Expression of Adipocyte/Macrophage Fatty Acid–Binding Protein in Tumor-Associated Macrophages Promotes Breast Cancer Progression

Breast cancer is the most common cancer in women worldwide and its incidence has increased dramatically from 641,000 cases in 1980 to more than 1.6 million in 2010 (1). Only in the United States, there are an estimated 249,260 new breast cancer cases in 2016 (2). Despite improvements in breast cancer screening and therapy, it still kills nearly half-a-million women annually, about 90% of whom die from distant metastases (3). Therefore, there is great interest in the identification of new cellular and molecular mechanisms that mediate breast cancer growth and metastasis.

It is well-documented that tumor stromal cells actively participate in tumor progression (4, 5). Tumor-associated macrophages (TAM) are the most abundant myeloid cells infiltrating the tumor microenvironment (6). Clinical and experimental evidence has shown that a high density of TAM is associated with both poor prognosis and decreased survival in breast cancer patients. Conversely, deletion of macrophages in mice via genetic and therapeutic approaches results in reduced mammary tumor growth and even tumor regression (7). Overall, these studies corroborate that TAM play a pivotal role in breast/mammary cancer development, but the underlying molecular mechanisms accounting for the protumor activity of TAM remain unclear.

Macrophages are notoriously heterogeneous, exhibiting functional plasticity in different disease contexts (8, 9). Mirroring Th1/Th2 (T helper cell) polarization, activated macrophages are commonly classified as M1/M2 dichotomy, two extremes in a spectrum of functional states (10, 11). M1 macrophages induced by Th1 cytokines are characterized as having proinflammatory and antitumor activities, whereas M2 macrophages, which are driven by Th2 cytokines, increase angiogenesis and exhibit protumor functions (12). In breast cancer, TAM consist of distinct subpopulations with specialized functions in different tumor regions (13, 14). For example, Ly6C⁺ monocytes can differentiate into MHCII⁺CD11c⁺ subsets, displaying an M1-like antitumor phenotype in early-stage tumors. In contrast, a great majority of macrophages exhibit an MHCII⁻CD11c⁻ M2-like protumor phenotype in late stages of tumor development (15). Emerging evidence suggests that different environmental cues or host-produced molecules influence TAM phenotype, skewing them to either a tumor-promoting or a tumor-inhibiting phenotype (16). Therefore, uncovering new factors that are able to shape TAM phenotype and function may lead to novel strategies for breast cancer immunotherapy.

Known as intracellular lipid chaperones facilitating lipid distribution and responses, fatty acid–binding proteins (FABP) are key mediators in the regulation of metabolic and inflammatory pathways inside cells (17, 18). We and others have demonstrated that FABPs, in particular epidermal FABP (E-FABP or FABP5) and adipocyte/macrophage FABP (A-FABP or FABP4), play critical roles in various disease models, including metabolic diseases, chronic inflammation, and cancer development (19–21). Interestingly, new evidence from our recent studies suggests that different macrophage subsets have unique FABP expression patterns (19), implying that FABPs may serve as new factors that influence macrophage phenotype and function. In the present study, we report that A-FABP is upregulated in certain subsets of TAM in breast/mammary tumors, which enhances their ability to promote tumor growth and metastasis through IL6–dependent pathways. Thus, A-FABP may serve as a new functional marker for protumor TAM.

A-FABP WT (FABP4^+/+) mice and A-FABP KO (FABP4^−/−) mice were bred and maintained in the animal facility of the University of Louisville with authorized protocol from the Institutional Animal Care and Use Committee (IACUC). Human samples including normal/benign breast tissues and malignant breast tissues were collected and double-blindly provided by Dr. Edward Sauter. All patients were given written informed consent under an Institutional Review Board–approved protocol in accordance with U.S. Common Rule. Details for tumor cells, syngeneic tumor models are described in the Supplementary Methods.

For analyzing the association of FABPs with the overall survival of breast cancer patients, we performed the analysis using PROGgene. See detailed information in the Supplementary Methods.

Both macrophage cell lines and bone marrow–derived macrophages were used in this study. See detailed information of macrophage culture and transfection assays in the Supplementary Methods.

Confocal staining was performed to determine A-FABP expression and macrophage infiltration in human breast tumor tissues. See the Supplementary Methods for the details.

Total RNA was extracted using the RNeasy Mini Kit (Qiagen). cDNA synthesis was performed using the QuantiTect Reverse Transcription Kit (Qiagen). The quantitative real-time PCR was carried out with SYBR Green PCR Master Mix on the StepOnePlus Real-Time PCR Systems. For miR-29b expression, qPCR was conducted by TaqMan microRNA assays and normalized by U44/48. Target expression was analyzed using the ΔCT approach. See detailed primer sequences in the Supplementary Table S1.

For cell surface staining, immune cells in different organs or in the tumor stroma were stained with anti-CD4 (GK1.5), anti-CD8 (53-6.7), anti-NK1.1 (PK136), anti-CD19 (1D3), anti-F4/80 (BM8), anti-CD11b (M1/70), anti-Ly6C (HK1.4) and anti-MHCII (M5/114.15.2). For intracellular A-FABP staining, cells were stained with anti–A-FABP antibody (MAB3150) after fixation and permeabilization. Cells were acquired by BD FACS Fortessa and BD FACS AriaII Cell Sorter. All the data were analyzed by FlowJo 10.1 as previously described (20).

E0771 tumor cell colony formation assay was performed to determine whether A-FABP expression in TAM promotes E0771 colony formation. Details are in the Supplementary Methods.

The procedures of SDS-PAGE and immunoblotting were described as our previous studies (19). Briefly, to measure the expression of target molecules, proteins were acquired from cell lysates and quantified by BCA assay. Anti–mouse-IL6 and anti-NFκB were from Abcam. Anti–A-FABP was from R&D System. Other antibodies, including anti-STAT3, antiphospho-STAT3 (Tyr705), β-actin, GAPDH, were from Cell Signaling Technology.

A total of 5 × 10⁴ 293T cells were seeded into 24-well plates for overnight and cotransfected with 2μg of IL6 3-UTR luciferase plasmid (Plasmid #52884, Addgene) with miR-29b or its scramble control. pRL-TK Renilla luciferase plasmid (Promega) was cotransfected as an internal control. Firefly luciferase activity was measured 48 hours after transfection by using a dual luciferase reporter assay system (Promega) according to the manufacturer's instruction. Values were corrected to account for differences in transfection efficiency by pRL-TK Renilla luciferase activity. All luciferase data represent mean±SD of three independent experiments.

ELISA kits for mouse IL6, IL10, or TNFα (Biolegend) were used to detect respective cytokines in mouse serum and cell cultural supernatants according to the manufacturer's instruction.

miRNA microarray was performed to analyze the miRNA expression profile between WT and A-FABP^−/− macrophages. See detailed information in the Supplementary Methods.

For analysis of microRNA data, the raw microarray images were imported into Partek Genomics Suite (version 6.6) with a customized normalization method. The cleaned data matrix was then exported to R (version 3.1) and analyzed by Limma packages. The significantly differentially expressed miRNAs were those that showed at least 2-fold change in expression with a false discovery rate (FDR)– corrected P value less than 0.05. For other in vitro or in vivo experiments, a two-tailed, unpaired Student t test was performed for the comparison of results. A P value of <0.05 is regarded as statistically significant.

To investigate the emerging roles of FABPs in breast cancer, we first analyzed publicly accessible raw databases from Gene Expression Omnibus (GEO) and The Cancer Genome Atlas using online tool PROGgene to determine whether FABP expression was associated with breast cancer survival. We found that high expression of A-FABP (FABP4), but not other members of FABP family in breast tumor tissues, was significantly associated with low overall survival of total breast cancer patients (n = 596), triple-negative patients, and single negative (ER⁻, PR⁻, or Her2⁻) patients (Supplementary Fig. S1 and Supplementary Table S2). As A-FABP was primarily expressed in stromal cells, including adipocytes and macrophages, we further analyzed FABP expression pattern in the stroma of breast tissues using GEO dataset GSE9014 (22). We found that A-FABP was the most strongly upreglulated FABP member in breast cancer tissues as compared to normal controls (P = 3.06E−16; Fig. 1A). Moreover, the upregulation of A-FABP in the stroma was independent of subtypes of breast cancer (Fig. 1B). To verify the potential procancer activity of A-FABP in breast cancer, we analyzed A-FABP expression in benign and malignant breast tissues. Compared with normal/benign breast tissues A-FABP expression was elevated in cancer tissues (Fig. 1C). Significantly more A-FABP positive cells were accumulated in the stroma of malignant tumors regardless of ER status (Fig. 1D). These data strongly suggest that A-FABP may represent a new unappreciated factor in tumor stromal tissues promoting breast cancer progression.

If A-FABP upregulation was able to promote breast cancer progression, we speculated that A-FABP deficiency might inhibit tumor growth and metastasis. To this end, we employed A-FABP deficient (A-FABP^−/−) mouse models to assess the impact of A-FABP deficiency on mammary tumor growth and metastasis. Of note, A-FABP^−/− mice developed normally and exhibited similar lipid metabolism, adipokines and immunophenotypes as their WT counterparts under the standard chow diet (Supplementary Fig. S2A–S2F; refs. 23–25). E0771 tumor cells derived from a C57BL/6 mouse mammary adenocarcinoma were orthotopically injected into the mammary fat pad of A-FABP^−/− mice and their WT littermates. E0771 tumors grew much slower and average tumor weight was 3-fold less in A-FABP^−/− mice than in WT mice (Fig. 2A and B). A-FABP deficiency also significantly reduced E0771 cell-derived lung metastasis (Fig. 2C and D). To determine whether host-derived A-FABP represents a general mechanism for promoting tumor progression, we further evaluated tumor growth and metastasis in WT and A-FABP^−/− mice using different types of tumor cells, including mouse mammary tumor cells (MMT) and colon cancer cells (MC38). Similar to the E0771 model, both tumors exhibited significant decrease in tumor growth and lung metastasis when A-FABP was not present in host mice (Fig. 2E–H). Notably, E0771 tumor cells themselves expressed A-FABP whereas the other tested cell lines had low or no A-FABP expression (Supplementary Fig. S2G and S2H). However, regardless of A-FABP expression status in tumor cells, A-FABP deficiency in the host inhibited tumor growth and metastasis. These data collectively suggest that host-derived A-FABP plays a critical role for promoting tumor progression.

To investigate how A-FABP promotes tumor progression, we first sought to identify which population of stromal cells was responsible for the upregulation of A-FABP expression. In naïve mice without tumors, we evaluated different populations of leukocytes separated from the spleen and found that A-FABP expression was predominantly confined to F4/80⁺ macrophages (Fig. 3A). Confocal microscopy staining confirmed that A-FABP was expressed in the cytoplasm of bone marrow–derived macrophages from WT mice (Fig. 3B). Interestingly, when macrophages were cultured in the presence of tumor supernatants, A-FABP expression in macrophages was further enhanced (Fig. 3C). By mimicking in vivo crosstalk of tumor/macrophages, we co-cultured macrophages with tumor cells in Transwell inserts and confirmed that macrophages responded to tumor exposure with elevation of A-FABP expression (Fig. 3D). In contrast, when adipocytes, which also highly expressed A-FABP, were cocultured with tumor cells (Supplementary Fig. S3A), A-FABP expression in adipocytes was downregulated (Supplementary Fig. S3B), suggesting that macrophages, but not adipocytes, are the major populations responsible for the upregulation of A-FABP in the tumor stroma. To substantiate this speculation, macrophages separated from mammary tumors or the spleen of tumor-bearing mice exhibited enhanced expression of A-FABP as compared with those from naïve mice (Fig. 3E). Furthermore, we measured A-FABP expression using surgically-removed benign and malignant human breast tissues and confirmed the elevated expression of A-FABP in CD163⁺ TAM in patients with invasive breast cancer (Supplementary Fig. S3C). Taken together, our in vitro and in vivo data suggest that TAM account for the elevated A-FABP expression in the tumor stroma.

Next, we sought to determine whether A-FABP–expressing TAM were responsible for the enhanced tumor growth using clodronate-mediated macrophage depletion assays (26, 27). When mice were intraperitoneally injected with clodronate liposomes or control liposomes, splenic macrophages were successfully depleted in WT and A-FABP^−/− mice (Supplementary Fig. S4A). Using this approach, we depleted macrophages in WT and A-FABP^−/− mice and measured transplanted tumor growth in these mice. Importantly, macrophage depletion significantly reduced E0771 tumor growth in WT mice, but not in A-FABP^−/− mice (Fig. 3F). The marked differences in E0771 tumor weight between WT and A-FABP^−/− mice also disappeared after macrophage depletion (Fig. 3G). Consistently, we observed similar results with MC38 tumors after clodronate treatment with significant depletion of TAM in the tumor stroma (Fig. 3H–J). Of note, macrophage depletion with clodronate had no obvious impact on the infiltration of other immune cell populations, including CD8⁺ T cells, CD4⁺ T cells, B cells and NK cells, in the tumor stroma (Supplementary Fig. S4B–S4E), suggesting a specific role of macrophages in promoting tumor progression. Altogether, these data indicate that A-FABP is upregulated in macrophages that contribute to tumor progression.

Considering the heterogeneous phenotype and function of TAM, we hypothesized that A-FABP might be differentially expressed in different subsets of macrophages. Using high-power confocal microscopy, we observed that A-FABP (red) was expressed in half of the TAM (green) in the tumor stroma of human invasive breast tumor tissues (Fig. 4A). To determine the phenotype of A-FABP⁺ macrophages, we gated on CD11b⁺F4/80⁺ macrophages and identified distinct macrophage subsets using MHCII and Ly6C as markers (Q1 to Q4) in the spleen of naïve WT and A-FABP^−/− mice (see gating strategy and FACS definition of Q1 to Q4 in Supplementary Fig. S5A). We separated each subset with a flow sorter and analyzed A-FABP expression levels. We found that A-FABP was preferentially expressed in the Q4 subset of splenic macrophages, which exhibited the F4/80⁺CD11b⁺MHCII⁻Ly6C⁻ phenotype (Fig. 4B). When we analyzed the peripheral blood using the same gating strategy, monocytes in the blood were divided into two major subsets: Q1 and Q4 (Supplementary Fig. S5B). Consistently, A-FABP was preferentially expressed in the Q4 subset of the peripheral monocytes similar to their counterparts in the spleen (Fig. 4C). When we further analyzed tumor infiltrating macrophage subsets in tumors, about 40% of TAM were located in the Q4 quadrant (Supplementary Fig. S5C). Importantly, A-FABP was also upregulated in this specific subset of TAM (Fig. 4D). Thus, A-FABP seemed to be preferentially expressed in the monocytes/macrophage subset, which exhibited the phenotype of CD11b⁺F4/80⁺MHCII⁻Ly6C⁻. These results imply that A-FABP may facilitate tumor progression by enhancing the protumor activity of this specific TAM subset.

We next examined the dynamic changes in monocytes/macrophages by focusing on the Q4 subset in tumor-bearing mice over a 3 to 4 weeks period (days 0, 3, 12, and 24 after tumor implantation). We observed that the percentage of the Q4 subset quickly increased in the spleen and blood after tumor cell implantation, and reached peak levels at day 12. However, it decreased dramatically to the basal levels around 4 weeks when mice were sacrificed at the IACUC-required endpoints (Fig. 4E and F). In sharp contrast, the Q4 subset in the tumor stroma increased slowly at the early stage (before day 12), but expanded dramatically at the late stage after tumor cell implantation (Fig. 4G; Supplementary Fig. S5D), suggesting a protumor function of the Q4 subset. To demonstrate that Q4 macrophages were directly involved in promoting mammary tumor growth, we separated splenic Q4 subset from WT and A-FABP^−/− mice using a flow sorter, and co-injected with E0771 cells into A-FABP^−/− mice, respectively. Strikingly, Q4 macrophages from WT mice, but not A-FABP^−/− mice, significantly promoted E0771 tumor growth (Fig. 4H). Tumor size and weight in A-FABP^−/− mice coinjected with WT Q4 macrophages were comparable with these of WT mice. In contrast, Q4 macrophages from A-FABP^−/− mice had no protumor effect (Fig. 4I and J). All these results indicate that A-FABP is highly expressed in the Q4 macrophage subset, which is accumulated in the tumor stroma directly promoting tumor progression in an A-FABP–dependent manner.

To dissect how A-FABP exerts its protumor function in TAM, we separated major TAM subsets from the tumor stroma using a flow sorter as described above (Supplementary Fig. S5C), and analyzed well-known tumor-related molecules produced by TAM. Interestingly, high expression of A-FABP in the Q4 subset of TAM did not impact the expression of well-known markers for M1 and M2, such as iNOS, Arg-1, IL10. and TNFα (Fig. 5A–D), nor did it influence other tumor-related molecules in macrophages, including IL12, CXCR-1, TGFβ, MMP-9, and MMP-1 (Supplementary Fig. S6A–S6E). However, A-FABP⁺ Q4 TAM exhibited elevated levels of IL6 expression as compared with A-FABP^−/− counterparts (Fig. 5E). By measuring circulating levels of cytokines in naïve and tumor-bearing mice, we observed that tumors increased the levels of IL6, but not IL10 and TNFα, in WT mice, but not in A-FABP^−/− mice (Fig. 5F–H), suggesting an A-FABP–dependent effect in IL6 upregulation in vivo. To verify this effect, we measured IL6 production first by knocking down A-FABP expression with A-FABP shRNA in WT macrophages (Fig. 5I–J), and then by employing A-FABP-deficient macrophages (Fig. 5K–L). Results from both real-time PCR and Western blotting confirmed that IL6 upregulation was controlled by A-FABP expression in macrophages. To further confirm the A-FABP–regulated IL6 production, we analyzed their association using human tumor samples. In the laser-dissected breast cancer stroma samples (GSE9014), we found that A-FABP was significantly positively correlated with IL6 expression (Fig. 5M). This positive correlation was also confirmed in a dataset using whole breast tumor tissues (GSE18229 with 223 breast cancer patients; Supplementary Fig. S6F). All these in vitro and in vivo evidence indicates that A-FABP controls IL6 production by TAM.

To further investigate the molecular mechanisms by which A-FABP regulates IL6 production in TAM, we conducted miRNA arrays using WT and A-FABP^−/− macrophages due to the emerging role of miRs in regulating macrophage function and immune responses in the tumor microenvironment (28, 29). Principal component analysis (PCA) showed that miR clusters of WT macrophages displayed a distinct pattern than that of A-FABP^−/− macrophages, suggesting biological differences due to the deficiency of A-FABP (Supplementary Fig. S7A). Heat map analysis identified a group of miRs that were differentially expressed between WT and A-FABP^−/− macrophages (Supplementary Fig. S7B). Among them, upregulation of miR-29b in A-FABP^−/− macrophages (Fig. 6A) particularly attracted our attention as our previous studies have shown that A-FABP enhances activation of NFκB, which negatively regulates miR-29b expression (30, 31). We thus speculated that A-FABP may regulate tumor-promoting IL6 production through the NFκB/miR29b pathway in macrophages. To this end, we first demonstrated that A-FABP deficiency significantly impaired tumor-stimulated NFκB activation in macrophages (Fig. 6B). Using real-time PCR, we next confirmed that miR-29b was indeed elevated when A-FABP was either transiently or stably knocked down in macrophages (Fig. 6C and D). Importantly, when macrophages were transfected with NFκB p65 shRNA (Fig. 6E), knocking down of NFκB not only enhanced miR-29b expression (Fig. 6F), but also inhibited IL6 production at both mRNA and protein levels (Fig. 6G and H). Considering that miRs negatively regulate target gene expression through direct binding to their 3′UTR regions, we identified a target sequence on IL6 3′UTR (414-420), which paired with miR-29b using the miR target prediction algorithm TargetScan. To validate the direct regulation of miR-29b binding to IL6 3′UTR, we cotransfected 293T cells with IL6-3′UTR luciferase plasmid and miR-29b or scramble controls, and demonstrated that miR-29b was able to significantly reduce IL6 3′UTR luciferase activity as compared with the scramble control (Fig. 6I). Moreover, when miR-29b was overexpressed in macrophages (Fig. 6J), IL6 production was significantly inhibited at both mRNA and protein levels (Fig. 6K and L). The negative association of A-FABP with miR-29b expression was further verified in human breast cancer samples when we analyzed a well-annotated cohort of 207 breast cancer patients with matched mRNA and microRNA expression profile (GEO dataset: GSE22220; P < 0.0001; Supplementary Fig. S7C). In addition, we found that higher miR-29b expression was positively associated with better breast cancer survival probability in this dataset (Supplementary Fig. S7D), which further supports the functional role of the A-FABP/miR-29b pathway in human breast cancer patients. Collectively, all these data suggest that A-FABP regulates IL6 production, at least partially, via the NFκB/miR-29b pathway in macrophages.

Considering the well-characterized pathway of IL6/STAT3 in promoting tumor growth and metastasis (32, 33), we next determined whether A-FABP⁺ macrophages promote tumor cell invasiveness by focusing on IL6/STAT3 signaling. When E0771 tumor cells were cocultured with A-FABP⁺ WT or A-FABP^−/− macrophages, respectively, we observed that WT macrophages, but not A-FABP^−/− macrophages, greatly induced STAT3 phosphorylation in E0771 cells (Fig. 7A). Of note, STAT3 phosphorylation could be blocked by anti–IL6–neutralizing antibody, suggesting that TAM activate tumor cells through A-FABP-dependent IL6/STAT3 signaling. Moreover, we performed tumor-colony formation assays as further verification of the protumor function of A-FABP⁺ TAM. As shown in Fig. 7B and C, macrophage expression of A-FABP significantly promoted tumor colony formation, which was able to be blocked by anti-IL6–neutralizing antibody. As miR-29b inhibited IL6 production in macrophages, we cocultured E0771 tumor cells with miR-29b- or scramble-transfected macrophages and showed that miR-29b-transfected macrophages significantly reduced STAT3 phosphorylation and colony formation of E0771 cells when compared with scramble controls (Supplementary Fig. S8A–S8C), also confirming the functional role of A-FABP/miR-29b/IL6 signaling in enhancing the invasiveness of tumor cells in vitro.

To further test protumor function of A-FABP⁺ macrophages in vivo, we cocultured E0771 tumor cells with either A-FABP⁺ or A-FABP⁻ macrophages, and respectively implanted them into the mammary fat pads of mice for observation of tumor growth and metastasis. Indeed, tumor cells conditioned with A-FABP⁺ macrophages grew much faster than those conditioned with A-FABP⁻ macrophages (Fig. 7D and E). In addition, tumor cells treated with A-FABP⁺ macrophages showed enhanced capability to metastasize to the lung as compared to these treated with A-FABP⁻ macrophages (Fig. 7F and G; P < 0.001). As E0771 tumor cells, but not host cells from lung tissues, exhibited unique a CD44⁺CD24⁻ phenotype (Supplementary Fig. S8D–S8E), we also quantitatively analyzed CD44⁺CD24⁻ tumor cells metastasizing to lung tissues by flow cytometric staining. The data clearly showed that A-FABP depletion significantly decreased tumor growth and metastatic spread (Supplementary Fig. S8F–S8G). Taken together, our data confirmed that A-FABP expression in macrophages is critical for tumorigenic signaling, and tumor progression, thereby establishing the protumor function of A-FABP in macrophages.

Cancer immunotherapy, in particular macrophage-directed immunotherapy, has been proposed as a breakthrough strategy for eliminating cancer (34). However, due to the complexity of macrophage biology, uncovering novel cellular and molecular mechanisms determining TAM phenotype and function will be critical for translation of these seminal ideas into the clinical setting. Herein, we identified A-FABP as a new tumor-promoting factor in the context of breast cancer. Breast stromal levels of A-FABP were similar among different subtypes of breast cancer, but significantly higher than noncancer controls, suggesting that A-FABP upregulation in tumor stroma represents a general mechanism of protumor effects regardless of subtypes of breast cancer. Most importantly, we demonstrated that A-FABP promoted tumor progression via enhancing the protumor activity of TAM, specifically within a subset of TAM that exhibit a CD11b⁺F4/80⁺Ly6C⁻MHCII⁻CD11c⁻ phenotype. Mechanistically, A-FABP expression in TAM facilitated the protumor IL6/STAT3 signaling through regulation of the NFκB/miR-29b pathway. Thus, the current study not only establishes A-FABP as a new marker for protumor macrophages, but also increases our mechanistic understanding of several important aspects of macrophage-based tumor immunotherapy.

TAMs are notoriously heterogeneous exhibiting both anti- and protumor activities. However, the identification of functional subsets has been challenging. Although the classical M1/M2 dichotomy has served as a model of macrophage function, the in vivo setting is very complex due to functional overlap and lack of specificity. For example, Arginase I, usually classified as a M2 marker since it converts arginine to tumor-promoting ornithine, is also upregulated in LPS/IFNγ-stimulated M1 macrophages (35). Through analysis of publicly accessible databases, we found that high expression of A-FABP in human breast tumor tissues was significantly associated with a lower survival rate of breast cancer patients. Mice deficient for A-FABP exhibited reduced tumor growth and metastasis as compared with their WT littermates. Macrophages and adipocytes are the two main populations that highly express A-FABP. Our data demonstrated that A-FABP upregulation in tumor was primarily restricted to a subset of TAM, but not in adipocytes in the tumor stroma. Using clodronate-mediated macrophage depletion assay we showed that macrophage expression of A-FABP was key to their protumorigenic effect. Phenotypically, A-FABP was highly expressed in the Q4 subset of monocytes/macrophages in both mice and humans. Functionally, A-FABP⁺ TAM separated from mammary tumors exhibited enhanced production of tumor-promoting molecules (e.g., IL6) as compared with A-FABP^−/− TAM. Moreover, we further showed that A-FABP expression in macrophages was critical for tumorigenic signaling and colony formation of mammary tumor cells in vitro and their growth and metastasis in vivo. Thus, our data suggest that A-FABP may serve as a new functional marker for protumor macrophages.

As the most abundant stromal cells in breast/mammary tumors, TAM account for up to 50% of the cells in the tumor mass (36, 37). The question remains as to how TAMs are dynamically regulated during tumor development. Although mounting evidence suggests that TAM originate primarily from bone marrow-derived circulating monocytes (38, 39), the factors controlling the dynamic changes in heterogeneous subsets of TAM in the tumor stroma remain unclear. Most circulatory mouse monocytes are CD11b⁺F4/80⁺MHCII⁻Ly6C⁺ inflammatory monocytes (Q1 subset) and CD11b⁺F4/80⁺MHCII⁻Ly6C⁻ patrolling monocytes (Q4 subset). Although published studies have shown that Ly6C⁺ monocytes are able to differentiate into different subsets of macrophages in tumor models and in muscle injury (15, 40), distinct populations of macrophages in inflammatory tissues can also be sequentially mobilized from the circulation with recruitment of Ly6C⁺ monocytes in the early stage, whereas Ly6C⁻ monocytes appear in the late stage (41, 42). Our data showed that A-FABP was preferentially expressed in CD11b⁺F4/80⁺MHCII⁻Ly6C⁻ macrophages (Q4 subsets) in different tissues, including the spleen, PBMCs, and tumors. By dynamic analysis of macrophage phenotypes in syngeneic mouse models, our previous and current studies demonstrated that macrophages, which were recruited in the tumor stroma in the early stages (1 week after tumor implantation), mainly exhibited a CD11b⁺F4/80⁺Ly6C⁺ phenotype with high expression of E-FABP and exerted antitumor function. In contrast, macrophages that accumulated in the late stage of tumors (2–4 weeks after tumor implantation) exhibited a CD11b⁺F4/80⁺MHCII⁻Ly6C⁻ phenotype with high levels of A-FABP expression and promoted tumor growth and metastasis. In addition, the dynamic changes in macrophage phenotype within the tumor stroma corresponded well to the changes of circulating monocytes in the peripheral tissues, suggesting that TAM could also be recruited from monocytes in the periphery. In addition, we also found that A-FABP is highly expressed in the CD14^dimCD16⁺ monocytes in humans. Thus, measuring A-FABP⁺ monocytes/macrophages in peripheral blood may provide new evidence for the evaluation of tumor progression in the early stages of tumor development.

How do TAM exert their pro- or antitumor activity? In syngeneic mouse models, we noticed that mammary tumor formation was critical in the first week after tumor cell implantation. Once tumors were established, tumor cells exhibited exponential growth. Indeed, early-infiltrating TAM have been shown to exhibit antitumor activity through E-FABP–mediated type I IFNβ responses (19). In the current study, we noticed that significant tumor growth differences began around 2 weeks (14–17 day) after tumor implantation between WT and A-FABP^−/− mice. Moreover, TAM at these stages supported tumor progression by upregulation of A-FABP, which promoted tumor growth via IL6/STAT3 signaling. In line with our observations, the tropic role of A-FABP in promoting tumor growth was confirmed in other models (43, 44). Of note, FABPs constitute a family of 9 proteins with different patterns of tissue distribution. Members of FABPs are traditionally thought to exhibit similar functions via transporting fatty acids and coordinate lipid responses inside cells. However, our research demonstrated an unexpected but novel finding that suggests opposing roles of individual FABPs in regulating macrophage phenotype and function in the tumor setting. Simply put, E-FABP expressed in a subset of CD11b⁺F4/80⁺MHCII⁺Ly6C⁺ macrophages enhances their antitumor function, whereas A-FABP expressed in a subset of CD11b⁺F4/80⁺MHCII⁻Ly6C⁻ macrophages facilitates their protumor function.

Mounting evidence suggests an important role of miRs in regulating immune responses and tumor development and progression. Among them, miR-29b has been shown to act as a tumor suppressor in many types of cancer including breast cancer (45, 46). Through direct inhibition of oncogenic genes, such as STAT3 and cyclin-dependent protein kinase 6, in tumor cells, miR-29b can greatly inhibit tumor growth (45, 47). Moreover, macrophages transfected with miR-29b is associated with M1-associated transcripts (29). These studies suggest a beneficial effect of miR-29b in inhibiting tumor progression by targeting both tumor cells and immune cells. In the current studies, we demonstrated for the first time that expression of A-FABP in macrophages downregulated miR-29b through enhanced NFκB activity, which substantiated our previous observation that activation of NFκB repressed miR-29b expression in leukemia cells (30). We further showed that upregulated miR-29b in A-FABP^−/− macrophages was responsible for the inhibition of IL6/STAT3 signaling and subsequent tumor colony formation, providing a molecular mechanism by which A-FABP promotes the protumor activity of TAM through regulation of the miR-29b/IL6 axis.

Targeting TAM for tumor immunotherapy has been proposed as a promising clinical treatment (34, 48), but this has been difficult to achieve in practice due to the heterogeneous and dynamic nature of macrophages in the tumor environment. For example, strategies of either blocking TAM recruitment or depleting TAM and their progenitors have not proven clinically feasible (49). Considering that TAM can exhibit both pro- and antitumor activity depending on different disease settings, identification of functional subsets is key for the design of effective immunotherapies. In the current study, we identified A-FABP as a functional marker for protumor TAM. Thus, inhibition of A-FABP may offer a novel strategy for cancer treatment by specifically targeting tumor-promoting TAM. Indeed, inhibition of A-FABP activity through small-molecule inhibitors has been shown effective in treatment of diabetes and atherosclerosis. It will be of great interest to test their therapeutic efficacy in the treatment of breast cancer.

In summary, our studies suggest A-FABP as a new predictor for breast cancer progression through enhancement of the protumor activity of TAM. Specifically, A-FABP is preferentially expressed in a subset of macrophages (CD11b⁺F4/80⁺MHCII⁻Ly6C⁻) that are directly involved in promoting tumor growth. Mechanistically, this macrophage subset accumulated in the tumor stroma enhances tumor progression through A-FABP–mediated miR-29b/IL6/STAT3 cascade (Supplementary Fig. S9). Thus, A-FABP may serve as a new functional marker for protumor TAM, and targeting A-FABP⁺ TAM may represent a novel strategy for tumor immunotherapy.

No potential conflicts of interest were disclosed.

Conception and design: A. Triplett, D.A. Bernlohr, M.P. Cleary, J. Suttles, B. Li

Development of methodology: J. Hao, F. Yan, Y. Zhang, B. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Hao, Y. Zhang, Y. Sun, J. Zeng, E. Sauter

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Hao, F. Yan, A. Triplett, Y. Zhang, K.A.T. Silverstein, Q. Zheng, E. Sauter, B. Li

Writing, review, and/or revision of the manuscript: J. Hao, K.A.T. Silverstein, D.A. Bernlohr, M.P. Cleary, N.K. Egilmez, E. Sauter, S. Liu, J. Suttles, B. Li

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Hao, Y. Sun, S. Liu, B. Li

Study supervision: S. Liu, B. Li

Other (processed the samples and provided the resulting data for analysis): D.A. Schultz

The authors thank the Mayo Clinic Genome Analysis Core for the help with performing microRNA expression array of macrophages and Minnesota Supercomputing Institute for computing resources used for the microarray analysis. This work was supported by NIH R01 grants CA18098601A1 (to B. Li), CA17767901A1 (to B. Li), and AI048850 (to J. Suttles).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.