FOSL2 promotes VEGF-independent angiogenesis by transcriptionnally activating Wnt5a in breast cancer-associated fibroblasts

Abstract

Cancer-associated fibroblasts (CAFs), a predominant component of the tumor microenvironment, contribute to aggressive angiogenesis progression. In clinical practice, traditional anti-angiogenic therapy, mainly anti-VEGF, provides extremely limited beneficial effects to breast cancer. Here, we reveal that FOS-like 2 (FOSL2), a transcription factor in breast CAFs, plays a critical role in VEGF-independent angiogenesis in stromal fibroblasts. Methods: FOSL2 and Wnt5a expression was assessed by qRT-PCR, western blotting and immunohistochemistry in primary and immortalized CAFs and clinical samples. FOSL2- or Wnt5a-silenced CAFs and FOSL2-overexpressing NFs were established to explore their proangiogenic effects. Invasion, tubule formation, three-dimensional sprouting assays, and orthotopic xenografts were conducted as angiogenesis experiments. FZD5/NF-κB/ERK signaling activation was evaluated by western blotting after blocking VEGF/VEGFR with an anti-VEGF antibody and axitinib. Dual luciferase reporter assays and chromatin immunoprecipitation were performed to test the role of FOSL2 in regulating Wnt5a expression, and Wnt5a in the serum of the patients was measured to assess its clinical diagnostic value for breast cancer patients. Results: Enhanced FOSL2 in breast CAFs was significantly associated with angiogenesis and clinical progression in patients. The supernatant from CAFs highly expressing FOSL2 strongly promoted tube formation and sprouting of human umbilical vein endothelial cells (HUVECs) in a VEGF-independent manner and angiogenesis as well as tumor growth in vivo. Mechanistically, the enhanced FOSL2 in CAFs was regulated by estrogen/cAMP/PKA signaling. Wnt5a, a direct target of FOSL2, specifically activated FZD5/NF-κB/ERK signaling in HUVECs to promote VEGF-independent angiogenesis. In addition, a high level of Wnt5a was commonly detected in the serum of breast cancer patients and closely correlated with microvessel density in breast tumor tissues, suggesting a promising clinical value of Wnt5a for breast cancer diagnostics. Conclusion: FOSL2/Wnt5a signaling plays an essential role in breast cancer angiogenesis in a VEGF-independent manner, and targeting the FOSL2/Wnt5a signaling axis in CAFs may offer a potential option for antiangiogenesis therapy.

Keywords: FOSL2; VEGF-independent angiogenesis; Wnt5a; cancer-associated fibroblasts.

Figure 1

FOSL2 expression is upregulated and positively correlates with angiogenesis in breast cancer CAFs. (A) The known human transcription factors were downloaded from a public database (http://humantfs.ccbr.utoronto.ca/), the upregulated genes in CAFs were identified using microarray data, and the downregulated genes of preeclampsia from the GEO database of GSE99007 were analyzed by bioinformatics. The Venn diagram shows twenty dysregulated transcription factors in CAFs. (B) Heatmap of the dysregulated transcription factors in 20 paired breast CAFs and NFs detected by Agilent mRNA microarrays (fold changes>1.5; p<0.05, CAFs vs NFs). (C, D) western blot analysis to determine the levels of FOSL2 protein in 12 paired primary CAFs and their NFs isolated from breast tumor tissues (C) and in 2 paired immortalized NFs and CAFs (D). β-actin was the loading control. (E) Percentages of specimens with low or high expression of FOSL2 according to stage. (F) Kaplan-Meier survival curve of overall survival in 117 patients with carcinoma according to stromal FOSL2 expression. (G) Representative IHC staining image showing the increased microvessel density in breast cancer tissues with high FOSL2 expression. (H) Quantity of MVD in breast tumor tissues with high or low FOSL2 expression. (I) The correlation between the levels of FOSL2 and CD31 in breast tumor samples from TCGA (r=0.55, p<0.001). (J) GSEA analysis of TCGA database showing a positive correlation between FOSL2 expression levels and the enrichment of angiogenesis-related genes.

Figure 2

FOSL2 promotes CAF angiogenesis in vitro. (A-B) FOSL2 levels were detected by qRT-PCR or western blotting in the indicated cells (**p<0.01). (C) Representative images of HUVEC recruitment (left panel), HUVEC tube formation (middle panel), and HUVEC spheroid spouting (right panel) using conditioned media (CMs) are shown (scale bar, 100 μm). (D-F) Quantification analysis of the recruitment of HUVECs (D), the average branch number (E), and the cumulative sprout length (F) of the different groups are shown.

Figure 3

FOSL2 promotes CAF angiogenesis in a VEGF-independent manner. (A) VEGFA levels were detected by western blotting and normalized to β-actin in the indicated cells. (B) ELISA was used to determine the VEGFA concentrations in the supernatants of the indicated cells. (C, E) Representative images of the formation of HUVEC tubes following incubation with CM derived from CAFs or NFs treated with VEGF Ab; the average branch number was calculated. (D, F) Representative images of the spheroid spouting of HUVECs following incubation with CM collected from CAFs or NFs treated with VEGF Ab; the cumulative sprout length was calculated (scale bar, 100 μm).

Figure 4

FOSL2 is regulated by the estrogen/cAMP/PKA signaling axis. (A) The mRNA levels of CYP19A1 were determined by qRT-PCR in CAFs and NFs. Gene expression was normalized by β-actin (**p<0.01). (.B) Testosterone (100 nM) was added to the culture medium of breast NFs and CAFs. E2 concentrations were determined by chemiluminescence immunoassay (**p<0.01). (C) The production of cAMP in NFs and CAFs was detected by an ELISA Kit (R&D System, USA). Production of cAMP was increased under treatment with E2 (100 nM) in NFs (**p<0.01). (D-F) Protein levels of FOSL2 were detected in CAFs stimulated with SQ22536 (100 μM) (D) and H89 (5 μM) (E) or NFs treated with E2 (100 nM) (F) for 12 h. (G) CYP19A1 levels were detected by qRT-PCR in CAFs transfected with CYP19A1 siRNA (siCYP19A1) and control cells (**p<0.01). (H, I) FOSL2 levels were evaluated by qRT-PCR (H) and western blotting (I) analysis in the indicated cells (*p<0.05, **p<0.01).

Figure 5

FOSL2 transcriptionally activates Wnt5a. (A) The predicted target genes regulated by FOSL2 from the Cistrome database and a set of genes related to angiogenesis and the reported proteins in the extracellular matrix downloaded from AmiGO Gene Ontology were used in our analysis to acquire the FOSL2-regulated angiogenesis genes. Sixty-eight putative FOSL2 targets are shown. (B) The significantly altered cytokine genes (fold changes>1.5 times, CAFs vs NFs) were identified by microarray analysis for primary CAFs and NFs. (C-D) Expression of Wnt5a was determined by qRT-PCR (C) or western blotting (D) in the indicated stromal fibroblasts (**p<0.01). (E) Schematic diagram of canonical FOSL2-binding motif (JASPAR Database) and potential FOSL2 responsive elements (E1, E2, E3) in the Wnt5a promoter. Full-length and truncated Wnt5a promoter or E2 mutated promoter are shown. (F) Transcriptional activities of FOSL2 on Wnt5a were determined by luciferase reporter assay as CAFs were transfected with full-length or truncated Wnt5a promoter (upper panel) and E2 wild-type or mutated reporter plasmids (down panel). (G) ChIP assays were performed to analyze FOSL2 binding to the Wnt5a promoter using an anti-FOSL2 antibody. The E2 site was significantly enriched in comparison with the E1 or E3 site. (H) western blotting to determine Wnt5a collected from CM derived from the indicated cells.

Figure 6

Wnt5a is involved in the FOSL2-mediated promotion of angiogenesis in CAFs. (A-F) Cell invasion (A), tubule formation (C), and spheroid spouting (E) abilities of HUVECs were tested in the presence of CM from FOSL2-silenced CAFs treated with or without rWnt5a and the control CAFs or CM from NFs transfected with FOSL2 under treatment with or without BOX5 and their control NFs (scale bar, 40 μm). Quantification analysis of the recruitment of HUVECs (B), the average branch number (D), and the cumulative sprout length (F) of the different groups are shown.

Figure 7

Wnt5a derived from CAFs activates FZD5/NF-κB/ERK signaling in endothelial cells. (A) FZD5 levels were detected by qRT-PCR in HUVECs transfected with FZD5 siRNA (siFZD5) and control cells (**p<0.01). (B) Cell invasion and tubule formation of the indicated HUVECs (control HUVECs, HUVECs treated with rWnt5a, and FZD5-silenced HUVECs under treatment with rWnt5a) were assessed (scale bar, 100 μm). (C-D) Quantification analysis of the recruitment of HUVECs (C) and the average branch number (D) of the different groups of HUVECs are shown. (E-F) HUVECs were treated with different CM. The CM was as follows: CAF CM of scramble and loss of FOSL2 or Wnt5a in CAFs, CAF CM with siFZD5, and FBS-free medium with rWnt5a. The total and phosphorylated proteins of P65/NF-κB, ERK, VEGFR2, PI3K/AKT and β-catenin in HUVECs were determined by western blotting. β-actin was used as a loading control. Levels of p-P65 and p-ERK (F) were quantified as relative pixel intensity to β-actin.

Figure 8

FOSL2 modulates breast tumor angiogenesis and tumor growth in xenograft mouse models and clinical breast cancer patients. (A) Representative images of blood vessels on the tumor surface (labeled with black arrow) of mice injected with the mixture of MDA-MB-231 and indicated stromal fibroblasts under treatment with or without VEGF Ab, rWnt5a or BOX5. The blood vessel structures in the H&E and IHC staining of CD31 in tumors of each group of mice are shown (scale bar, 50 μm). (B-C) Representative tumor size (B) and tumor weight (C) for each group of xenograft mice (7 mice in each group) are shown (**p<0.01). (D) The levels of Wnt5a in the serum of 71 patients were positively correlated with the microvascular density in the tumor tissues of breast cancer patients. The red circles represent patients whose primary CAF/NF counterparts were successfully isolated from the tissue. (E) FOSL2 protein levels in breast tumor tissues were positively correlated with Wnt5a in tumor tissue extracts. The red squares represent patients whose primary CAFs/NFs were successfully isolated from the tissue. (F) FOSL2 protein levels were closely correlated with Wnt5a in the supernatant of 12 pairs of primary CAFs and NFs.