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

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.


Introduction
Nasopharyngeal carcinoma (NPC) accounts for 73,000 deaths in 2018, and Southeast Asia exhibits the highest incidence (1,2). Commonly contributing factors in NPC development include Epstein-Barr virus (EBV) infection, genetic susceptibility, and lifestyle (2). Clinically, radiotherapy and chemotherapy are recommended for early-stage NPC and nonmetastatic NPC patients (3). However, therapeutic options for metastatic NPC patients are limited. Metastatic NPC appears to be a heterogeneous group of tumors with a wide range of survival, and lung, liver, and bone are the most common sites of distant metastases (4). Targeted therapy is recognized as an effective approach to further prolong the survival of NPC patients. Nevertheless, several clinical trials show that targeting the vascular endothelial growth factor (VEGF) signaling by bevacizumab, or targeting the epidermal growth factor (EGF) signaling by cetuximab, did not show clinical benefits in NPC patients, compared with conventional chemoradiotherapy (5)(6)(7). Hence, novel molecular-targeted therapies for NPC are urgently warranted. The mechanistic study of NPC metastasis is the foundation of developing novel targeted therapies. Currently, NPC metastasis studies are mostly 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 TAMmediated 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-2 hi tumors.

FGF-2 expression and TAMs infiltration in human NPC tissues.
In the TME, various signaling molecules orchestrate cancer metastasis via autocrine, paracrine, and endocrine mechanisms (14). To investigate the molecular mechanism of NPC metastasis, we applied a gene expression profiling analysis associated with metastasis-related growth factors to screen NPC and various other cancer types. Twenty-three metastasis-related growth factors were selected for the screening. Tissue RNA expression data sets of major cancer types including colon adenocarcinoma (COAD), breast invasive carcinoma (BRCA), lung adenocarcinoma (LUAD), ovarian cancer (OV), stomach adenocarcinoma (STAD), skin cutaneous melanoma (SKCM), pancreatic adenocarcinoma (PAAD), liver hepatocellular carcinoma (LIHC), and kidney renal clear cell carcinoma (KIRC) and their corresponding controls were downloaded from The Cancer Genome Atlas (TCGA). Tissue RNA expression data sets of NPC and its corresponding control with accession no. GSE12452 were downloaded from the Gene Expression Omnibus (GEO). We compared a panel of selected genes among these expression profiles in each tumor type with its respective adjacent control tissues. Interestingly, VEGFB in NPC showed similar expression levels among various cancers. VEGFC and TNFB were expressed at low levels in NPC but high levels in most of the non-NPC cancer types. Surprisingly, FGF2, a potent mitogenic factor of the FGF family, was exclusively highly expressed in NPC ( Figure 1A). TCGA analysis further confirmed the lower FGF2 expression in non-NPC cancer tissues ( Figure 1B and Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi. org/10.1172/jci.insight.157874DS1). Next, we applied the correlation study of FGF2 expression in various clinical stages of NPCs compared with normal nasopharyngeal tissue (NNT) using the GSE12452 data set containing 31 NPC tissue specimens and 10 healthy NNT specimens. It demonstrated the higher expression of FGF2 in all stages of NPCs ( Figure 1C). No expression difference among various NPC stages was observed ( Figure 1C).
To further identify the cell type origin of FGF-2 in NPC tissues, we collected 3 NNTs, 10 rhinitis tissues, and 6 NPC tissues from patients receiving a nasopharyngoscopy test. The demographic information of these patients was shown (Supplemental Table 1). Histological analysis showed significantly high FGF-2 expression in NPC tissues compared with non-NPC tissues. Quantification analysis under the supervision of an experienced pathologist showed over a 7-fold increase of FGF-2 + signals in NPCs relative to NNTs or to rhinitis tissues ( Figure 1D). Moreover, the major cellular component expressing FGF-2 in the NPC microenvironment was epithelial cells, indicating that FGF-2 originated from NPC cancer cells ( Figure  1D). To further validate the source of FGF-2 production in the TME, we applied a negative selection strategy to isolate NPC cancer cells without knowing cancer cell surface marker expression. Cancer cells were isolated from 3 fresh NPC tissues using a magnetic-activated cell sorting (MACS) kit. As expected, cancer cells showed significantly high FGF-2 expressions ( Figure 1E). These results demonstrate the distinct FGF-2 expression in NPC cancer cells.
To investigate potential structural changes induced by FGF-2 signaling, various cellular components were analyzed by staining with fibroblast-specific protein 1 + (FSP1 + ), CD163 + signals, CD31 + , and neuron-glial antigen 2 + (NG2 + ) signals. Interestingly, fibroblasts were not abundant cellular components in NPC tumor tissues ( Figure 1F). In contrast, macrophage infiltration into NPC tumor tissues was highly increased. Quantification analysis showed that over a 4-fold increase of CD163 + signals was found in NPCs relative to NNTs ( Figure 1F). A higher number of vessels was observed in NPCs, and the pericyte coverage of these vessels is similar to that in NNTs ( Figure 1F). These data demonstrate a significant increase of TAM infiltration in NPC tissues. To further validate that these findings are specific to NPC tissue, we collected 5 rhinitis fresh tissues and 6 NPC fresh tissues. Indeed, FGF2 and CD163 were significantly expressed in NPC tissues compared with rhinitis tissues. There was no change of FSP1 expression between these 2 groups ( Figure 1G). In GSE12452, analysis revealed an impeccable correlation between FGF2 and CD163 expression, suggesting an FGF-2-induced inflammation ( Figure 1H). These data show that NPC-derived FGF-2 correlates with TAMs infiltration in the TME.
FGF-2 promotes NPC metastasis. We then tested FGF-2 protein levels in various human tumor cell lines. Compared with melanoma, breast cancer, hepatocellular carcinoma, squamous cell carcinoma, and lung cancer cell lines, NPC cell lines SUNE-1 and 5-8F showed dramatically high levels of FGF-2 ( Figure 2A), validating the results from clinical NPC samples and the database analysis. To investigate the role of FGF-2 in promoting tumor growth, recruiting TAMs, and metastasis in NPC tumors, we next chose natural FGF-2 high-expressing human NPC 5-8F cells and performed FGF2-specific short hairpin RNA-knockdown (shRNA-knockdown) experiments. As expected, stable transfection of FGF2-specific shRNA effectively suppressed FGF-2 production (Supplemental  Figure 2I). Interestingly, FGF-2-expressing tumor-bearing mice showed markedly higher CTCs, tumor clones in blood culture, and pulmonary metastasis, compared with that in vector-transfected controls (Figure 2, J and K, and Supplemental Figure 2, K and L). Using gain-and loss-of-function models, we provide compelling evidence that FGF-2 contributes to promoting angiogenesis, TAMs infiltration, and tumor metastasis.
FGF-2 drives macrophage migration via pericyte-secreted factors. We next analyzed various cell types to identify FGFR expression. In both human and mouse cells, FGFR1-3 were highly expressed in both fibroblasts and pericytes ( Figure 3, A and B). Of note, FGF-2 stimulated the proliferation of pericytes and fibroblasts from both human and mouse origins (Supplemental Figure 3A). These results are in agreement with published literature (21). Only marginal expression levels were found in tumor cells, endothelial cells, and macrophages ( Figure 3, A and B). These findings show that fibroblasts and pericytes in the tumor stroma -but not tumor cells, per se -express FGFRs. This is consistent with the fact that FGF-2 knockdown in NPC cells did not affect their migratory capacity ( Figure 2, B and C). Since fibroblasts and pericytes distinctively express receptors of FGF-2 in the TME, we hypothesized that FGF-2-promoted tumor metastasis requires assistance from fibroblasts or pericytes. To test that hypothesis, we treated various cells with FGF-2 and cocultured them with tumor cells. Surprisingly, in both human NPC cells and mouse tumor cells, coculture with FGF-2-stimulated fibroblasts or pericytes did not enhance the tumor migration rate in vitro ( Figure 3, C and D), suggesting a more complex interaction among cell components. To investigate fibroblasts and pericytes highly expressing FGFR in the tumor metastasis process, we next examined the contribution of various cell types to tumor cell migration using a coculture system. We found that tumor cells cocultured with macrophages increased their migration by 2 fold, suggesting a macrophage-tumor cell interaction mechanism of metastasis ( Figure 3, E and F). Given the fact that macrophages did not express FGFRs ( Figure 3, A and B) and that FGF-2 stimulation did not alter macrophage migration (Supplemental Figure 3, B and C), we hypothesized that FGF-2 indirectly activates macrophages via fibroblasts or pericytes for tumor cell migration. We collected the conditioned medium of FGF-2-treated fibroblasts or pericytes, and we stimulated macrophages with these media. Surprisingly, significant morphological changes and a markedly increased migration rate of macrophages were observed only in the pericyte-conditioned medium group (Figure 3, G-I), suggesting that macrophages can be activated by pericyte-derived, but not fibroblast-derived, factors. Furthermore, precultured with this macrophage-activating medium, macrophages strongly drove tumor migration ( . In addition, the major receptor for CCL11, CCR3 (25), was barely expressed in macrophages but was highly expressed in basophils and eosinophils (Supplemental Figure 4A). These results suggest that CXCL14, rather than CCL11, may mediate the pericyte-specific activation of macrophages. To further validate the pericytes as the major sources of CXCL14 production in an in vivo model, we isolated host cells -including pericytes, TAMs, and endothelial cells -from the TME using NG2, F4/80, and CD31 markers (Supplemental Figure 4, B-D). We confirmed that the NG2 + cell population was the critical cell type to produce Cxcl14 in FGF-2-high tumors ( Figure  4E). In contrast, as a downstream executor of the pericyte-macrophage axis, F4/80 + macrophages did not contribute to FGF-2-induced Cxcl14 expression ( Figure 4E). Similarly, the NG2cell population, including tumor cells, produced negligible levels of Cxcl14 in both FGF-2 + and FGF-2tumors ( Figure 4E). These findings demonstrate that pericytes are the primary source of CXCL14 in the FGF-2-expressing TME.
We further analyzed the potential transcription factors regulating Cxcl14 using a prediction tool PROMO (26). Genome-wide microarray analysis of FGF-2-stimulated pericytes revealed that, in all the potential regulators of Cxcl14, aryl hydrocarbon receptor (Ahr) was the most upregulated transcription factor ( Figure 4I). Indeed, FGF-2-stimulated pericytes, rather than fibroblasts, expressed a high level of Ahr, supporting that AHR confers pericyte-specific Cxcl14 expression (Supplemental Figure 4E). The increased expression was further validated by quantitative PCR (qPCR) ( Figure 4J). Knockdown of Ahr using siRNA significantly impaired FGF-2-induced Cxcl14 expression ( Figure 4J and Supplemental Figure 4F). To provide experimental evidence for validating how AHR physically interacts with the Cxcl14 promoter, we analyzed the mouse Cxcl14 promoter region and discovered a canonical AHR-binding site at -278 bp. ChIP assay using the Cxcl14 promoter fragment containing the binding site and exon fragment demonstrated that AHR binds to the Cxcl14 promoter ( Figure 4K). These findings suggest an FGF-2/FGFR1/AHR/CXCL14 axis in pericytes ( Figure 4L).
CXCL14 promotes TAM infiltration and polarization. To investigate the functional impact of CXCL14 on macrophages, we performed in vitro experiments on macrophage migration via wound healing assay and chemotaxis assay. As expected, CXCL14 significantly recruited macrophages and promoted migration (Figure 5, A and B). It is known that, in the TME, TAMs with M2 phenotype are associated with tumor growth and invasion (27). Although FGF-2 is correlated with TAM infiltration in NPCs (Figure 1, F-H), the phenotypic characteristics of TAMs have not been identified. We extracted an equal amount of scrambled controland FGF2 shRNA-transfected 5-8F tumor tissues and performed FACS analysis of various immune cells (Supplemental Figure 5A). The number of CD45 + cells was significantly decreased by two-thirds in shFGF2 NPC tumors ( Figure 5C), validating the inflammatory effect of FGF-2 in the TME. Interestingly, the proportion of F4/80 + cells in the composition of the various types of immune cells did not change ( Figure 5D), suggesting that the inflammatory effect of FGF-2 is not limited in macrophages. Notably, MHCII + cells occupied a greater proportion in the FGF2 shRNA-transfected 5-8F TME compared with that in the control group ( Figure 5D), suggesting that FGF-2 affects DCs infiltration (Supplemental Figure 5B). Further FACS analysis showed a dramatic reduction of total TAMs ( Figure 5E). Surprisingly, CD206 + M2-like TAMs, but not CD86 + M1-like TAMs, were reduced in the FGF-2 knockdown group ( Figure 5F), supporting the role of FGF-2 on TAMs polarization toward M2 phenotype. To validate these FACS results, we isolated F4/80 + cells using MACS and then detected CD206 and CD86 RNA expression. Knockdown of FGF-2 in the TME dramatically reduced CD206 expression in TAMs ( Figure 5G). IHC staining showed a similar CD206 reduction in NPC xenograft tumor tissues ( Figure 5H). These findings demonstrate that FGF-2 promotes TAMs recruitment and polarization toward the M2 phenotype.
We next analyzed the role of CXCL14 on macrophage polarization. Indeed, direct stimulation of macrophages with CXCL14 significantly increased CD206 expression and decreased CD86 expression, while FGF-2 did not alter the expression of these markers ( Figure 5, I and J). In contrast, using the conditioned medium of FGF-2-stimulated pericytes, we could reproduce CXCL14-induced macrophage polarization ( Figure 5, I and J). We further verified the polarization effect of CXCL14 at the protein level ( Figure  5K). These results indicate that pericyte-derived CXCL14 is the mediator of FGF-2-induced macrophage recruitment and polarization.
Selective depletion of pericytes prevents CXCL14 expression and TAM infiltration. To study the inflammatory impact of pericytes on TAMs in vivo, we applied an NG2-thymidine kinase (NG2-TK) mouse model in which chondroitin sulfate proteoglycan 4 (Cspg4, Ng2) gene promoter controls the expression of herpes simplex 1 virus TK in BALB/c mice. This strain allows ganciclovir-inducible ablation of NG2 + pericytes (28). Due to the BALB/c background of this strain, we constructed another tumor cell model that is compatible with BALB/c background, mouse breast cancer 4T1, and stably transfected with human FGF-2 or empty vectors. This tumor cell line pair recapitulate the results from human 5-8F and mouse T241 pairs in vitro ( Figure 6, A-C). Interestingly, similar to previous results, 4T1-FGF-2 tumors in WT mice show increased vasculatures with pericyte coverage compared with the 4T1-vector control group ( Figure 6D). In NG2-TK mice, after ganciclovir treatment, tumoral pericytes were completely ablated ( Figure 6D). Of note, vasculature density was significantly reduced without altering the vessel diameter ( Figure 6D), probably due to the high interstitial pressure in the TME. To detect the CXCL14 production levels in tumors, we collected the total RNA of tumor tissue and performed qPCR analysis. Cxcl14 was significantly expressed in FGF-2-expressing tumors and was down to an undetectable level after pericytes ablation ( Figure 6E). These results provide compelling evidence that FGF-2 specifically promotes CXCL14 expression in NG2 + pericytes and that NG2 + pericytes are the critical source of CXCL14 in vivo.
We further investigated the role of pericyte and the role of CXCL14 in TAM infiltration. As expected, in vector tumors, TAMs were observed, and removal of pericytes further reduced them to trace amounts ( Figure 6F). Compared with vector tumors, overexpression of FGF-2 significantly promoted TAMs infiltration, while depletion of pericytes completely blocked TAMs ( Figure 6F). These results confirmed that pericytes in the FGF-2 TME are critical for TAM infiltration. To further investigate the role of CXCL14 on TAM infiltration in vivo, we injected CXCL14 protein intratumorally. Interestingly, CXCL14 administration significantly increased TAMs in pericyte-depleted FGF-2 tumors ( Figure 6F). These CXCL14-recruited TAMs were further isolated and identified as CD206 + M2 phenotype ( Figure  6, G and H). Given that pericytes have been shown to be the sole CXCL14 source in the FGF-2 TME ( Figure 4E and Figure 6E), these results suggest that pericyte-derived CXCL14 promotes the recruitment and M2-polarization of TAMs in vivo. Next, we explored metastatic activities in this pericyte depletion model. Interestingly, although ganciclovir ablated Cxcl14 expression and TAM infiltration, tumor metastasis was increased (Supplemental Figure 6, A-C). One would reasonably speculate that decrease of TAM may reduce tumor metastasis. However, in addition to immune regulation, pericytes also play an important role in vascular coating. Various studies from other groups and our group have shown that pericyte ablation increases vascular leakage and, hence, tumor metastasis (29). These results suggest that a simple deletion of pericytes to treat tumor metastasis is not ideal. Instead, targeting the later steps of the FGF-2/pericyte/ CXCL14/TAMs axis might be a better approach.
TAM-dependent metastasis of high FGF-2 tumors. We next investigated the impact of TAMs in promoting metastasis using a pharmacological approach. To define the causational relation between TAMs and NPC metastasis, NPC xenograft tumor-bearing mice were treated with clodronate liposomes to deplete TAMs. Expectedly, clodronate treatment ablated the total number of TAMs in 5-8F tumor tissues ( Figure 7A). A significantly lower number of CD206 + TAMs was found in clodronate treated 5-8F tumor-bearing mice ( Figure 7A). In FGF2-shRNA transfected group, clodronate further reduced the low level of macrophage infiltration ( Figure 7A). Importantly, CTCs, tumor clones from blood culture, and pulmonary metastases were markedly inhibited in clodronate-treated 5-8F tumor-bearing mice (Figure 7, B and C, and Supplemental Figure 7, A and B), supporting TAM's critical role in NPC pulmonary metastasis. To generalize these findings in FGF-2 expressing tumors, we treated FGF-2overexpressing and control tumor-bearing mice with clodronate. Similarly, the control liposome did not significantly affect TAM infiltration, and the vector tumor group had a significantly lower number of TAMs in TME (Supplemental Figure 7C). FGF-2 tumor-bearing mice showed elevated CTC levels and increased tumor clones from blood culture, and approximately 80% of them developed pulmonary metastasis (Supplemental Figure 7, D-F, and Figure 7D). Vector tumor-bearing mice had lower levels of CTC and pulmonary metastasis (Supplemental Figure 7, D-F, and Figure 7D). The depletion of TAMs markedly decreased the metastasis rate of both vector and FGF-2 tumor-bearing mice (Supplemental Figure 7, D-F, and Figure 7D). These findings show that FGF-2-promoted pulmonary metastasis through a TAM-dependent mechanism.

Discussion
Despite the increased need to understand the role of the TME in promoting tumor invasion and metastasis, key questions regarding how crosstalk between nontumor cell components contribute to tumor metastasis require further investigation. Particularly, vascular pericytes located between blood vessels and tumor cells may easily communicate with various cell types and become initiators of the metastatic cascade. The role of pericyte in the TME is diverse. As a major perivascular cell type in tumor microvessels, pericyte participates in angiogenesis and increases vessel maturation and stability, which support transportation of nutrients and oxygen for tumor growth. Additionally, pericyte coverage of microvessels impedes tumor cell intravasation. Recently, the inflammatory role of pericytes in the TME receives increasing attention. For example, pericyte-derived IL-33 promotes TAM infiltration (27). In malignant glioma, immature pericytes possess T cell inhibitory capability via expressing multiple immunosuppressive mediators (30). In primary CNS lymphoma, pericyte-derived CXCL9 and CXCL12 increase tumor-infiltrating lymphocytes, including CD8 + T cells (31). It seems that the inflammatory role of pericytes is context dependent. Moreover, pericytes may regulate the immune microenvironment through indirect mechanisms. For example, targeting pericytes induces leaky, dysfunctional microvessels, indirectly increasing hypoxia and resulting in myeloid-derived suppressor cell infiltration (32). Considering that pericytes' phenotype varies with tumor types, and that pericyte closely communicates with other TME components, it is not difficult to understand that its inflammatory role is complex and somehow paradoxical. Indeed, although targeting pericytes has been proposed as a potential therapeutic option for treating solid tumors alone or together with antiangiogenic drugs (33)(34)(35), clinical trials blocking pericytes have failed to improve patients' outcomes (36). We believe that in-depth studies of molecular mechanisms of pericyte-derived signaling molecules in the modulation of the TME will support us for understanding the complex role of tumoral pericytes and allow us to discover new therapeutic options.
In the current work, we have taken an unbiased approach to define pericyte-derived inflammatory signaling molecules upon FGF-2 challenge. We found that CXCL14 is highly upregulated and is a potent facilitator for TAM recruitment and polarization. Interestingly, although FGFRs are expressed at the same levels in both fibroblasts and pericytes, FGF-2-induced CXCL14 is exclusively expressed in pericytes, indicating that CXCL14 is one of the pericyte-specific inflammatory mediators. As a nonglutamic acid-leucine-arginine chemokine, CXCL14 has broad biological activities. It primarily contributes to the regulation of immune cell migration and also executes antimicrobial immunity. CXCL14 receptor remains an enigma, although recent research suggested that ACKR2 is required for CXCL14 signaling (24). It is reported that different CC or CXC chemokines can form heterodimers, and cross-family CC or CXC heterodimers have been reported (37,38). That might explain the difficulties of CXCL14 receptor identification. Our work shows that CXCL14 promotes TAM recruitment and polarization, promoting tumor metastasis. These results are in line with clinical studies and the current knowledge of this chemokine (39). Notably, besides its inflammatory role, CXCL14 in the TME may directly stimulate malignant cells and contribute to EMT and tumor metastasis (24,40). This role requires further validation in FGF-2-expressing tumors.
One of the striking findings is that, using a cross-data set approach, we identified NPC for expressing high levels of FGF-2. To our knowledge, this is the first time that NPC has been identified as a natural FGF-2-expressing human tumor type. The progress of molecular-targeted therapies in NPC significantly falls behind than that in other types of tumor, and several trials using bevacizumab or cetuximab in NPC treatment have failed to provide better clinical benefits than conventional therapies (5,6). This work might provide mechanistic insights for the limited clinical outcomes of anti-VEGF and anti-EGFR treatments in NPC. It also offers potentially novel targets such as the FGF-2/CXCL14 axis for treating NPC. Of note, in addition to FGF-2 signaling, NPC may also affect pericytes through other angiogenic factors, such as angiopoietins. Whether these factors act together with FGF-2 on pericytes and metastasis has not been explored in depth and needs to be further investigated.
JCI Insight 2022;7(10):e157874 https://doi.org/10.1172/jci.insight.157874 targeting TAMs by reducing or reprogramming them has shown promising activity in some clinical trials (43)(44)(45). However, in the NPC field, immunotherapy is mainly focused on T lymphocytes. Other types of immune cells, such as macrophages, have not been explored. Interestingly, a cohort of 108 NPC patients shows that the expression of macrophage inhibitory factor, a highly conserved cytokine that inhibits macrophage migration, can independently predict the survival of NPC patients (46), indicating that macrophages might be involved in NPC progression or metastasis. In our work, macrophage-depleting reagent clodronate liposomes significantly reduce NPC metastasis. Notably, due to the lack of spontaneous mouse NPC models or mouse NPC cell lines, we exploited other types of mouse tumor for FGF-2 overexpression experiments and genetically modified mouse model experiments. These experiments cannot fully recapitulate the characteristics of human NPC -rather, they illustrate the generalized mechanism of FGF-2 in various tumors. Although our present work is originated from and focused on NPC, we believe that these mechanistic principles may also apply to other solid cancers that express FGF-2. Our results suggest that targeting TAMs would be a potent antimetastasis therapy in NPC or other FGF-2-expressing tumors.
Animals. Female C57BL/6 and BALB/c-nude mice at the age between 6 and 8 weeks old were purchased from GemPharmatech, and they were maintained under a 12-hour dark/12-hour light cycle with food and water provided ad libitum. C.FVB-Tg(Cspg4-TK*)1Rkl/J (NG2-tk on BALB/c) mice were provided by Raghu Kalluri at the Metastasis Research Center, University of Texas MD Anderson Cancer Center, Houston, Texas, USA. All animals were randomly assigned to groups before experiments. The experimenter was not blind to the assignment of the groups and the evaluation of the results. No samples, animals, or data were excluded.
Human patient samples. Fresh samples were collected from patients receiving nasopharyngoscopic biopsy. Human NPC cancer cells and noncancer cells were isolated from fresh NPC tissues using a human tumor cell isolation kit (catalog 130-108-339, Miltenyi Biotec).