FGF-2 signaling in nasopharyngeal carcinoma modulates pericyte-macrophage crosstalk and metastasis

Abstract

Molecular signaling in the tumor microenvironment (TME) is complex, and crosstalk among various cell compartments in supporting metastasis remains poorly understood. In particular, the role of vascular pericytes, a critical cellular component in the TME, in cancer invasion and metastasis warrants further investigation. Here, we report that an elevation of FGF-2 signaling in samples from patients with nasopharyngeal carcinoma (NPC) and xenograft mouse models promoted NPC metastasis. Mechanistically, tumor cell-derived FGF-2 strongly promoted pericyte proliferation and pericyte-specific expression of an orphan chemokine (C-X-C motif) ligand 14 (CXCL14) via FGFR1/AHR signaling. Gain- and loss-of-function experiments validated that pericyte-derived CXCL14 promoted macrophage recruitment and polarization toward an M2-like phenotype. Genetic knockdown of FGF2 or genetic depletion of tumoral pericytes blocked CXCL14 expression and tumor-associated macrophage (TAM) infiltration. Pharmacological inhibition of TAMs by clodronate liposome treatment resulted in a reduction of FGF-2-induced pulmonary metastasis. Together, these findings shed light on the inflammatory role of tumoral pericytes in promoting TAM-mediated metastasis. We provide mechanistic insight into an FGF-2/FGFR1/pericyte/CXCL14/TAM stromal communication axis in NPC and propose an effective antimetastasis therapy concept by targeting a pericyte-derived inflammation for NPC or FGF-2hi tumors.

Keywords: Cancer; Immunology; Macrophages; Oncology; Pericytes.

Figure 1. FGF-2 is distinctively expressed and correlates with TAM infiltration in human NPC.

(A) Cross–data set quantitative heatmap of selected genes of various types of cancer and their adjacent control healthy tissues. Arrow points to distinctively upregulated genes in NPC. Log₂ fold changes were used for quantification. (B) Transcriptomic expression levels of FGF2 in human LUAD tissues, BRCA tissues and their adjacent healthy tissues. Sample number: control-LUAD/LUAD/control-BRCA/BRCA=347/483/291/1085. (C) Transcriptomic expression levels of FGF2 in various stages of human NPC tissues and their adjacent healthy tissues. Sample number: control/StageT1/StageT2/StageT3=10/16/11/4. (D) Human normal nasopharyngeal tissues (NNT), rhinitis tissues, and NPC tissues were stained with H&E and an anti–FGF-2 antibody (brown). Sample number: NNT/Rhinitis/NPC=3/10/6. Scale bar in upper panel: 500 μm. Scale bar in middle and lower panels: 50 μm. Quantification of FGF-2⁺ signals and FGF-2⁺ signals in stromal and epithelial components (n = 8 random fields per group). (E) NPC cancer cells were sorted by MACS from freshly tissues. qPCR quantification of FGF2 mRNA (n = 3 samples per group). (F) NNT rhinitis tissues and NPC tissues were stained. Sample number: NNT/Rhinitis/NPC=3/10/6. Scale bar in upper and middle panels: 50 μm. Scale bar in lower panel: 100 μm. Quantification of FSP1⁺ (brown), CD163⁺ (brown), CD31⁺ (red), and NG2⁺ (green) and coverage rate of NG2⁺ pericytes (n = 8 random fields per group). (G) qPCR quantification of FGF2, CD163, CD31, NG2, and FSP1 mRNA in freshly collected tissues. Sample number: Rhinitis/NPC=5/6. (H) Correlation of FGF2 and CD163 expression of human NPCs and their control healthy tissues. Sample number: Control/NPC=10/31. *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired 2-tailed Student’s t test (B, D, E, G, and H) or 1-way ANOVA with Tukey’s multiple-comparison analysis (C, D, and F). Data are presented as mean ± SD.

Figure 2. FGF-2 promotes TAM infiltration and tumor metastasis in mice.

(A) FGF-2 protein expression levels in various human tumor cell lines, including SK-MEL-5 melanoma, MCF-7 breast cancer, Hep3B hepatocellular carcinoma, A-431 squamous cell carcinoma, A549 lung cancer, SUNE-1 NPC, and 5-8F NPC (n = 3 samples per group). (B, C, G, and H) Migration and chemotactic ability of scrambled and FGF2 shRNA–transfected NPC cancer cells (B and C) and of vector and FGF-2 overexpressing T241 tumor cells (G and H). (D and I) Xenograft tumor tissues were stained with H&E, an anti-F4/80 antibody (brown), an anti-CD31 antibody, and an anti-NG2 antibody (n = 8 mice per group). Scale bar in upper panel: 50 μm. Scale bar in middle panel: 50 μm. Scale bar in lower panel: 100 μm. Quantification of F4/80⁺ signals, NG2⁺ signals, and coverage rate of NG2⁺ pericytes (n = 8 random fields per group). (E and J) Micrographs of representative cell culture dishes after incubation with blood samples from 5-8F shScrambled or 5-8F shFGF2 tumor–bearing mice (E) and from vector or FGF-2–overexpressing tumor–bearing mice (J). Blue signal indicates the crystal violet–positive tumor colonies. Scale bar: 1 cm (n = 3 samples randomly chosen from 8 mice per group). (F and K) H&E staining in the lung from 5-8F shScrambled or 5-8F shFGF2 tumor–bearing mice (F) and from vector or FGF-2–overexpressing tumor–bearing mice (K). Scale bar in upper panel: 3 mm. Scale bar in lower panel: 100 μm. Quantification of total microscopic lung metastases and various sizes of metastases (n = 3 samples randomly chosen from 8 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired 2-tailed Student’s t test (B–K) or 1-way ANOVA with Tukey’s multiple-comparison analysis (A). Data are presented as mean ± SD.

Figure 3. Pericyte-dependent mechanism of FGF-2–induced macrophage activation.

(A and B) qPCR quantification of human and mouse FGFR1, FGFR2, FGFR3, and FGFR4 mRNA levels in various cell types, including 5-8F NPC cell line, hTERT-immortalized dermal fibroblasts, isolated primary pericytes, HUVEC endothelial cells, THP-1 monocyte/macrophage cell line, mouse T241 cell line, mouse MS5 stromal fibroblasts, mouse lung isolated primary pericytes, mouse liver isolated primary endothelial cells, and murine RAW 264.7 monocyte/macrophage cell line. (C and D) Human and mouse tumor cell migration of tumor cells cocultured with various cell types in the presence or absence of FGF-2. Vehicle- or FGF-2–treated tumor cells serve as controls (n = 8 samples per group). (E and F) Human and mouse tumor cell migration of tumor cells cocultured with or without various cell types (n = 8 samples per group). (G and H) Conditioned medium of pericytes or fibroblasts in the presence or absence of FGF-2 was collected. Mouse macrophage migration (n = 8 samples per group) and chemotactic ability (n = 6 samples per group) of macrophages treated with various conditioned medium are shown. (I) Morphological changes of macrophage administrated with vehicle or the conditioned medium of FGF-2–treated pericytes. Quantification of macrophage structural changes (n = 8 random fields per group). (J and K) Human and mouse tumor cell migration of tumor cells cocultured with macrophages, which activated with FGF-2–treated pericyte conditioned medium. Tumor cells receiving the FGF-2–treated pericyte conditioned medium serve as controls (n = 8 samples per group). ***P < 0.001 by unpaired 2-tailed Student’s t test (C, D, and G–K) or 1-way ANOVA with Tukey’s multiple-comparison analysis (A, B, E, and F). Data are presented as mean ± SD.

Figure 4. FGF-2 induces CXCL14 expression in pericytes via FGFR1/ERK/AHR signaling.

(A) Heatmap of selected genes by inflammatory cytokine/chemokine profiling of vehicle- and FGF-2–treated primary mouse pericytes (n = 3 samples per group). Arrow points to upregulated Cxcl14 gene. (B) Volcano plot of inflammatory gene profiling of vehicle- and FGF-2–stimulated pericytes (n = 3 samples per group). (C and D) Expression levels of Ccl11 and Cxcl14 in vehicle- and FGF-2–stimulated isolated primary pericytes and MS5 fibroblasts (n = 3 samples per group). (E) qPCR quantification of Cxcl14 mRNA levels in F4/80⁺ TAMs, NG2⁺ pericytes, CD31⁺ endothelial cells, and NG2^– population isolated from T241-vector and T241–FGF-2 tumors (n = 3 samples per group). (F) qPCR quantification of Cxcl14 mRNA levels in vehicle- and FGF-2–stimulated pericytes in the presence or absence of FGFR1, FGFR2, and FGFR3 specific inhibitors, and pan-FGFR inhibitor (n = 3 samples per group). (G) After 0, 15, 30 minutes of stimulation, FGF-2 induced phosphorylation of AKT and ERK in pericytes. β-Tubulin marks the loading level in each lane. These experiments were repeated twice. (H) qPCR quantification of Cxcl14 mRNA levels in vehicle- and FGF-2–stimulated pericytes in the presence or absence of MEK1/2, ERK1/2, and AKT specific inhibitors (n = 3 samples per group). (I) Volcano plot of predicted transcription factors which bind to Cxcl14 promoter in genome-wide expression profiling of vehicle- and FGF-2–stimulated pericytes (n = 3 samples per group). (J) qPCR quantification of Cxcl14 mRNA levels in vehicle- and FGF-2–stimulated pericytes in the presence or absence of Control or Ahr-specific siRNA (n = 3 samples per group). (K) ChIP assay of AHR binding to the Cxcl14 gene promoter. Nonimmune IgG and Cxcl14 exon 2 regions served as controls (n = 3 samples per group). (L) Mechanistic diagram of the FGF-2/FGFR1/ERK/AHR/CXCL14 signaling pathway. **P < 0.01, ***P < 0.001 by unpaired 2-tailed Student’s t test (C–E and K) or 1-way ANOVA with Tukey’s multiple-comparison analysis (F, H, and J). Data are presented as mean ± SD.

Figure 5. CXCL14 recruits, activates, and polarizes TAMs.

(A and B) Mouse macrophage migration (n = 8 samples per group) and chemotactic ability (n = 6 samples per group) of macrophage treated with or without CXCL14. (C) Quantification of CD45⁺ cells in xenograft shScrambled- and shFGF2-transfected NPC tumors (n = 5 samples per group). (D) Pie charts of percentage of various inflammatory cells in xenograft shScrambled- and shFGF2-transfected NPC tumors (n = 5 samples per group). CD45⁺CD11b⁺F4/80⁺ macrophage population, CD45⁺MHCII⁺CD11b⁺CD11c⁺ DC population, CD45⁺CD11b⁺Ly6G^hiLy6C^int granulocytic subsets of myeloid-derived suppressor cell population, CD45⁺CD11b⁺Ly6G^–Ly6C⁺ monocytic subsets of myeloid-derived suppressor cell population, CD45⁺B220⁺ B cell population, and CD45⁺CD11b^–CD49b⁺ NK cell population were analyzed. (E and F) Quantification of CD45⁺CD11b⁺F4/80⁺ TAM population, CD45⁺CD11b⁺ F4/80⁺CD206⁺ M2-like TAM population, and CD45⁺CD11b⁺F4/80⁺CD86⁺ M1-like TAM population (n = 5 sample per group). (G) qPCR quantification of CD206 and CD86 mRNA levels in F4/80⁺ TAMs isolated from xenograft shScrambled- and shFGF2-transfected NPC tumors (n = 3 samples per group). (H) Tumor tissues were stained with an anti-CD206 antibody (brown). Scale bar: 50 μm. Quantification of CD206⁺ signals (n = 8 random fields per group). (I and J) qPCR quantification of CD206 and CD86 mRNA levels in macrophages that were activated with FGF-2–treated pericyte conditioned medium or CXCL14. Vehicle- and FGF-2–stimulated macrophages serve as controls (n = 3 samples per group). (K) CXCL14- or FGF-2–treated pericyte conditioned medium–induced CD206 upregulation and CD86 downregulation in macrophages. β-Actin marks the loading level in each lane. These experiments were repeated twice. *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired 2-tailed Student’s t test (A–C and E–J). Data are presented as mean ± SD.

Figure 6. Genetic depletion of pericytes ablates CXCL14 and TAM infiltration in the TME.

(A) Growth rates of 4T1-vector and 4T1–FGF-2–overexpressing tumor cells in vitro. (B and C) Cell migration (n = 8 samples per group) and chemotactic ability (n = 6 samples per group) of 4T1-vector and 4T1–FGF-2–overexpressing tumor cells. (D) Tumor-bearing WT and NG2-TK mice were administrated with ganciclovir when the tumor reached 0.5 cm³. H&E staining and immunofluorescence localization of CD31 (red), NG2 (green), and DAPI (blue) signals in 4T1-vector and 4T1-FGF-2–overexpressing tumor–bearing WT and NG2-TK mice (n = 6 mice per group). Scale bar in upper panel: 50 μm. Scale bar in lower panel: 100 μm. Quantification of CD31⁺ signals, NG2⁺ signals, pericyte coverage, and average vessel diameters (n = 8 random fields per group). (E) qPCR quantification of Cxcl14 mRNA levels of 4T1-vector and 4T1–FGF-2–overexpressing tumor tissues from WT and NG2-TK mice (n = 6 mice per group). (F) F4/80 (brown) IHC in vector and FGF-2 tumor with or without NG2⁺ pericyte depletion and in CXCL14-administrated, NG2⁺ pericyte–depleted FGF-2 tumor (n = 6 mice per group). Scale bar: 50 μm. Quantification of F4/80⁺ signals (n = 8 random fields per group) (G) CD206 (brown) IHC in FGF-2 tumor with or without NG2⁺ pericyte depletion or CXCL14 administration (n = 6 mice per group). Scale bar: 50 μm. Quantification of CD206⁺ signals (n = 8 random fields per group) (H) qPCR quantification of Cd206 mRNA levels in F4/80⁺ TAMs from various tumor groups (n = 3 samples per group). *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired 2-tailed Student’s t test (A–D) or 1-way ANOVA with Tukey’s multiple-comparison analysis (E–H). Data are presented as mean ± SD.

Figure 7. Pharmacological TAM depletion diminishes FGF-2–induced NPC metastasis.

(A) Micrographs of H&E and IHC staining with F4/80 (brown) or CD206 (brown) in 5-8F shScrambled or 5-8F shFGF2 tumors implanted in clodronate-treated and nontreated mice. Scale bar: 50 μm. Quantification of F4/80⁺ and CD206⁺ signals (n = 8 random fields per group). (B) Micrographs of representative cell culture dishes after incubation with blood samples from 5-8F shScrambled or 5-8F shFGF2 tumor–bearing mice receiving vehicle or clodronate liposomes. Blue signal indicates the crystal violet-positive tumor colonies. Scale bar: 1 cm. (C) H&E staining in the lung from 5-8F shScrambled or 5-8F shFGF2 tumor–bearing mice. Scale bar in upper panel: 3 mm. Scale bar in lower panel: 100 μm. Quantification of total microscopic lung metastases and various sizes of metastases (n = 3 samples randomly chosen from 6 mice per group). (D) H&E staining in the lung from vector or FGF-2–overexpressing tumor–bearing mice. Scale bar in upper panel: 3 mm. Scale bar in lower panel: 100 μm. Quantification of total microscopic lung metastases and various sizes of metastases (n = 3 samples randomly chosen from 6 mice per group). **P < 0.01, ***P < 0.001 by 1-way ANOVA with Tukey’s multiple-comparison analysis (A–D). Data are presented as mean ± SD.

Figure 8. Schematic diagram of pericyte-associated FGF-2/FGFR1/AHR/CXCL14 axis recruits and polarizes TAMs in facilitating NPC metastasis.

(A) NPC cancer cells often produce FGF-2, and FGF-2 primarily targets pericytes and fibroblasts. In FGF-2⁺ tumors, vascular-associated pericytes and CAFs express various inflammatory regulating cytokine/chemokines. Among them, CXCL14 is produced exclusively by pericytes through FGF-2/FGFR1/AHR signaling. CXCL14 signaling recruits and polarizes TAMs into an M2-like phenotype. M2-like TAMs facilitate tumor cell intravasation and pulmonary metastasis.