B (PDGFB; green) and for smooth muscle 22 alpha (SM22, aka transgelin) for vascular smooth muscle cells (vSMC, red). g Quantification of relative PDGFB protein expression in Cxcr4^+/− heterozygous embryos (black bars; n = 12 arteries, n = 3 veins) versus Cxcr4^−/− homozygous mutant embryos (purple bars; n = 15 arteries, n = 4 veins). Values are expressed as a percentage of the heterozygous control condition and images were acquired from three to five individual mice per condition. h Schematic diagram of a zebrafish larva with the red box highlighting the area imaged in the lower panels. Lower panels: Whole mount in situ hybridization of the mid-trunk of 56 hpf zebrafish injected with control (left) or cxcl12b (right) RNA at the one-cell stage, showing upregulation of pdgfba transcript in response to exogenous cxcl12b. Red and blue dotted lines in the panels indicate the dorsal aorta and cardinal vein, respectively. Images are representative of 20 embryos. i Western blot of whole embryo protein lysate from 54 hpf zebrafish injected with either control (left) or cxcl12b (right) RNA, probed for pdgfb (top) or alpha tubulin (bottom), showing upregulation of pdgfb protein levels in response to exogenous overexpression of cxcl12b. Images are representative of data from three individual experiments. j Schematic diagram illustrating the proposed model for endothelial-autonomous chemokine signaling driving increased endothelial PDGFB ligand production, thereby indirectly promoting vSMC acquisition by arteries. Putative upstream regulators of CXCL12 and CXCR4 are noted in red. Scale bars = 100 µm (f) and 50 µm (h). Box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5x above and below the interquartile range. All data points are shown as individual dots, with outliers shown above or below the whiskers. P-values are indicated above statistically significant datasets. Statistics in panels c, e were run using one-way ANOVA and g using two-way ANOVA to calculate the P-values.

a Top: Schematic diagram of a zebrafish larva with the red box highlighting the area imaged below. Bottom: Representative whole mount in situ hybridization (WISH) image of a 60 hpf zebrafish probed for klf2a. Red dashed lines indicate the dorsal aorta; blue dashed lines indicate the cardinal vein. b, c Fluorescent images of transverse sections through E9.5 mice: IHC for GFP was done to amplify signal from a Klf2-GFP knockin allele where the GFP is fused to the N-terminus of Klf2 (Klf2; green) and for platelet endothelial cell adhesion molecule-1 (PECAM) to mark the endothelium (red). Nuclei are labeled with DAPI (blue). Arrows highlight Klf2-positive endothelial nuclei in the cardinal vein (b) and vitelline vein (c)^32,84. d, e Quantitative qPCR measurement of CXCR4 (d) and PDGFB (e) transcript levels in HUVEC cells cultured in vitro as a confluent cell monolayer and treated with either control (black columns) or KLF2 (blue columns) siRNA. Values are expressed as a percentage of control and are representative of three individual experiments. Samples are generated by pooling 20 individual embryos per condition. f–h Representative western blot images of CXCR4 and PDGFB protein levels (f), and quantification of relative CXCR4 (g) and PDGFB protein levels (h) from HUVEC cells cultured in vitro in a confluent cell monolayer and treated with either control or KLF2 siRNA. Values in g, h are averaged from three individual experiments and expressed as a percentage of control. I, J Quantification of vSMC number associated with the dorsal aorta (i) or cardinal vein (j) in the mid-trunk of 5 dpf control wild type/heterozygous siblings (black columns; n = 44 (i, j)) or klf2a^Δ8/Δ8 mutant (blue columns; n = 14 (i, j)) animals. Values are averaged from data collected from three separate experiments and are expressed as a percentage of the control sibling average. k Quantification of dorsal aorta (DA, left columns) and cardinal vein (CV, right columns) width in the mid-trunk of 5 dpf control sibling (black columns; n = 90) or klf2a^Δ8/Δ8 mutant (blue columns; n = 60) animals. Values are averaged from data collected from three separate experiments and are expressed as a percentage of the average artery width in the control siblings. l, m Confocal images of the anterior trunk of 5 dpf Tg(tagln:eGFP), Tg(kdrl:mCherry-CAAX) control sibling (l) and klf2a^Δ8/Δ8 mutant (m) zebrafish embryos with eGFP-positive vSMCs (green) and mCherry-CAAX-positive endothelial cells (magenta). Arrows point to vSMCs associated with the CV in control versus klf2a^Δ8/Δ8 mutants. Scale bars = 300 µm (panels b, c) and 75 µm (panels l, m). Box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5x above and below the interquartile range. All data points are shown as individual dots, with outliers shown above or below the whiskers. P-values are indicated above statistically significant datasets. Statistics in panels d, e were generated via two-way ANOVA and panel k were run using two-way ANOVA; all other panels employ student’s t-tests to calculate the P-values.

a During the early development of arterial and venous blood vessels the CXCL12 ligand, its receptor CXCR4, and the vascular smooth muscle cell (vSMC) chemoattractant PDGFB are all more highly expressed on arteries than on veins. In contrast, KLF2 is more highly expressed on primitive veins than arteries. b A proposed molecular pathway for preferential recruitment of vSMCs to arteries. In arteries, autocrine activation of endothelial CXCR4 by its CXCL12 ligand results in increased production of PDGFB by arterial endothelium, promoting vSMC recruitment to arteries. Expression of KLF2 in primitive veins suppresses expression of CXCL12 and CXCR4, preventing upregulation of PDGFB and limiting vSMC recruitment to veins. The cues directing preferential expression of KLF2 in veins remain unclear, but may involve different types of flow (i.e., pulsatile vs. laminar).