Critical role for NF-kappaB-induced JunB in VEGF regulation and tumor angiogenesis

Abstract

Regulation of vascular endothelial growth factor (VEGF) expression is a complex process involving a plethora of transcriptional regulators. The AP-1 transcription factor is considered as facilitator of hypoxia-induced VEGF expression through interaction with hypoxia-inducible factor (HIF) which plays a major role in mediating the cellular hypoxia response. As yet, both the decisive AP-1 subunit leading to VEGF induction and the molecular mechanism by which this subunit is activated have not been deciphered. Here, we demonstrate that the AP-1 subunit junB is a target gene of hypoxia-induced signaling via NF-kappaB. Loss of JunB in various cell types results in severely impaired hypoxia-induced VEGF expression, although HIF is present and becomes stabilized. Thus, we identify JunB as a critical independent regulator of VEGF transcription and provide a mechanistic explanation for the inherent vascular phenotypes seen in JunB-deficient embryos, ex vivo allantois explants and in vitro differentiated embryoid bodies. In support of these findings, tumor angiogenesis was impaired in junB(-/-) teratocarcinomas because of severely impaired paracrine-acting VEGF and the subsequent inability to efficiently recruit host-derived vessels.

Figure 1

Inherent vascular defects in junB^−/− tissues due to impaired VEGF expression. (A) Analyses of vascular network formation of wild-type and junB^−/− yolk sacs (E 9.5), differentiated EBs (day 11) and allantois explants (40 h) from early headfold-stage embryos by β-galactosidase staining for yolk sacs and allantois explants or whole mount immunohiostochemical staining for CD31. Size bar corresponds to 120 μm for the yolk sacs, 70 μm for EBs and 40 μm or 16 μm for wild-type and junB^−/− allantois explants, respectively. (B) Ectopic VEGF can rescue the phenotype of junB^−/− EBs. Top, black and white photographs of CD31-stained EBs derived from wild-type or junB^−/− ES cells differentiated for 11 days. Every second day, 50 ng/ml recombinant human VEGF had been added to the cultures. Size bar corresponds to 40 μm. Bottom, quantification of EBs differentiated for 10 and 11 days. EBs that display an elaborate organized vascularization according to the criteria of Vittet et al (1997) were counted. Error bars represent s.d. values of at least 30 different EBs analyzed for each time point. P=0.022 (d10) or P=0.021 (d11) versus wild type. (C) Quantitative RT–PCR analysis of junB mRNA from wild-type ES cells and MEFs and End cells. ES cells were treated with 75 μM CoCl₂, and MEFs and End cells were incubated under hypoxic conditions (Hx, 1.5% O₂) for the indicated time points; before hypoxia induction, MEFs were starved for 36 h (0.5% FCS). Relative gene expression is given; expression of normoxic wild-type cells was set to 1. (D) Quantitative RT–PCR analysis of VEGF mRNA from wild-type and junB^−/− ES cells; MEFs and End cells treated as described in (C). Enhanced VEGF transcript levels in hypoxic cells (black bars) are indicated as fold difference compared to the expression in normoxic wild-type cells (white bars), which was set to 1. Error bars in (C) and (D) show s.e.m. values of at least three independent experiments with P=0.0289 (End cells) or P=0.0005 (ES cells), P=0.0448 (MEFs) for junB^−/− versus wild-type cells.

Figure 2

Hx responses of c-Fos, c-Jun and HIF are not affected in JunB-deficient MEFs, and the JunB induction is independent of HIF. (A–C) Immunoblotting was performed for the various proteins as indicated on the left of each panel. Fifty micrograms of nuclear extracts for AP-1 members or total extract for HIF members, respectively, was used, which were isolated from wild-type and junB^−/− MEFs kept under hypoxia mimicked by treatment with 100 μM CoCl₂ for the indicated time points. RCC1 or HSC70 served as a control for equal quality and loading of nuclear extracts or total extracts, respectively.

Figure 3

junB induction is achieved via NF-κB and is independent of HIF. (A) NF-κB is rapidly translocated to the nucleus in response to hypoxia. Immunofluorescence staining for p65 on wild-type (wt) or HIF-1α^−/− MEFs kept under normoxic (Nx) or hypoxic conditions (Hx, 1.5% O₂) for 20 min. White arrows (Hx) depict NF-κB-positive nuclei. Size bar, 25 and 12.5 μm for wt and HIF-1α^−/− cells, respectively. (B) F9 cells were cotransfected with junB reporter constructs, junB-wt and junB-mNF-κB, and p65 expression vector (CMV-p65, black bars) or control vector (pGL3, white bars). Promoter activity 16 h post-transfection is shown in relation to the luciferase activity of the vector control, which was set to 1. A cotransfected Renilla luciferase reference gene was used for normalization. Error bars of at least three independent experiments show s.d. values. ^*P=0.0028 for mutant versus junB-wt. (C) Protein levels of JunB in wild-type or ΔN+ cells kept under hypoxia for the indicated time points were determined by immunoblot analysis using 50 μg of nuclear extract. Two different exposures (low and high) are given to show basal as well as induced JunB levels. RCC1 served as a control for equal quality and loading. Hx, hypoxia. (D) HIF protein levels are not affected by loss of NF-κB. Immunoblotting was performed for HIF-1α, HIF-1β and HIF-2α by using 50 μg of total protein from wild-type and ΔN+ MEFs kept under hypoxia mimicked by treatment with 100 μM CoCl₂ for the indicated time points. HSC70 served as a control for equal quality and loading of extracts. (E) Hx-mediated junB and VEGF induction is strongly diminished in fibroblasts with repressed NF-κB activity (ΔN+). Northern blot analyses of wild-type MEFs or cells with suppressed NF-κB activity exposed to hypoxia (Hx). Ethidium bromide-stained agarose gel is shown as loading control. (F) junB and VEGF can be induced by Forskolin despite the absence of NF-κB. Quantitative RT–PCR analyses of wild-type MEFs or cells with suppressed NF-κB activity treated with 10 μM Forskolin for the indicated time points were performed using specific primers for junB or VEGF.

Figure 4

JunB binds to and activates the murine VEGF promoter. (A) Left panel, top: schematic representation of the murine VEGF promoter indicating the positions of the relevant consensus-binding sites for AP-1 (black ellipse for AP-1 and gray diamond for CRE) and HIF (gray bar). Left panel, bottom: EMSA of complexes formed by in vitro-translated JunB:Fos dimers (D) on the consensus AP-1 sites (TRE) and different potential AP-1-binding sites within the VEGF promoter. Location of these sites is indicated on the top. For control, the DNA elements were incubated with reticulocyte extract only (R). Middle panel: EMSAs of complexes formed with the consensus AP-1 site (TRE) and the VEGF-AP-1 element located at −1093 using nuclear extracts from wild-type (wt) and junB^−/− fibroblasts grown under normoxic or hypoxic (Hx) conditions for 16 h. For cross-competition experiments, a 50-fold excess of non-labeled competitor TRE or −1093, as indicated on the top, was used. For supershift analysis, extracts were preincubated with a specific antiserum recognizing JunB (α-JunB). Arrows depict specific complexes. No specific protein–DNA complexes were obtained with an oligonucleotide representing a mutated version of the VEGF-AP-1 site (data not shown). Individual lanes were numbered for convenience. Right panel, top: residual AP-1-binding activity in junB^−/− nuclear extracts due to c-Jun. EMSA of complexes formed with the VEGF-AP-1 element located at −1093 using nuclear extracts from wild-type (wt) and junB^−/− fibroblasts grown under normoxic or hypoxic (Hx) conditions for 16 h. For supershift analysis, extracts were preincubated with a specific antiserum recognizing c-Jun (α-c-Jun). Right panel, bottom: in parallel, EMSAs with an Sp1 element were performed to control for equal quality and amount of extracts used. (B) ChIP analysis of the VEGF promoter was performed using an antibody specific for JunB. Specific primer sets were used to discriminate between the promoter region harboring the proximal TREs (−1140 to −733) and a control region (co) in the 5′UTR that does not contain any TRE (+654 to +844). JunB binding to the VEGF promoter region between −1140 and −733 was detected and was enhanced in End cells treated with 200 μM CoCl₂ for 4 h. (C) F9 cells were cotransfected with a JunB expression vector and luciferase reporters containing no promoter (pGL3), a JunB target (5 × TRE), wild-type VEGF or VEGF containing the mutated AP-1 site at −1093 (VEGF-mAP-1) as indicated. The fold induction upon transactivation was calculated as described in Figure 3B. (D) Hypoxia-dependent transactivation of the wild-type and VEGF reporters with mutated AP-1 (VEGF-mAP-1), mutated HIF (VEGF-mHRE) or both mutated binding sites (VEGF-mAP-1/mHRE) was measured in transiently transfected HepG2 cells. Twenty-four hours post-transfection, cells were incubated for 12 h under normoxic or hypoxic (Hx) conditions. Relative luciferase activities (RLU) are given. Fold activation of reporter plasmid under hypoxia over normoxia is depicted on top of the columns. Error bars of at least three independent experiments show s.d. values. ^*P=0.0001 (C) and P<0.01 (D) for mutant versus wild-type VEGF reporter, respectively.

Figure 5

Reduced tumor volume of JunB-deficient teratocarcinomas, lack of large blood vessels and impaired VEGF expression. (A) Photographs of a junB^−/− and a wild-type (wt) teratocarcinoma showing severe hemorrhaging in wild-type but not in the junB^−/− teratocarcinomas. (B) Tumor progression of wild-type (wt) and junB^−/− teratocarcinomas monitored for the indicated time points post injection by measuring the tumors in two dimensions (width by length) and subsequent calculation of the volumes. Error bars of at least three tumors of each genotype show s.d. values. ^*P=0.001 (d17 and d21). (C) CD31 (α-CD31, red signal) immunofluorescence staining on cryosections of two different tumor areas derived from wild-type (wt) and junB^−/− ES cells. To determine the vessel sizes, three independent tumors of wild-type and junB^−/−genotype, and of two stages, day 17 and day 24, as indicated in table, were analyzed by measuring the vessel size using the software Lucia Archive (Nikon). Five randomly chosen optical fields of each tumor were analyzed. Table shows the percentage of vessels being larger than 25 μm in diameter. Wilcoxon's rank sum test comparing all vessel diameters of day 24 wild-type and junB^−/− teratocarcinomas reveals that diameters of JunB-deficient vessels differ significantly from those of wild-type vessels (P=0.006). (D) VEGF (α-VEGF, green signal) immunofluorescence staining on cryosections of two different tumor areas derived from wild-type (wt) and junB^−/− ES cells. For CD31 and VEGF, stainings of wild-type and junB^−/− tumors isolated at day 24 are shown. Nuclei were counterstained with Hoechst 33342 (blue signal). Size bar, 100 μm.

Figure 6

Model for hypoxia-induced transcription factors HIF, JunB and NF-κB acting in concert on the murine VEGF promoter to confer maximal VEGF expression. The model is based on previous reports and our own findings; c-Fos or c-Jun may represent the most likely dimerization partners of JunB.