Tumor cell-derived osteopontin promotes tumor fibrosis indirectly via tumor-associated macrophages

Abstract

Background

High fibrosis of the tumor microenvironment (TME) not only impedes the effective infiltration of T cells but also serves as a physical barrier to inhibit the penetration of chemotherapy drugs. Triple-negative breast cancer (TNBC) is characterized by significant infiltration of tumor-associated macrophages (TAMs) and high fibrosis. However, the mechanism of high fibrosis in such tumors is still under debate.

Methods

We first investigated the correlation between tumor-derived osteopontin (OPN) and tumor fibrosis as well as TAM enrichment using a tumor model characterized by OPN genetic inactivation or overexpression. We further compared the effects of macrophage depletion on tumor fibrosis in mice bearing TNBC tumors (4T1^WT or 4T1^{Spp1 − KO}). To elucidate the mechanism by which TAMs promote tumor fibrosis, we evaluated their potential to recruit cancer-associated fibroblasts (CAFs) through in vitro migration assays and compared the production of transforming growth factor-beta 1 (TGFβ1) among different TAM subpopulations.

Results

Our study revealed that OPN secretion by tumor cells correlates positively with both tumor fibrosis and TAM enrichment. Specifically, within the enriched TAM population, Ly6C⁺CD206⁻ TAMs recruit CAFs via CCL5 secretion, while Ly6C⁻CD206^high TAMs secrete TGFβ1 to activate CAFs. Blocking the tumor cell-derived OPN can effectively prevent tumor fibrosis.

Conclusions

This study shows that tumor-derived OPN primarily drives TAM enrichment in mouse cancer model, indirectly promoting tumor fibrosis through Ly6C⁺CD206^−/low and Ly6C⁻CD206^high TAMs. Our findings have potential application in preventing tumors from excessive fibrosis and enhancing the efficacy of immunotherapy and chemotherapy.

Introduction

Tumor fibrosis is a significant factor that promotes cancer progression, invasion, and resistance to immunotherapy and chemotherapy. TME is a heterogeneous and complex ecosystem composed of both cellular and non-cellular components, including various immunosuppressive cell subgroups and metabolites. TAMs and CAFs are the main cellular constituents of the TME. TAMs promote tumor immune evasion and enhance chemoresistance and cancer progression by stimulating angiogenesis, releasing pro-inflammatory cytokines, inhibiting cytotoxic T cell responses, and remodeling the extracellular matrix (ECM) [1,2,3,4,5]. TAMs may also promote tumor progression though CAFs. It has been reported that TAMs and CAFs were co-localize in several types of tumors, with TAM infiltration positively correlated with CAF enrichment in tumor tissues [6, 7]. Studies on pancreatic ductal adenocarcinoma (PDAC) and small cell lung cancer (SCLC) models have identified specific macrophage subpopulations that promote fibrosis and exhibit immunosuppressive functions [8,9,10]. CAFs produce collagen and other ECM components, which may promote tumor progression by maintaining tumor cell stemness and reshaping metabolism [11]. The high-density, orderly arranged collagen fibers at the tumor edge suppress antitumor immune responses by constraining CD8 T cell infiltration [10, 12, 13], with highly fibrotic tumors are generally characterized by immune “cold” [14,15,16]. Fibrous matrices in solid tumors also create a physical barrier that hinders the penetration of chemotherapeutic drugs, posing significant challenges to tumor chemotherapy [17].

Previous studies have shown a positive correlation between OPN and fibrosis in both human and animal tumor models [18, 19]. OPN, encoded by the secreted phosphoprotein 1 (Spp1) gene, is a multifunctional phosphoglycoprotein produced by various cell types, including lymphohematopoietic cells (such as T cells, B cells, NK cells, NKT cells, and myeloid cells), osteoblasts, osteocytes, epithelial cells, and neurons [20]. OPN is overexpressed in multiple human tumors, such as lung cancer, prostate cancer, breast cancer, colorectal cancer, and liver cancer, closely related to disease progression and poor prognosis [20, 21]. OPN facilitates tumor fibrosis through its interaction with CD44 and integrin αVβ3 on fibroblast surfaces, leading to the activation of the Twist1 gene and upregulation of integrin β1, α-smooth muscle actin (αSMA), and fibroblast activation protein (FAP) [22]. Additionally, OPN can bind to integrin αVβ3 on mesenchymal stem cells (MSCs), activating the MZF1/TGFβ1 signaling pathway and increasing the expression levels of fibrosis-related genes such as αSMA, Tenascin-C, vimentin, and FSP1, leading to the phenotypic transformation of MSCs into CAFs [23]. These studies highlight OPN’s role in promoting tumor fibrosis through direct activation of CAFs without the involvement of TAMs. It has also been reported that OPN can chemotactic macrophages and maintain their M2-type polarization [24, 25], and TAMs can promote tumor fibrosis [10]. Given the significant enrichment of TAMs in TNBC [26], we propose that in this highly fibrotic tumor, OPN promotes fibrosis likely more significantly through an indirect mechanism mediated by TAMs, rather than only directly acting on CAFs. Additionally, further investigation is required to clarify the source of OPN in this process.

In this study, we systematically investigated the role and underlying mechanisms of OPN in tumor fibrosis using a variety of mouse tumor models that differ in OPN production and fibrotic characteristics. Our findings revealed that the degree of fibrosis is positively correlated with the ability of tumor cells, but not recipient cells, to produce OPN. Furthermore, the role of OPN in promoting tumor fibrosis is mediated by TAMs. Mechanistically, OPN produced by tumor cells recruits and activates CAFs by enriching Ly6C⁺CD206⁻ TAMs to produce CCL5 and Ly6C⁻CD206^high TAMs to produce TGFβ1, respectively. Our results suggest that inhibiting OPN secretion by tumor cells holds significant potential for restricting tumor fibrosis and enhancing the efficacy of tumor immunotherapy and chemotherapy.

Results

Tumor cell-derived OPN aggravates tumor fibrosis and TAM accumulation

We first investigated the sources and role of OPN in tumor fibrosis using a mouse model of TNBC. This model was created by injecting a highly fibrotic TNBC cell line (4T1), known to produce high levels of OPN [26], into the mammary gland. OPN-deficient tumor cells were generated by genetic inactivation of Spp1 using the CRISPR/Cas9 system in 4T1 cells (Figure S1A and S1B). Although OPN deficiency did not significantly affect tumor cell viability or apoptosis in vitro (Figure S1C and S1D), 4T1^{Spp1 − KO} tumors exhibited a marked reduction in αSMA⁺ cells and collagen accumulation (determined by Sirus red staining) compared to 4T1^WT tumors (Fig. 1A). The reduction of αSMA⁺ cells and collagen accumulation in 4T1^{Spp1 − KO} tumors was associated with significant inhibition of tumor growth (Figure S2) and decreased infiltration of CD11b⁺F4/80⁺ TAMs (Fig. 1B). We also measured CD11b⁺F4/80⁺Ly6C⁺ and CD206^high M2-type TAMs, which have been reported to associated with hepatic [27, 28], pulmonary [27, 29], cardiac [30, 31] and pancreatic [8] fibrosis. Although there was no significant difference in proportion, 4T1^{Spp1 − KO} tumors had significantly fewer CD11b⁺F4/80⁺Ly6C⁺ TAMs compared to the 4T1^WT tumors (Fig. 1C). Furthermore, TAMs in 4T1^{Spp1 − KO} tumors contained significantly fewer CD206^high M2-type TAMs than those in the 4T1^WT tumors (Fig. 1D). No significant difference was detected in the proportions of CD11b⁺Ly6C⁺ monocytes and CD11b⁺Ly6G⁺ neutrophils in peripheral blood between 4T1^{Spp1 − KO} and 4T1^WT tumor-bearing mice (Fig. 1E). These results indicate that OPN produced by tumor cells play an important role in the formation of fibrotic stroma in the 4T1 mouse TNBC tumor.

Fig. 1

Tumor cell-derived OPN positively correlated with the degree of tumor stromal fibrosis. (A) Representative immunohistochemical (IHC) staining images (left, scale bars = 20 μm) and the quantification (right) of αSMA and Sirus red-stained images in 4T1^WT and 4T1^{Spp1 − KO} tumor tissues of Balb/c mice (n = 3). Original magnification, ×500. (B) Representative flow cytometric plots (top) and frequency and number (bottom) of CD11b⁺F4/80⁺ macrophages in tumors (n = 5). (C) Representative flow cytometric plots (left) and frequency and number (right) of F4/80⁺Ly6C⁺ cells in tumors (n = 5). (D) Representative flow cytometric plots (left) and frequency (right) of F4/80⁺CD206^high M2 type macrophages in tumors (n = 5). (E) Representative flow cytometric plots (left) and frequency (right) of CD11b⁺Ly6C⁺ and CD11b⁺Ly6G⁺ cells in peripheral blood (n = 5). (F) Mice were inoculated respectively with 4T1, CT26, RM1 and B16-F0. All tumor samples were collected at the point of tumor volume reached 800 mm [3] and fixed for IHC staining. The staining results of αSMA and Sirus red were quantificated and presented as integral optical density (IOD)/Area (n = 3). Original magnification, ×500 (scale bars: 20 μm). (G) Based on transcriptome sequencing data of four tumor cell lines, gene set enrichment analysis was used to screen the potential genes positively associated with tumor fibrosis. (H) Relative mRNA levels of Spp1 in 4T1, CT26, RM1 and B16 cells measured by RT-qPCR (n = 3 biological replicates). Data are presented as means ± SEM, or representative staining profiles or photographs. p values were determined by unpaired two-tailed Student’s t test (A-E), or one-way ANOVA (F, H). Shown are data from a representative of three independent experiments (A–E, H). *p < 0.05, **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, no significance

We next assessed whether transgenic expression of OPN in a mouse melanoma cell line (B16-F0), which has low endogenous OPN production [26], promotes tumor fibrosis. B16 melanoma cells with transgenic overexpression of OPN (B16^Spp1 were generated via lentiviral transduction and validated for stable OPN overexpression at both mRNA and protein levels compared to vector-transduced cells (B16^NC) (Figure S3A and S3B). Consistent with observations in 4T1 tumors, no significant differences in viability or apoptosis were observed between B16^Spp1 and B16^NC cells (Figure S3C and S3D). However, B16^Spp1 tumors exhibited significant increases in αSMA⁺ cells and collagen accumulation compared to B16^NC tumors (Figure S4A). Additionally, flow cytometry analysis revealed a significant increase in the proportion of CD11b⁺F4/80⁺ TAMs and CD206^high M2 TAMs in B16^Spp1 tumors compared to B16^NC tumors (Figure S4B, S4C).

In line with the observations made through genetic inactivation or overexpression of OPN, tumor fibrosis was found to be positively correlated with OPN production levels by tumor cells across various tumor types. Among the four tumor models examined, tumor fibrosis, indicated by the amounts of αSMA⁺ cells and collagen (identified by Sirus Red staining), was most severe in 4T1 tumors, followed by CT26 (colon cancer), RM1 (prostate cancer), and least severe in B16-F0 tumors (Fig. 1F). Transcriptome sequencing (RNA-seq) analysis of these 4 tumor cell lines revealed that, among the genes associated with tumor fibrosis, including integrin alpha 1 (Itga1), Itga8, Itgb4, platelet-derived growth factor beta polypeptide (Pdgfb), suppressor of cytokine signaling 2 (Socs2), interleukin 33 (Il33), lysyl oxidase like 1 (Loxl1), C-X-C motif chemokine receptor 4 (Cxcr4), Spp1, signal transducer and activator of transcription 1 (Stat1), and G protein subunit beta 4 (Gnb4), the expression level of Spp1 was found best correlated with the degree of fibrosis (Fig. 1G). The highest OPN expression was observed in 4T1 cells, followed by CT26 and RM1, with almost undetectable levels in B16-F0 cells (Fig. 1G and H). Collectively, these results indicate that OPN produced by tumor cells, rather than recipient cells, plays a critical role in driving tumor fibrotic stroma formation. This finding aligns with previous reports that OPN produced by TAMs is primarily the non-secreted isoform [26, 32, 33].

OPN promotes tumor fibrosis through mechanisms highly dependent on TAMs

TAMs have been reported to contribute to tumor fibrosis [10]. Given that OPN-induced tumor fibrosis was accompanied by increased TAM accumulation (Fig. 1 and S4), we sought to determine whether TAMs are involved in OPN-induced tumor fibrosis. To this end, we compared the effects of macrophage depletion on tumor fibrosis in mice bearing 4T1^WT or 4T1^{Spp1 − KO} tumors. Macrophage depletion in tumor-bearing mice was achieved by intravenous injection of clodronate-liposomes (CL) as previously reported [26], and the efficacy of macrophage depletion, including CD206⁺ M2 TAMs, was confirmed by flow cytometry and IHC analysis of spleen and tumor tissues (Figure S5A-C). Immunohistochemical staining revealed that, compared to 4T1^{Spp1 − KO} tumors, 4T1^WT tumors from PBS-treated mice exhibited significantly higher expression of fibrotic proteins such as FSP1, αSMA, PDPN, and collagen, which were markedly reduced, to the levels comparable to the 4T1^{Spp1 − KO} tumors, in 4T1^WT tumors from CL-treated mice (Fig. 2A). However, macrophage depletion with CL had minimal or no effect on the expression of these fibrotic proteins in the 4T1^{Spp1 − KO} tumors (Fig. 2A). Furthermore, dual staining for CD90.2 and CD68 showed a clear positive correlation and spatial association between CD90.2⁺ fibroblasts and CD68⁺ TAMs, and both of which were significantly reduced in the 4T1^WT tumors to the levels comparable to the 4T1^{Spp1 − KO} tumors following macrophage depletion (Fig. 2B). Consistent with the immunohistochemical staining results, in PBS-injected controls, the frequency of proliferating CD45⁻CD90.2⁺Ki67⁺ CAFs was significantly higher in 4T1^WT tumors compared to 4T1^{Spp1 − KO} tumors. Macrophage depletion with CL resulted in a marked decrease in the frequency of CD45⁻CD90.2⁺Ki67⁺ CAFs in 4T1^WT tumors to levels comparable to those in 4T1^{Spp1 − KO} tumors but had no detectable effect in 4T1^{Spp1 − KO} tumors (Fig. 2C). The correlation analysis revealed a significant positive correlation between the proportion of TAMs and that of CAFs (Fig. 2D). As expected, the degree of tumor fibrosis was positively correlated with tumor progression, as shown by increased tumor growth and lung metastasis, and reduced infiltration by CD4⁺ and CD8⁺ T cells (Figure S6A-C). These findings suggest that the tumor-derived OPN promotes tumor fibrosis primarily through its effects on TAMs.

Fig. 2

Tumor cell-derived OPN increase tumor stromal fibrosis through mechanisms highly dependent on TAMs. (A-C) Balb/c mice were inoculated 4T1^WT or 4T1^{Spp1 − KO} cells and treated as indicated in Fig S5. Tumors were excised on the 24th day and subsequently analyzed using immunofluorescence (IF), IHC staining, and flow cytometry. (A) Representative images (left, scale bars = 20 μm) and the quantification (right) of fibrosis related IHC staining for FSP1, PDPN, αSMA and Sirus Red in tumor tissues (n = 5). Original magnification, ×500. (B) IF staining multiplex imaging (left, scale bars = 50 μm) and the quantification (right) of CD90.2 in tumor tissues (n = 5), visualization of CD90.2 (red), CD68 (green) and cell nuclei (blue). Original magnification, ×200. (C) Representative flow cytometric plots (left) and frequency (right) of CD90.2⁺Ki67⁺ CAFs within the CD45⁻ population in tumors (n = 5). (D) Percentage of TAMs and CAFs in tumor tissues from 4T1^WT or 4T1^{Spp1 − KO} -bearing mice with different treatments (left) and Pearson correlation analysis between TAMs and CAFs (right). Data are presented as means ± SEM, or representative staining profiles or photographs. p values were determined by one-way ANOVA (A-C), Spearman’s test (D). Shown are data from a representative of three independent experiments (A–C). *p < 0.05, **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, no significance

TAMs recruit and activate CAFs in tumors by secreting CCL5 and TGFβ1, respectively

To elucidate the mechanisms by which TAMs promote tumor fibrosis, we first investigated the potential of TAMs to recruit CAFs using an in vitro migration assay. CD45⁻CD90.2⁺ CAFs isolated from tumor tissues were placed in the upper chamber of a transwell system, while CD45⁺CD11b⁺F4/80⁺ TAMs were positioned in the lower chamber. After a 24-hour co-incubation, the migration of CAFs was assessed. The number of CAFs migrating through the membrane was significantly higher in cultures with TAMs compared to the non-TAM controls (culture medium, CM) (Fig. 3A). We then assessed the ability of TAMs to produce CCL5, a chemokine known to recruit CAFs [34]. RT-qPCR analysis revealed that CCL5 expression was significantly higher in TAMs (CD45⁺CD11b⁺F4/80⁺), particularly in the Ly6C⁺CD206^−/low TAM subset, than in 4T1 tumor cells (Fig. 3B and C). Additionally, blocking the CCL5/CCR5 axis with a CCR5 receptor antagonist, maraviroc, significantly reduced the ability of TAMs to chemoattract CAFs (Fig. 3D). These findings suggest that TAMs may promote tumor fibrosis by recruiting CAFs through the secretion of CCL5.

Fig. 3

TAMs of Ly6C⁻CD206^high and Ly6C⁺CD206^−/low synergistically spur tumor fibrosis. (A) Transwell migration assay of CAFs, in which CAFs sorted from tumors were added to upper chambers, and the lower chambers contained serum-free medium (Culture Medium, CM) with or without TAMs. Shown are representative membrane staining (left, scale bars = 20 μm), and numbers of CAFs (right) migrating to the lower compartment (n = 3 biological replicates). Original magnification, ×500. (B) Relative mRNA levels of CCL5 in 4T1^WT cells and CD11b⁺F4/80⁺ macrophages which were sorted from tumors measured by RT-qPCR (n = 3 biological replicates). (C) Relative mRNA levels of CCL5 in four subtypes of macrophages sorted from tumors measured by RT-qPCR (n = 6 biological replicates). (D) Transwell migration assay of CAFs, in which CAFs sorted from tumors were pretreated with or without CCR5 antagonist (Maraviroc, 100 nm/ml) and added to upper chambers, and TAMs were placed in the lower chambers. Shown are representative membrane staining (left, scale bars = 20 μm), and numbers of CAFs (right) migrating to the lower compartment (n = 3 biological replicates). Original magnification, ×500. (E) Relative mRNA levels of TGFβ1 in four subtypes of TAMs measured by RT-qPCR (n = 6 biological replicates). (F) The levels of TGFβ1 in cultured medium derived from Ly6CCD206^−/low or Ly6CCD206^high TAMs detected by ELISA (n = 3 biological replicates). (G) Representative flow cytometric plots (upper), frequency and number (bottom) of Ly6C⁺CD206^−/low and Ly6C⁻CD206^high cells in tumors (n = 5). (H) Relative mRNA levels of αSMA, FSP1, FAP, ITGβ1 and PDPN in CAFs pretreated with TGFβ1 (10 ng/ml) or PBS for 72 h measured by RT-qPCR (n = 3 biological replicates). (I) Representative αSMA, FSP1 and PDPN western blots using lysates of CAFs pretreated with TGFβ1 (10 ng/ml) or PBS for 72 h. (J) Relative mRNA levels of αSMA, FSP1 and PDPN in CAFs co-cultured with TAMs treated with anti-TGFβ1 antibody or isotype control antibody for 72 h (n = 3 biological replicates). (K) Representative αSMA, FSP1 and PDPN western blots using lysates of co-cultured CAFs from (J). Data are presented as means ± SEM, or representative staining profiles or photographs. p values were determined by one-way ANOVA (C, E), unpaired two-tailed Student’s t test (A, B, D, F), or two-way ANOVA (G, H, J). Shown are data from a representative of three independent experiments (A–K). *p < 0.05, **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, no significance

TGFβ1 is a key pro-fibrotic factor produced by macrophages, but the TAM subsets that promote fibrosis via TGFβ1 vary across different disease settings [8, 35, 36]. Thus, we compared different TAM subsets for TGFβ1 production and found that among the TAM subsets, including Ly6C⁻CD206^−/low, Ly6C⁻CD206^high, Ly6C⁺CD206^−/low, and Ly6C⁺CD206^high, the expression of TGFβ1 in the Ly6C⁻CD206^high subset was significantly higher than in the other subsets, followed by the Ly6C⁺CD206^−/low subset (Fig. 3E). Ly6C⁻CD206^high TAMs also produced significantly more TGFβ1 protein than Ly6C⁻CD206^−/low TAMs when cultured in vitro (Fig. 3F). These results indicate that Ly6C⁻CD206^high and Ly6C⁺CD206^−/low macrophage populations have the potential to synergistically promote tumor fibrosis. Consistent with prior findings, OPN deficiency in tumor cells reduces tumor fibrosis (Figs. 1A and 2A and 2B), and significantly decreases the number and proportion of Ly6C⁻CD206 ^high macrophages, as well as the number of Ly6C⁺CD206^−/low macrophages, though the proportion of the latter remains unchanged (Fig. 3G). Furthermore, incubation of CAFs with recombinant TGFβ1 resulted in a significant increase in the expression of fibrotic markers, including αSMA, FSP1, and PDPN, compared to controls (Fig. 3H and I). Although macrophages can promote the activation of fibroblasts by secreting a variety of cytokines [37,38,39,40,41], among the cytokines secreted by OPN-enriched TAMs, the activation of CAFs is primarily dependent on TGFβ1, as blocking TGFβ1 action in the co-culture system of TAMs and CAFs inhibits the transcription and protein expression of fibrosis-related genes, such as αSMA, Fsp1, and Pdpn, in TAM-induced CAFs mostly (Fig. 3J and K). These findings suggest that TAMs recruit and activate CAFs through the secretion of CCL5 and TGFβ1, respectively, in 4T1 tumors, thereby promoting tumor fibrosis. Additionally, CCL5 and TGFβ1 are predominantly produced by distinct TAM subpopulations, all of which can be enriched by OPN derived from tumor cells. Thus, blocking or neutralizing the activity of tumor cell-derived OPN or its interaction with related TAM subpopulations may alleviate tumor fibrosis. Consistent with findings in mice, database analysis revealed that OPN expression in TNBC tissues is significantly higher than in adjacent normal tissues, and OPN mRNA levels are positively correlated with the mRNA levels of fibrosis-related genes, such as αSMA, Fsp1, and Pdpn (Figure S7A, S7B). This suggests that targeting OPN and its interaction with related TAM subpopulations holds therapeutic potential for human TNBC fibrosis.

Discussion

In this study, using 4T1^WT and 4T1^Spp1-KO cell lines along with corresponding mouse models, we demonstrated that tumor-derived OPN predominantly drives the enrichment of TAMs in TNBC. Among them, Ly6C⁺CD206^−/low TAMs secrete CCL5 to chemotaxis CAFS, while Ly6C⁻CD206^high TAMs secrete TGFβ1 to activate CAFs. Therefore, OPN indirectly promotes TNBC fibrosis through the cooperation of these two TAM subsets. Our findings offer a mechanistic insight into the regulation of TNBC fibrosis by OPN, which holds significant implications for mitigating tumor fibrosis and enhancing the efficacy of tumor immunotherapy and chemotherapy.

TAMs and CAFs are key components of the TME, influencing immune responses [42, 43]. The highly fibrotic microenvironment restricts anti-tumor T cell infiltration and chemotherapy drug penetration. Although some drivers of CAF activation have been identified, further research is required to elucidate upstream factors affecting CAF chemotaxis and activation. Co-localization of TAMs and CAFs has been observed in several tumor types, implying that their interactions are critical for tumor progression [6, 7, 44]. For instance, studies on neuroblastoma show that high macrophage and fibroblast co-localization correlates with bone marrow metastasis, high Ki67 index, MYCN amplification, low tumor differentiation, and high-risk classification [44]. Spatial transcriptomics of colon cancer revealed higher enrichment of FAP⁺ CAFs and SPP1⁺ TAMs in tumor tissues compared to normal tissues, showing significant co-localization. Notably, SPP1⁺ TAMs express CD206, and both cell types show active signaling pathways related to collagen fiber formation, ECM remodeling, and TGFβ response [7].

Fibroblasts play a crucial role in regulating macrophage migration and function. Contracted fibroblasts generate deformation fields within the fibrous collagen matrix to initiate and direct macrophage migration [45]. Studies on melanoma have shown that collagen expressed in endothelial and stromal cells promotes Ly6C⁻ macrophage survival, differentiation, and proliferation through the interaction of with leukocyte-associated immunoglobulin-like receptor 1 (LAIR1), a collagen receptor expressed on monocytes and macrophages [46]. Information about macrophages promoting fibrosis mostly comes from research on tissue repair processes. Inflammatory monocytes and macrophages are key regulators of tissue fibrosis, playing a crucial role in the occurrence, maintenance, and regression of tissue damage. Fibroblasts are the main cellular source of pathological ECM deposition during fibrosis [47, 48]. Pro-fibrotic macrophages coordinate scar formation by interacting with fibroblasts [49]. Although macrophages can promote fibrosis, they also serve as “primary regulators” of myofibroblast function and ECM degradation. Therefore, it is essential to elucidate the roles of different macrophage subpopulations in fibrosis. In damage repair models, sensory nerves regulate the survival and IL-10 production of TIM4⁺ macrophages by secreting TAFA4, promoting the anti-inflammatory function of macrophages and preventing fibrosis after tissue injury [50]. Tissue resident macrophages (TRMs) in PDAC models exhibit phenotypes associated with ECM remodeling, maintaining tissue homeostasis and driving fibrosis. Eliminating TRMs can improve survival rates of tumor-bearing mice and reduce tumor invasiveness [8]. A subpopulation of macrophages that promotes fibrotic and immunosuppressive functions has been identified in SCLC [51]. However, the specific subpopulation of macrophages that primarily interacts with fibroblasts and promotes fibrosis, as well as the factors influencing fibrosis-related macrophage enrichment in tumors, remain controversial.

OPN has been previously reported to play a critical role in the fibrosis process by directly acting on fibroblasts across various fibrotic disease models and tumors. Specifically, OPN can enhance the migration and proliferation of human lung fibroblasts, upregulate the expression of tissue inhibitor of matrix metalloproteinases (TIMPs) and type I collagen in fibroblasts, downregulate matrix metalloproteinase 1 (MMP-1) expression, and promote an extracellular matrix deposition microenvironment [52]. In the myelofibrosis model, inhibition of monocyte-derived OPN secretion via ERK1/2 signaling or administration of an OPN monoclonal antibody reduces bone marrow reticular network protein deposition and attenuates splenic fibrosis development in mice. However, the precise mechanisms underlying OPN’s promotion of myelofibrosis remain to be elucidated [53]. In a metabolically dysfunction-associated steatohepatitis (MASH) model, CEBPA alleviates hepatic fibrosis by negatively regulating OPN expression in hepatocytes. Kupffer cells appear not to significantly contribute to this process, as their depletion does not alleviate the fibrosis exacerbation induced by CEBPA deficiency [54]. OPN is overexpressed in various human tumors and is highly correlated with tumor progression and metastasis [20, 21]. In the TME, OPN is produced by tumor and stromal cells [20]. Some functions of OPN have been elucidated. However, the contribution of OPN to tumor fibrosis is still partially recognized. Previous studies have shown that tumor cell-derived OPN binding to CD44 or integrin αVβ3 can reprogram normal fibroblasts into FAP⁺/FSP1⁺/αSMA⁺ cells or transform MSCs into αSMA⁺/FSP1⁺/Vimentin⁺ CAFs [22, 23, 55]. However, the impact and mechanisms of OPN on fibrosis in highly fibrotic tumors, such as the 4T1 model, remain unclear. This study found that tumor-derived OPN positively correlates with tumor fibrosis and TAM enrichment. Specifically, OPN expression in tumor cells determines macrophage infiltration. In mouse tumor models including 4T1, B16, RM1, and CT26, 4T1 cells secrete the most OPN, resulting in the most significant macrophage infiltration in tumors. Removing OPN expression from 4T1 cells reduced macrophage infiltration, while overexpressing OPN in B16 cells increased it. Subsequently, this project investigated the role of different TAM subpopulations in recruiting and activating CAFs. Recent studies have shown that tumor-derived CCL5 recruits CAFs in colorectal cancer, increasing the proportion of αSMA⁺ CAFs, and ultimately promoting collagen accumulation in tumor tissues and tumor progression [34, 56]. Consistent with these findings, we observed that the Ly6C⁺CD206^−/low subgroup recruits CAFs by secreting CCL5. TGFβ1 is the most significant pro-fibrotic factor produced by macrophages. TGFβ1 can polarize fibroblasts into myofibroblasts, enhancing ECM deposition and tissue remodeling [35]. For example, monocyte-derived macrophages, but not TRMs, are involved in pulmonary fibrosis development [36]. In the PDAC model, TRMs promote fibrosis and tumor progression by secreting TGFβ1, and their elimination improves survival and reduces tumor invasiveness in tumor-bearing mice [8]. This study found that Ly6C⁻CD206^high and Ly6C⁺CD206^−/low TAMs, particularly Ly6C⁻CD206^high TAMs, activate CAFs by secreting TGFβ1. Thus, Ly6C⁺CD206^−/low and Ly6C⁻CD206^high TAMs have the potential to synergistically promote hyperfibrosis in 4T1 tumors.

In conclusion, unlike previous reports that OPN promotes fibrosis by directly acting on fibroblasts in other tumors and other fibrosis model, this study reveals that in TNBC, a highly fibrotic tumor, tumor cell-derived OPN mainly promotes fibrosis through TAMs. Blocking tumor cell-derived OPN can effectively inhibit excessive tumor fibrosis and improve the efficacy of immunotherapy and chemotherapy. Anti-OPN antibodies will have a direct anti-fibrotic effect, but our recently published literature has confirmed that blood-derived TAMs (Ly6C⁺) have anti-tumor effects, while resident TAMs (Ly6C⁻LYVE-1⁺) have pro-tumor effects. OPN can promote the proliferation of resident TAMs by binding with LYVE-1 on resident TAMs [26]. Together with the results of this study, we speculate that anti-LYVE-1 antibodies block the interaction between OPN and LYVE-1, it will not only block the fibrosis caused by OPN more precisely but also be more conducive to immunotherapy.

Limitations of the study

Besides the effect of TGFβ1 on activating CAFs, we recognized that other factors secreted by TAMs such as PDGF [37], fibroblast growth factor12 (FGF12) [38], IL-10 [39], IL-1α/β, tumor necrosis factor-α (TNFα) [40] and MMPs (MMP9, MMP12, MMP13) [41] play an important role in activating fibroblasts. While our study demonstrated that different subpopulations of TAMs play a crucial role in OPN-induced fibrosis through the secretion of CCL5 and TGFβ1, we did not investigate if other cytokines secreted by TAMs influence this process and their importance. Furthermore, this study did not explore more effective and precise interventions in the fibrotic process associated with OPN to maximize the efficacy of tumor immunotherapy.

Materials and methods

Animals

6 to 8 weeks old female Balb/c and C57BL/6 mice were purchased from Charles River Laboratories (Beijing, China) and maintained in a specific pathogen–free environment with food and water ad libitum. All animals received care in compliance with the guidelines outlined in the Guide for the Care and Use of Laboratory Animals. All procedures were approved by the Jilin University Animal Care and Use Committee.

Cell lines

4T1, CT26, RM1 and B16 cells were purchased from ATCC and maintained in RPMI-1640 (Carlsbad, CA, USA) or DMEM (Carlsbad, CA, USA) supplemented with 10% FBS (Waltham, MA, USA), 100 U/mL penicillin and 100 U/mL streptomycin. The above cells were cultured in a humidified incubator at 37°C with 5% CO2. The plasmid pL-CRISPR.EFS.GFP (57818, Addgene, Waterton, MA, USA) was used as the DNA template for sgRNA transcription in vitro. The sequence of the Spp1-targeting sgRNA was 5’-CACCGTCTGCGGCAGGCATTCTCGG-3’ and 5’-AAACCCGAGAATGCCTGCCGCAGAC-3’. The vector was enzymatically digested by Esp3I (Thermo Fisher, USA), then the target fragment was collected and purified. Secondly, the primers were annealed to form a double-strand structure, and connected with the target fragment. Finally, 4T1 cells were stably transfected with the complex plasmid. GFP positive monoclonal cells were screened for OPN expression by PCR and western blot. Mouse Spp1 (NM_001204201.1) expression vector GV493 was obtained from GeneChem (Shanghai, China).

Cell viability and apoptosis assays

1.5 × 10³ tumor cells were plated in 96-well plates in culture medium. After 24-, 48- and 72-hours, viability was determined by Enhanced Cell Counting Kit-8 (Beyotime, China) according to the manufacturer’s instructions. 5 × 10⁵ tumor cells were plated in 6-well plates in culture medium. After 24-, 48- and 72-hours, cells were collected and stained with AnnexinV-APC/7-AAD Apoptosis Detection Kit (Biolegend, USA), apoptosis was detected by flow cytometry.

Mouse tumor and macrophage depletion models

To set up the tumor-bearing mouse model, 4T1 cells (5 × 10⁵ per mouse in 100 μL PBS) were administered by subcutaneous injection into the 4th mammary gland fat-pad of the Balb/c mice. CT26, RM1 and B16 cells (5 × 10⁵ per mouse in 100 μL PBS) were administered by subcutaneous injection into the right flank of C57BL/6 mice which had previously shaved. The tumors were measured every 3 days. To deplete macrophage, Clodronate-encapsulated liposomes (Amsterdam, The Netherlands) were administered through intravenously injection 24 h prior tumor inoculation and continuing twice weekly for a dose of 100 μL per 10 g weight. Efficient depletion was confirmed by measuring CD11b⁺F4/80⁺ cells in spleen and tumor tissues. Tumor volume was determined by measuring length (L) and width (W) and calculated as Volume = LW²/2.

Tissue harvest and dissociation

For single cell isolation from tumor tissues, the tumors were diced into 1 mm [3] pieces, collected into the tubes, which were filled with digestion solution, and dissociated tissues with gentleMACS™ Octo Dissociator (standardized program). Spleens harvested from tumor–bearing mice were triturated and incubated on ice with red blood cell (RBC) lysing buffer for 10 min to remove RBCs. Peripheral blood mononuclear cells (PBMCs) were isolated by incubating on ice with RBC lysing buffer for 15 min. All the samples were filtered through a cell strainer (100 μm) to provide a single-cell suspension and stored at 4 ℃ for downstream experiments.

Flow cytometry and cell soring

Flow cytometry was used to determine the phenotypes of mouse cells using various combinations of the following fluorochrome-conjugated mAbs: anti-mouse CD45, CD4, CD8, CD19, CD49b, CD11b, F4/80, Ly6C, Ly6G, CD206 and CD90.2 from BioLegend. Prior to flow cytometry staining, samples were incubated with anti-mouse CD16/32 (BioLegend,101320) at room temperature for 10 min to block Fc receptor. The single-cell suspension samples were stained with fluorochrome-conjugated mAb and washed twice with FACS buffer. The samples were collected on a fluorescence-activated cell sorter (Cytek Biosciences, Cytek) and analyzed by FlowJo software (TreeStar). Macrophages were primary sorting by CD11b Microbeads (Miltenyi, Germany) according to the manufacturer’s instructions. CD11b⁺ cells were collected and stained with F4/80, Ly6C and CD206 mAbs. CAFs and TAMs were co-cultured at 1:1 ratio and/or treated with 20 μg/ml TGFβ1-blocking antibody (Invitrogen, MA1-21595) or isotype control antibody for 72 h prior to RT-qPCR and WB analysis. TAMs of different phenotypes and CAFs were separated by flow cytometry sorting.

RNA isolation and RT-qPCR

RNAs were isolated using Trizol (Thermo Fisher, USA) following the manufacturer’s protocol. Quantitative real-time PCR was performed using the SYBR Premix ExTaq (TaKaRa, Tokyo, Japan), and the relative expression of each target gene was normalized by β-actin as previously described [57]. The following primers were used for each target gene for real-time PCR:

SPP1: 5’-GCAAGAAACTCTTCCAAGCAA-3’

5’-TGAGATTCGTCAGATTCATCCG-3’

CCL5: 5’-GCTGCTTTGCCTACCTCTCC-3’

5’-TCGAGTGACAAACACGACTGC-3’.

TGFβ1: 5’-CTTCAATACGTCAGACATTCGGG-3’.

5’-GTAACGCCAGGAATTGTTGCTA-3’

αSMA: 5’-GGCACCACTGAACCCTAAGG-3’.

5’-ACAATACCAGTTGTACGTCCAGA-3’

FSP1: 5’-TGAGCAACTTGGACAGCAACA-3’

5’-CTTCTTCCGGGGCTCCTTATC-3’

PDPN: 5’-GTTTTGGGGAGCGTTTGGTTC-3’.

5’-CATTAAGCCCTCCAGTAGCAC-3’

ITGβ1: 5’-ATGCCAAATCTTGCGGAGAAT-3’.

5’-TTTGCTGCGATTGGTGACATT-3’

FAP: 5’-GGCTGGGGCTAAGAATCCG-3’

5’-GCATACTCGTTCACTGGACAC-3’).

Relative expression levels for each target gene were calculated using the 2^−ΔΔCt method.

Western blotting

All samples were lysed on ice for 15 min with 100 μL RIPA lysis buffer (Invitrogen, P0013E). The samples were centrifuged for 10 min at 12,000 rpm, retaining the supernatant. The protein concentration was detected with the BCA protein assay kit (Thermo Fisher Scientific, 23227), and then boiled at 100 °C for 10 min. Equal amounts of proteins (30–60 μg) were separated by electrophoresis on 10% SDS–polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride membrane. The indicated OPN, αSMA, FSP1, PDPN, and Vinculin primary antibodies was detected and was visualized with peroxidase-conjugated secondary antibodies and enhanced chemiluminescence substrate (Pierce, Rockford, IL, USA).

Transwell assay

Tumor cells or sorted CAFs were seeded into the upper chambers with 8-μm pore inserts (Corning, USA). The lower chambers were filled with 500 μL of migration medium. After 24 h incubation, the upper chambers were removed and rinsed with PBS. The migrated cells on the basal side of the membrane were fixed with 4% paraformaldehyde at room temperature for 20 min, then stained with 0.2% crystal violet for 5–10 min. The chambers were gently washed three times with PBS to remove any excessive remaining crystal violet. Next, the basal side of membrane were examined under microscope, and the number of migrated cells was quantified by ImageJ software (Beckman Coulter).

Immunohistochemistry (IHC) and immunofluorescence (IF) staining

Lungs and tumor tissues were fixed with 4% paraformaldehyde and performed on paraffin-embedded, and sectioned for H&E (2.5 μm) and IHC (2.5 μm) staining. The markers: anti-CD68 (Abcam, ab125212), anti-FSP1 (Abcam, ab197896 ), anti-PDPN (Abcam, ab256559), anti-αSMA (Abcam, ab32575) and Sirus Red (Leagene, DC0041) were used for tumor tissues IHC staining. For IF (4 μm) staining, the tumor tissues were fixed with 4% paraformaldehyde for 72 h then transferred into 30% sucrose solution for dehydration. Place the observation side facing down into the tissue embedding cassettes, completely immersed with O.C.T. (Tissue-Tek, 4583), and transferred to -80 ℃ for freezing. The frozen tissues were labeled with antibodies against CD68 (Abcam, ab283654) followed by staining with Alexa Fluor 488–labeled anti-rabbit and CD90.2 followed by staining with APC–labeled anti-rat, respectively. Finally, counterstained with DAPI. All the images were captured using an optical microscope (Olympus, IX2-SL) or laser scanning confocal microscopy (ZEISS, LSM 880).

Data analysis from the cancer genome atlas (TCGA) data repository

Transcriptome data (log2TPM level, version 20240520) and corresponding clinical annotations were retrieved from the UCSC Xena platform (https://xenabrowser.net/datapages/) using the GDC_TCGA_BRAC dataset. TNBC cases were identified based on receptor status (ER-/PR-/HER2-) from clinical annotations.

Statistics

Statistics were performed on GraphPad Prism 8.3, and data are presented as means ± SEM. Two-tailed Student’s t-test and one-way analysis of variance (ANOVA) were used to analyzed statistical significance. p value of tumor growth curve was calculated using two-way ANOVA. p values < 0.05 were considered statistically significant. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. ns, no significance.