BACH family members regulate angiogenesis and lymphangiogenesis by modulating VEGFC expression

Abstract

Angiogenesis and lymphangiogenesis are key processes during embryogenesis as well as under physiological and pathological conditions. Vascular endothelial growth factor C (VEGFC), the ligand for both VEGFR2 and VEGFR3, is a central lymphangiogenic regulator that also drives angiogenesis. Here, we report that members of the highly conserved BACH (BTB and CNC homology) family of transcription factors regulate VEGFC expression, through direct binding to its promoter. Accordingly, down-regulation of bach2a hinders blood vessel formation and impairs lymphatic sprouting in a Vegfc-dependent manner during zebrafish embryonic development. In contrast, BACH1 overexpression enhances intratumoral blood vessel density and peritumoral lymphatic vessel diameter in ovarian and lung mouse tumor models. The effects on the vascular compartment correlate spatially and temporally with BACH1 transcriptional regulation of VEGFC expression. Altogether, our results uncover a novel role for the BACH/VEGFC signaling axis in lymphatic formation during embryogenesis and cancer, providing a novel potential target for therapeutic interventions.

Figure 1.. Spatial and temporal expression of bach2 paralog transcripts during zebrafish development.

(A) BACH putative binding sites are evolutionarily conserved in the VEGFC promoter region. Numbering is from the ATG (translation initiation) because of the difference in length of the mouse’s 5′ UTR. An arrow indicates transcription start site (TSS), and the location of the first exon is marked as a gray rectangle. The location of the BACH sites is as predicted by Genomatix Genome Analyzer MatInspector: Human (NM_005429.5 TSS at hg19, chr4:177713899 on the minus strand); Mouse (NM_009506.2 TSS at mm9, chr8:54077532 on the plus strand); and Zebrafish (NM_205734.1 TSS at Zv9, chr1:39270725 on the minus strand). (B) Semi-quantitative RT-PCR analysis of the indicated genes (bactin-β actin) in enriched GFP⁺ cells isolated by FACS from Tg(fli1:EGFP)^y1 embryos at two developmental time points, 21–24 hpf and 3 dpf (two independent experiments for each time point). (C) A lateral view of the trunk region of a wild-type zebrafish embryo at 20, 24, 30, and 48 hpf, detected with a specific bach2a or bach2b anti-sense mRNA probe. A red arrowhead indicates somite boundaries. Scale bar, 100 μm.

Figure S1.. Expression of bach2a at different stages during zebrafish development.

(A, B, C, D, E) Embryos at 20 (A), 24 (B), 30 (C, D), and 48 (E) hpf were examined by whole-mount in situ hybridization with an antisense bach2a riboprobe. Abbreviations: FB, forebrain; HB, hindbrain; M, myotomes; MB, midbrain; MHB, midbrain hindbrain boundary; OT, optic tectum; rHB, rostral hindbrain; S, somite; SB, somite-boundaries; TE, telencephalon. Scale bar, 500 μm.

Figure 2.. bach2a is essential for developmental angiogenesis in zebrafish embryos.

(A) Confocal images of the primordial hindbrain channel (PHBC, white arrow) of 30-hpf Tg(fli1:EGFP)^y1 embryos injected with control MO (10 ng), bach2a MO (3.75 ng), bach2b MO (3.75 ng), bach2b gRNA (125 ng), or vegfc MO (10 ng). Asterisk indicates the absence of PHBC. (B) Percentage of 30-hpf Tg(fli1:EGFP)^y1 embryos with intact PHBC formation after injection with control MO (10 ng, n_{Control MO} = 68), bach2a MO (3.75 ng, n_{bach2a MO} = 107; *P < 0.0001), bach2b MO (3.75 ng, n_{bach2b MO} = 48), bach2b gRNA (125 ng, n_{bach2b gRNA} = 42), or vegfc MO (10 ng, n_{vegfc MO} = 35; *P < 0.0001). Error bars, mean ± SEM. (C) Percentage of 30-hpf Tg(fli1:EGFP)^y1 embryos with intact PHBC formation after injection with control MO (10 ng, n_{Control MO} = 24) or an increased concentration of bach2b MO (5 ng, n_{bach2b MO 5ng} = 24) or (10 ng, n_{bach2b MO 10ng} = 24). Error bars, mean ± SEM; P > 0.99999. (D) Confocal projection at 30 hpf Tg(fli1:EGFP)^y1 of homozygous bach2a mutants (bach2a^mut−/−) from F2 bach2a^mut+/− incross. White arrow points at an intact PHBC detected in embryos injected with control MO (10 ng, bach2a^mut−/− + Control MO) and bach2a MO (3.75 ng, bach2a^mut−/− + bach2a MO). Asterisk indicates defects in PHBC development after injection with bach2b MO (3.75 ng, bach2a^mut−/− + bach2b MO) or bach2b gRNA (125 ng, bach2a^mut−/− + bach2b gRNA). Scale bar, 100 μm. (E) Percentage of randomly selected bach2a^+/− F2 incross progeny at 30 hpf with an intact PHBC formation injected with control MO (10 ng, n_{bach2amut + Control MO} = 50; P > 0.99999), bach2a MO (3.75 ng, n_{bach2amut + bach2a MO} = 75; *P < 0.0002), bach2b MO (3.75 ng, n_{bach2amut + bach2a MO} = 55; *P < 0.012) or bach2b gRNA (125 ng, n_{bach2amut + bach2b gRNA} = 137; *P < 0.001). After genotyping, offspring followed the expected Mendelian ratios of inheritance. Error bars, mean ± SEM. Kruskal–Wallis test in panels (B, C, E).

Figure S2.. CRISPR/Cas9–mediated editing of the bach2 paralog loci.

(A) Schematic illustration of the bach2b locus (XM_005160548.3) and the gRNA targeting exon 4. The protospacer-adjacent motif (PAM) sequence is shown in red letters and the cleavage site is indicated by a red arrowhead. (B) Agarose gel electrophoresis of PCR products amplified from 24-hpf wild-type (WT) or bach2b gRNA (125 ng, bach2b^mut)–injected Tg(fli1:EGFP)^y1 embryos using primer sets which flank the gRNA cut sites. Arrows indicate the lengths of the resultant DNA fragments. (C) Schematic diagram of the bach2a locus (GRCz10, chr17:15666119–15734205). The CRISPR guide was designed to target exon 2, with the cleavage site (red arrowhead) located 115 bp downstream of the ATG. (D) Representative genotyping PCR of a wild-type (WT) and a CRIPSR/Cas9–mediated homozygous bach2a knockout (bach2a^mut−/−) Tg(fli1:EGFP)^y1 embryo using primer sets, which flank the gRNA cut sites. Arrows indicate the lengths of the resultant DNA fragments. (E) PCR-sequencing chromatograph of the bach2a locus in a Tg(fli1:EGFP)^y1 embryo (WT) aligned with a homozygous bach2a knockout (bach2a^mut−/−). CRIPSR/Cas9–mediated genome modification at the zebrafish bach2a locus leads to the deletion of five base-pairs, which causes a frame shift in the protein sequence (translation shown below the DNA sequence). A stop codon (indicated by a dot in the translation) created 17 amino acids downstream of the deletion site, leading to the formation of a 57-amino acid–long, truncated protein. (F) Quantitative RT-PCR validation of bach2b expression conducted for wild-type (bach2a^mut+/+) and homozygous (bach2a^mut−/−) 6-dpf siblings obtained from F2 bach2a intercross. Error bars; mean ± SEM; P = 0.2466; Wilcoxon rank sum test (two independent experiments consisting of 20 single larvae in each group). (G) Semi-quantitative RT-PCR analysis of bach2a expression in wild-type (bach2a^mut+/+) and homozygous (bach2a^mut−/−) 6-dpf siblings obtained from F2 bach2a heterozygous intercross.

Figure S3.. Knockdown of bach2a causes morphological defects in zebrafish embryos.

(A) Lateral-view confocal micrographs of full-body, 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos after injection with control (10 ng, Control MO), bach2a (3.75 ng, bach2a MO), or vegfc (10 ng, vegfc MO) morpholino oligonucleotide, showing body curvature of bach2a morphants. Scale bar, 500 μm. (B) Percentage of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal morphology (no body curvature) after injection with control (10 ng, Control MO) or bach2a MO (3.75 ng, n_{Control MO} = 46; n_{bach2a MO} = 47 *P ≤ 0.0005). (C) Confocal images of the heart morphology of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos after injection with control, bach2a, or vegfc MO. H marks the heart and a white arrowhead, a pericardial edema. Scale bar, 500 μm. (D) Quantification of 2- and 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal heart morphology after injection with control MO (10 ng, Control MO) or bach2a MO (3.75 ng, 2 dpf: n_{Control MO} = 16; n_{bach2a MO} = 19; 3 dpf: n_{Control MO} = 47; n_{bach2a MO} = 58; *P < 0.0001). (E) Percentage of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal heart morphology after injection with control MO (10 ng) or vegfc MO (10 ng, n_{Control MO} = 54; n_{vegfc MO} = 41; *P < 0.0001). (F) Dorsal view of 5-dpf Tg(fli1:EGFP)^y1 zebrafish larvae after injection with control MO, bach2a MO, or vegfc MO. Red arrowhead indicates edema. Scale bar, 100 μm. (G) Percentage of 5-dpf Tg(fli1:EGFP)^y1 zebrafish larvae with edema after injection with control MO (10 ng) or bach2a MO (3.75 ng, n_{Control MO-10ng} = 22; n_{bach2a MO} = 28; *P < 0.05). (H) Quantification of 5-dpf Tg(fli1:EGFP)^y1 zebrafish larvae with edema after injection with control MO (10 ng) or vegfc MO (10 ng, n_{Control MO} = 22; n_{vegfc MO} = 53; *P < 0.0001). (I) Percentage of 2- and 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal blood flow rate after injection with control MO (10 ng) or bach2a MO (3.75, 2 dpf: n_{Control MO} = 14; n_{bach2a MO} = 19; 3 dpf: n_{Control MO} = 47; n_{bach2a MO} = 58; *P < 0.05). (J) Percentage of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal blood flow rate after injection with control MO (10 ng) or vegfc MO (10 ng: n_{Control MO} = 54; n_{vegfc MO} = 41; *P < 0.004). (B, D, E, G, H, I, J) Wilcoxon rank sum test in panels (B, D, E, G, H, I, J).

Figure S4.. No morphological defects in bach2b zebrafish morphants.

(A) Percentage of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos after injection with control MO (10 ng) or bach2b MO (3.75, 5, or 10 ng) showing normal heart morphology (n_{Control MO} = 18; n_{bach2b MO-3.75ng} = 30; n_{bach2b MO-5ng} = 30; n_{bach2b MO-10ng} = 10; P ≥ 0.3878). (B) Quantification of 5-dpf Tg(fli1:EGFP)^y1 zebrafish larvae with edema after injection with control MO (10 ng) or bach2b MO (3.75, 5, or 10 ng) (n_{Control MO} = 18; n_{bach2b MO-3.75ng} = 30; n_{bach2b MO-5ng} = 30; n_{bach2b MO-10ng} = 10; P > 0.9999). (C) Percentage of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos with a normal blood flow rate after injection with control MO (10 ng) or bach2b MO (3.75 and 5 ng) (n_{Control MO} = 18; n_{bach2b MO-3.75ng} = 30; n_{bach2b MO-5ng} = 30; P > 0.9999). (A, B, C) Kruskal–Wallis test in panels (A, B, C).

Figure 3.. bach2a is essential for parachordal cell (PAC) development in zebrafish embryos.

(A, B) Confocal projection of the trunk of 3-dpf Tg(fli1:EGFP)^y1 embryos showing PACs (white arrow) in control MO, bach2b, or bach2b gRNA-injected embryos but not after injection with bach2a or vegfc MO (white asterisk). Scale bar, 100 μm. (B) Number of PAC-containing segments (mean ± SEM) in 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos injected with control MO (10 ng) or bach2a MO (3.75 ng, n_{Control MO} = 46; bach2a MO, n_{bach2a MO} = 53; *P < 0.001). Error bars, mean ± SEM. (C) Number of PAC-containing segments in vegfc MO-injected morphants (10 ng, n_{Control MO} = 53, n_{vegfc MO} = 41; *P < 0.0001). Error bars, mean ± SEM. (D) Number of PAC-containing segments in 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos injected with indicated bach2b MO concentrations (3.75, 5, or 10 ng, n_{Control MO} = 18; n_{bach2b MO-3.75ng} = 30; n_{bach2b MO-5ng} = 30; n_{bach2b MO-10ng} = 10; P ≥ 0.2819). Error bars, mean ± SEM. (E) Quantification of PAC-containing segments in embryos injected with bach2b gRNA (125 ng, n_{Control MO} = 35 n_{bach2b gRNA} = 42; P = 0.0615). Error bars, mean ± SEM. (F) Confocal projection of the trunk region showing PAC-containing segments in Tg(fli1:EGFP)^y1-homozygous bach2a mutants (bach2a^mut−/−) from F2 bach2a^mut+/− incross. An asterisk indicates the absence of PACs in bach2a^mut−/− embryos injected with bach2b MO (bach2a^mut−/− + bach2b MO) or bach2b gRNA (bach2a^mut−/− + bach2b gRNA) and a white arrow, their presence. Scale bar, 100 μm. (G) Number of PAC-containing segments (mean ± SEM) in 3-dpf embryos randomly selected from bach2a^+/− F2 incross progeny injected with control MO (10 ng, n_{bach2amut + Control MO} = 50; P = 0.5514), bach2a MO (3.75 ng, n_{bach2amut + bach2a MO} = 75; *P < 0.0001), bach2b MO (3.75 ng, n_{bach2amut + bach2b MO} = 55; *P and **P < 0.0222), or bach2b gRNA (125 ng, n_{bach2amut + bach2b gRNA} = 137; *P and **P < 0.0001). After genotyping, offspring followed the expected Mendelian ratios of inheritance. (B, C, D, E, G) Wilcoxon rank sum test in panels (B, C, E) and Kruskal–Wallis test in panels (D, G).

Figure S5.. Expression of lyve1 is impaired in bach2a morphants.

(A) Drawing of zebrafish embryo. (B, C) Purple box indicates region imaged in panels (B, C). (B) Whole-mount in situ hybridization comparing the expression pattern (red arrowhead) of lyve1 in 24-hpf wild-type zebrafish embryos after injection with control or bach2a MO (red asterisk indicates absence of expression). (C) Whole-mount in situ hybridization using flt4 riboprobe of 24-hpf, wild-type zebrafish embryos after injection with control or bach2a MO. Scale bar, 100 μm.

Figure 4.. bach2a is necessary for thoracic duct (TD) development in zebrafish larvae.

(A) Confocal images of the TD in 4-dpf control MO- (10 ng), bach2a MO-(3.75 ng), vegfc MO- (10 ng), bach2b MO- (10 ng), or bach2a gRNA-injected Tg(fli1:EGFP)^y1 larvae. A white arrow indicates the presence of a TD and a white asterisk, its absence. Scale bar, 20 μm. (B) Percentage of 4-dpf Tg(fli1:EGFP)^y1 larvae with an intact TD after injection with control MO (10 ng) or bach2a MO (3.75 ng, n_{Control MO} = 46; n_{bach2a MO} = 52). Error bars, mean ± SEM; *P < 0.0001. (C) Percentage of larvae with a TD after injection with vegfc MO (10 ng, n_{Control MO} = 46; n_{vegfc MO} = 52). Error bars, mean ± SEM; *P < 0.0001. (D) Quantification of bach2b morphants with an intact TD after injection with indicated bach2b MO concentrations (3.75, 5, or 10 ng, n_{Control MO} = 18; n_{bach2b MO-3.75ng} = 30; n_{bach2b MO-5ng} = 30; and n_{bach2b MO-10ng} = 10; P ≥ 0.1356). Error bars, mean ± SEM. (E) Percentage of TD-containing larvae injected with bach2b gRNA (125 ng, n_{Control MO} = 35 n_{bach2b gRNA} = 42; P = 0.5461). Error bars, mean ± SEM. (F) Confocal images of a TD (white arrow) in 4-dpf Tg(fli1:EGFP)^y1: homozygous bach2a mutants (bach2a^mut−/−) derived from bach2a^mut+/− F2 incross. An asterisk indicates absence of a TD and a white arrow, its presence. Scale bar, 20 μm. (G) Analysis of Tg(fli1:EGFP)^y1 4-dpf progeny obtained from F2 bach2a heterozygous intercross. Random selection from the pool of siblings injected at the one-cell stage with control MO (10 ng, n_{bachmut + Control MO} = 50), bach2a MO (3.75 ng, n_{bach2amut + bach2a MO} = 75), bach2b MO (3.75 ng, n_{bach2amut + bach2b MO} = 55), or bach2b gRNA (125 ng, n_{bach2amut + bach2b gRNA} = 137) was found to maintain, after genotyping, the expected Mendelian ratios of inheritance. Error bars, mean ± SEM; *P < 0.0003. (B, C, D, E, G) Wilcoxon rank sum test in panels (B, C, E) and Kruskal–Wallis test in panels (D, G).

Figure S6.. Morphological abnormalities in homozygous bach2a mutants after bach2b knockdown.

(A) Fluorescent images of homozygous bach2a (bach2a^mut−/−) siblings obtained from F2 bach2a heterozygous intercross injected at the one-cell stage with control (10 ng, Control MO) or bach2b MO (3.75 ng). Scale bar, 500 μm. (A, B) Bright-field images corresponding to the larvae presented in panel (A). Red arrowhead indicates cardiac edema. Scale bar, 500 μm. (C) Lateral view of 6-dpf homozygous bach2a (bach2a^mut−/−) siblings obtained from F2 bach2a heterozygous intercross injected with control (10 ng, Control MO), bach2a (3.75 ng, bach2a MO), bach2b (3.75 ng, bach2b MO) morpholino, or bach2b gRNA (125 ng, bach2b gRNA). Red arrowhead indicates cardiac edema, whereas body edema is indicated by a black arrow. Scale bar, 500 μm.

Figure 5.. BACH1 promotes angiogenesis and lymphangiogenesis during ovarian tumor progression in mouse models.

Ex vivo analysis of subcutaneous xenografts and diaphragm specimens excised from CD-1 nude female mice implanted with control (Control) or BACH1 ectopically expressing (BACH1) human ovarian clear cell carcinoma ES2 cells. (A) Immunofluorescence labeling of blood vessels using anti-CD34 antibodies in diaphragm specimens excised from mice injected intraperitoneally with control or BACH1-expressing human ES2 cells. Scale bar, 100 μm. (B) Morphometric analysis of the diaphragm relative region covered by CD34⁺ blood vessels. Diaphragms were excised from mice inoculated intraperitoneally with either control (Control, n = 3) or BACH1-overexpressing (BACH1, n = 7) ES2 cells (mean ± SEM; *P = 0.0304). (C) Confocal z-projection images (Z dimension 7 μm) of control and BACH1-overexpressing subcutaneous-ES2 ovarian carcinoma xenografts subjected to LYVE1 immunofluorescence staining along with a modified CLARITY technique. Images demonstrate the complexity of the lymphatic vasculature. Scale bar, 100 μm. 3D reconstructions of the stacks are available in Videos 1 and 2. (D) Lymphatic vessel immunostaining, using anti-LYVE1 antibodies, of diaphragm specimens excised from mice injected intraperitoneally with control or BACH1-overexpressing ES2 ovarian carcinoma cells. A black arrow indicates infiltration of cells into the lymphatic vessel and an asterisk, their absence. Scale bar, 100 μm. (E) Morphometric analysis of the diaphragm relative region covered by LYVE1⁺ lymph vessels. Diaphragms were excised from mice inoculated intraperitoneally with either control (Control, n = 6) or BACH1-overexpressing (BACH1, n = 9) ES2 cells (mean ± SEM; *P = 0.0047). (F) Immunofluorescence double staining of LYVE1⁺ lymphatic vessels (red) and cytokeratin 7 (CK7, green) of a 4-μm-thick specimen sectioned from paraffin-embedded diaphragm excised from mouse inoculated intraperitoneally with either control or BACH1-overexpressing ES2 cells. Nuclei were counterstained with DAPI (blue). A white arrow indicates infiltration of tumor cells into the lymphatic vessels and an asterisk, their absence. Scale bar, 100 μm. (G) Transwell Matrigel invasion assay performed in vitro with ES2 ectopically expressing BACH1 and control cells. The crystal violet dye staining images of the lower chambers are shown. Scale bar, 100 μm. (G, H) Percentage of ES2 cells that invaded through the Matrigel matrix (as in panel G) normalized to total cell number (n = 2 for each group, in duplicates; mean ± SEM; *P = 0.0209). (B, E, H) Wilcoxon rank sum test in panels (B, E, H).

Figure S7.. BACH1 overexpression in mouse tumor models stimulates angiogenesis, lymphangiogenesis, and induction of VEGFC expression.

Analysis of subcutaneous tumor specimens (two mouse models: human ovarian clear cell carcinoma ES2 cells in CD-1 nude mice and mouse D122 Lewis lung carcinoma cells in C57BL/6 black mice) excised from mice implanted with either ectopic overexpression of HA-BACH1 or control cells expressing an empty vector (Control). (A) Western blot detecting HA-BACH1 and β-tubulin in protein extracts from control (−) or BACH1 (+)-ectopically expressing ES2 cell. (B) Immunofluorescence labeling of blood vessels using anti-CD34 antibodies of subcutaneous tumors excised from mice inoculated with control (Control) or BACH1 (BACH1)-ectopically expressing human ES2 cells. (C) Morphometric analysis of the relative coverage of CD34⁺ blood vessels within the tumor area (n = 5 in each group; mean ± SEM; *P < 0.0001). (D) Immunohistochemical staining of lymphatic endothelial cells using anti–LYVE1 antibodies. Scale bar, 100 μm. (E) Western blot detecting HA-BACH1 and β-tubulin in protein extracts from control (−) or Bach1 (+)-ectopically expressing mouse D122 Lewis lung carcinoma cells. (F) Immunofluorescence labeling of blood vessels using anti-CD34 antibodies of subcutaneous tumor specimens excised from mice inoculated with control (Control) or Bach1 (Bach1)-ectopically expressing mouse D122 Lewis lung carcinoma cells. (G) Morphometric comparison between the coverage of CD34⁺ blood vessels within the tumor area of control (Control, n = 9) or Bach1-overexpressing (Bach1, n = 8) D122 Lewis lung cells (mean ± SEM; *P < 0.0001). (H) Immunohistochemical staining of lymphatic endothelial cells using anti-LYVE1 antibodies. (I) Quantitative RT-PCR validation of Vegfc mRNA levels in Bach1-overexpressing mouse D122 Lewis lung tumors compared with control tumor (n = 6 per group; mean ± SEM; *P < 0.05). (J) Immunohistochemical labeling of tumor sections using anti-VEGFC antibodies and counterstaining with hematoxylin (blue). Scale bar, 100 μm. (C, G, I) Wilcoxon rank sum test in panels (C, G, I).

Figure 6.. BACH1 and VEGFC genetically interact.

(A) Conservation of BACH sites in human, mouse, and zebrafish. The distal site is completely conserved. The proximal site is fully conserved between mouse and human, whereas there are three BACH sites at very close proximity in zebrafish. All proximal sites differ by one nucleotide from the consensus sequence. (B) Chromatin immunoprecipitation assay, followed by PCR measurements, was performed using primer mapping to the above human BACH proximal and distal regulatory sites and DNA precipitated with nonspecific IgG, HA-tag, or BACH1 antibodies. (C) Schematic representation of the wild-type human VEGFC promoter-driven luciferase (Luc) reporters (blue) (pVEGFCwt-Luc) and of three constructs deleted either from proximal (nt. −623 to −603, pVEGFCΔPro-Luc) or distal (nt. −2074 to −2054, pVEGFCΔDis-Luc) BACH-binding sites or a combination thereof (nt −623 to −603 and −2074 to −2054, pVEGFCΔProDis-Luc). Numbers refer to the nucleotide positions relative to ATG (translation initiation). (D) Quantification of dual–luciferase activity in human ES2 cells driven from pVEGFCwt-Luc, pVEGFCΔPro-Luc, pVEGFCΔDis-Luc, and pVEGFCΔProDis-Luc constructs. Relative luciferase activity is shown as a percentage of the pVEGFCwt-Luc value (mean ± SEM, n = 3). *P < 0.0001, Kruskal–Wallis test. (E) Immunofluorescence staining of human ES2 cells stably expressing either an empty pIRES vector (Control) or N-terminally HA-tagged BACH1 (BACH1) with antibodies directed against the HA tag (red, left panel) or against VEGFC (red, right panel). Nuclei were counterstained with DAPI (blue). Scale bar, 20 μm. TSS, transcription start site.

Figure 7.. BACH mediates angiogenesis and lymphangiogenesis in a VEGFC-dependent manner.

(A) Whole-mount in situ hybridization of 24-hpf, wild-type zebrafish embryos demonstrating the expression of vegfc mRNA after injection with control MO (Control MO 10 ng, red arrowhead) and the absence of its expression in embryos injected with specific MO targeting bach2a (3.75 ng, bach2a MO, red asterisk). Scale bar, 100 μm. (B) Confocal images of 30-hpf Tg(fli1:EGFP)^y1 embryos co-injected with specific MOs targeting bach2a (3.75 ng) and in vitro–transcribed vegfc mRNA (800 pg, bach2a MO + vegfc mRNA) demonstrating the restoration of PHBC (white arrow). (C) Percentage of rescued PHBC defects in 30-hpf Tg(fli1:EGFP)^y1 embryos after co-injection with bach2a MO (3.75 ng) and vegfc mRNA (800 pg, n_{Control MO} = 56; n_{bach2a MO} = 58; n_{bach2a MO + vegfc mRNA} = 56). Error bars, mean ± SEM; * or **P < 0.0001. (D) Rescue of parachordal cell (PAC) development in 3-dpf Tg(fli1:EGFP)^y1 embryos after the co-injection of bach2a MO and vegfc mRNA (bach2a MO + vegfc mRNA; PACs are indicated by a white arrow). Scale bar, 100 μm. (E) Quantification of the number of PAC-containing segments (mean ± SEM) in 3-dpf Tg(fli1:EGFP)^y1 embryos after the co-injection of bach2a MO (3.75 ng) and vegfc mRNA (n_{Control MO} = 47; n_{bach2a MO} = 27; n_{bach2a MO + vegfc mRNA} = 54; * or **P < 0.01). (F) Thoracic duct (TD) formation in 4-dpf Tg(fli1:EGFP)^y1 embryos co-injected with bach2a MO and vegfc mRNA (bach2a MO + vegfc mRNA; TD is indicated by a white arrow). Scale bar, 20 μm. (G) Percentage of 4-dpf Tg(fli1:EGFP)^y1 embryos showing normal TD after bach2a MO and vegfc mRNA injection (bach2a MO + vegfc mRNA). (n_{Control MO-10ng} = 32; n_{bach2a MO} = 65; n_{bach2a MO + vegfc mRNA} = 68 (Error bars, mean ± SEM; * or **P < 0.01. (H) Immunohistochemistry labeling of control (Control) or BACH1 (BACH1) ectopically expressing ES2 ovarian carcinoma xenograft specimens using anti-VEGFC antibodies and counterstaining with hematoxylin (blue). Black asterisk localizes the region magnified in the black frame. Scale bar, 100 μm. (I) Quantitative RT-PCR measurement of VEGFC and MMP1 mRNA expression in xenograft initiated either from control or BACH1-overexpressing ES2 cells (n = 5 in each group; mean ± SEM; *P < 0.05. (C, E, G, I) Kruskal–Wallis test in panels (C, E, G) and Wilcoxon rank sum test in panel (I).

Figure S8.. vegfc mRNA levels are not affected in bach2b morphants.

Whole-mount in situ hybridization presenting the expressions of vegfc in 24-hpf embryos injected with control or bach2b MO. Presence of expression is indicated by a red arrowhead. Scale bar, 100 μm.

Figure S9.. vegfc mRNA partially rescues morphological defects in bach2a morphants.

(A) Quantification of 3-dpf Tg(fli1:EGFP)^y1 zebrafish embryos injected with control (10 ng), bach2a (3.75 ng) MO, or co-injection with bach2a MO (3.75 ng) together with in vitro–transcribed vegfc mRNA (800 pg) showing normal body morphology (n_{Control MO} = 25; n_{bach2a MO} = 41; n_{bach2a MO + vegfc mRNA} = 43; * or **P < 0.05). (B) Percentage of normal heart morphology (3.5 dpf, b: n_{Control MO} = 54; n_{bach2a MO} = 28; n_{bach2a MO + vegfc mRNA} = 33; * or **P < 0.05). (C) Percentage of rescued body edema in 5-dpf larvae (n_{Control MO} = 25; n_{bach2a MO} = 66; n_{bach2a MO + vegfc mRNA} = 70; *P ≤ 0.0001 and **P < 0.05). (D) Quantification of normal blood flow rate of 3-dpf embryos (n_{Control MO} = 54; n_{bach2a MO} = 35; n_{bach2a MO + vegfc mRNA} = 28, *P < 0.05). (A, B, C, D) Kruskal–Wallis test in panels (A, B, C, D).

Figure S10.. Expression of VEGFA and VEGFB during tumor growth in mouse models is not BACH1 dependent.

(A) Quantitative RT-PCR measurement of VEGFA expression conducted on xenograft initiated either from control or BACH1-overexpressing ES2 cell (n = 5 in each group; mean ± SEM; P = 0.6857). (B) Quantitative RT-PCR validation of VEGFB expression conducted either on control or BACH1-overexpressing ES2 xenografts (n = 5 in each group; mean ± SEM; P = 0.0556). (C) Quantitative RT-PCR measurement of Vegfa mRNA levels in Bach1-overexpressing mouse D122 Lewis lung tumors compared with control tumor (n = 6 per group; mean ± SEM; P = 0.2500). (D) Quantitative RT-PCR validation of Vegfb mRNA levels in Bach1-overexpressing mouse D122 Lewis lung tumors compared with control tumor (n = 6 per group; mean ± SEM; P > 0.9999). (A, B, C, D) Wilcoxon rank sum test in panels (A, B, C, D).

Figure 8.. Expression of BACH1 and VEGFC correlates during human cancer progression.

(A) Correlation of BACH1 and VEGFC expression in specimens from melanoma patients clinically diagnosed with a primary tumor (n = 84) in comparison with those with metastatic stage melanoma (n = 356), as deduced from The Cancer Genome Atlas Research Network RNA sequencing (RNA-Seq) data. Mean ± SEM; *P ≤ 0.0008. (B) Correlation of BACH1 and VEGFC expression in specimens from lung adenocarcinoma (LUAD) clinically diagnosed with primary tumor stages I (n = 292), II (n = 133), and III (n = 95), as analyzed from The Cancer Genome Atlas Research Network RNA-Seq. data. Mean ± SEM; *P < 0.02. (A, B) Wilcoxon rank sum test in panel (A) and Kruskal–Wallis test in panel (B).