Matrix-Targeting Immunotherapy Controls Tumor Growth and Spread by Switching Macrophage Phenotype

Abstract

The interplay between cancer cells and immune cells is a key determinant of tumor survival. Here, we uncovered how tumors exploit the immunomodulatory properties of the extracellular matrix to create a microenvironment that enables their escape from immune surveillance. Using orthotopic grafting of mammary tumor cells in immunocompetent mice and autochthonous models of breast cancer, we discovered how tenascin-C, a matrix molecule absent from most healthy adult tissues but expressed at high levels and associated with poor patient prognosis in many solid cancers, controls the immune status of the tumor microenvironment. We found that, although host-derived tenascin-C promoted immunity via recruitment of proinflammatory, antitumoral macrophages, tumor-derived tenascin-C subverted host defense by polarizing tumor-associated macrophages toward a pathogenic, immune-suppressive phenotype. Therapeutic monoclonal antibodies that blocked tenascin-C activation of Toll-like receptor 4 reversed this phenotypic switch in vitro and reduced tumor growth and lung metastasis in vivo, providing enhanced benefit in combination with anti-PD-L1 over either treatment alone. Combined tenascin-C:macrophage gene-expression signatures delineated a significant survival benefit in people with breast cancer. These data revealed a new approach to targeting tumor-specific macrophage polarization that may be effective in controlling the growth and spread of breast tumors.

Figure 1. Tumor-derived tenascin-C favors tumor growth in association with macrophages.

(A) Representative western blot of tenascin-C (TNC) expression in NT193 TNC⁺ and TNC⁻ murine mammary tumor cell lines in culture before grafting. (B) Tumor growth of wild-type mice engrafted with TNC⁺ (orange) or TNC⁻ (purple) tumor cells; n=8/group. Correlation between (C) TNC⁺ and (D) TNC⁻ tumor size three weeks after engraftment and the concentration of tenascin-C expressed by tumor cells extracted from corresponding tumors; n=10/group. (E-F) Number and percentage of macrophages (Mp, CD45⁺F4/80⁺CD11b⁺CD11c⁺) in TNC⁺ and TNC⁻ tumors three weeks after engraftment and their correlation with the size of (H) TNC⁺ and (I) TNC^– tumors at the time of sacrifice; n=10/group. (J-K) Percentage of neutrophils (CD45⁺CD11b⁺CD115⁻Ly6C⁺Ly6G^hi) and monocytes (CD45⁺CD11b⁺CD115⁺Ly6C⁺) in TNC⁺ and TNC⁻ tumors three weeks after engraftment. (L) Tumor growth of TNC⁺ and TNC⁻ tumor-bearing mice that received an intravenous (iv) or intratumoral (it) injection of clodronate liposomes (CLO) or control liposomes (CTL); n=5-10/group. Data are representative of at least two independent experiments and are represented as the mean +/- SEM. Mann-Whitney non-parametric T test was used to compare data sets. *P<0.05; **P<0.01; ***P<0.001.

Figure 2. Tumor-derived tenascin-C skews the phenotype of tumor-associated macrophages.

(A) Representative plots and (B) percentage of M1-like (defined as IRF5⁺) and (C) M2-like (defined as CD206⁺) macrophages in TNC⁺ and TNC⁻ tumors three weeks after engraftment into wild-type mice; n=5/group. (D) Gene expression analysis of macrophages sorted from TNC⁺ or TNC⁻ tumors. Data are expressed relative to expression of the endogenous control gene Hprt; n = 6 (TNC⁺) or 8 (TNC⁻). (E-G) Expression of CD86, MHCI and MHCII on macrophages from TNC⁺ or TNC⁻ tumors; n=5-10/group. (H) Heatmap representation of IL1β, IL4, IL6, IL8, IL10, IL12, TGFβ secretion, and intracellular iNOS during coculture of naïve BMMs with TNC⁺ or TNC⁻ tumor cells for 5 days or following LPS activation. Data are expressed relative to unstimulated BMMs cultured for 5 days in the absence of tumor cells; n=3-6/group. (I) Representative images of tenascin-C, F4/80, and CD206 localization in TNC⁺ and TNC⁻ tumors three weeks after engraftment into wild-type mice. (J) Quantification of the number of F4/80⁺ cells in tenascin-C–rich tracks or in the tumor stroma (left panel), and the proportion of CD206⁺ cells in TNC⁺ or TNC⁻ tumors; n=5/group (right panel) (K) The fluorescence intensity of tenascin-C, F4/80, and CD206 in each tumor displayed in the corresponding image in Supplementary Figure S2F (dashed line). Scale bars: 50 μm. Data are representative of at least two independent experiments and are represented as the mean +/- SEM. Mann-Whitney non-parametric T test was used to compare data sets. *P<0.05; **P<0.01; ***P<0.001, ****P<0.0001.

Figure 3. Host-derived tenascin-C promotes an antitumoral macrophage phenotype.

(A) Percentage of macrophages (Mp, CD45⁺F4/80⁺CD11b⁺CD11c⁺) in TNC⁺ and TNC⁻ tumors three weeks after engraftment into wild-type (WT) or tenascin-C–null (KO) mice. (B) Percentage of M2-like (CD206⁺) and (C) MHCII+ macrophages in tumors three weeks after engraftment; n=5-10/group. (D) Quantification of the proportion of CD206⁺ cells amongst F4/80⁺ cells in tumor sections. (E) Representative images of tenascin-C, F4/80, and CD206 localization three weeks after tumor engraftment; n=5/group. Data are representative of at least two independent experiments and are represented as the mean +/- SEM, except for the box and whiskers graphs, representing the median, and minimum to maximum values. Mann-Whitney non-parametric T test was used to compare data sets. Scale bar: 100 μm. **P<0.01; ***P<0.001, ****P<0.0001.

Figure 4. Tenascin-C–conditioned macrophages fail to support T-cell proliferation and favor the polarization of Th17 cells.

Naïve wild-type T cells loaded with CFSE dye were cocultured for 10 days with TNC⁺ or TNC⁻ tumor cells in the presence (T+Mp+TNC^+/–) or absence (T+TNC^+/–) of naïve wild-type BMMs; n=10/group. (A) CFSE mean fluorescent intensity (MFI) indicating T-cell proliferation under the indicated conditions. (B) Closer assessment of the boxed dataset in (A). (C) Proliferation of naïve T cells cultured in presence of CD11b⁺ myeloid cells sorted from TNC⁺ (TNC⁺mye) or TNC⁻ tumors (TNC⁻ mye.); n=6/group. Quantification of secreted (E) IFNγ and (E) IL17 from cultures of naïve T cells with CD11b⁺ cells sorted from TNC⁺ and TNC⁻ tumors; n=6/group. Percentage of (F) Th1 (CD45⁺CD3⁺CD4⁺IFNγ⁺) and (G) Th17 (CD45⁺CD3⁺CD4⁺IL17A⁺) cells in TNC⁺ and TNC⁻ tumors three weeks after engraftment; n=10/group. Data are representative of at least two independent experiments and are represented as the mean +/- SEM, except for the box and whiskers graphs, representing the median, and minimum to maximum values. Mann-Whitney non-parametric T and parametric T tests were used to compare data sets. *P<0.05; **P<0.01; ***P<0.001.

Figure 5. Activation of TLR4 by the FBG domain of tenascin-C induces a phenotypic switch in macrophages.

(A) Expression of MHCII and (B) secretion of IL6 by BMMs after coculture with TNC⁺ or TNC⁻ tumor cells for 5 days in presence of increasing doses of TAK242; n=4/group. (C-D) Proliferation (by CFSE) of naïve purified T cells cocultured with BMMs and TNC⁺ or TNC⁻ tumor cells in the presence of TAK242; n=8/group. Expression of (E) MHCII and (F) secretion of IL6 by BMMs after coculture with TNC⁺ or TNC⁻ tumor cells for 5 days in the presence of anti-FBG at a molar ratio of 1:10, 1:1, or 5:1 to the concentration of tenascin-C produced by TNC⁺ tumor cells over 5 days of culture (green), isotype control (5:1), or 5 μM TAK242. Stimulation of BMMs with LPS in the absence of tumor cells (white) was not susceptible to anti-FBG treatment; n=3-6/group. (G-H) Proliferation of naïve purified T cells cocultured with BMMs and TNC⁺ or TNC⁻ tumor cells in the presence of anti-FBG; n=4-8/group. Data are representative of at least two independent experiments and are represented as the mean +/- SEM, except for the box and whiskers graphs, representing the median, and minimum to maximum values. Mann-Whitney non-parametric T was used to compare data sets. *P<0.05; **P<0.01; ***P<0.001.

Figure 6. Treatment with anti-FBG reorganizes the immune infiltrate and limits tumor growth and spread in vivo.

(A) Tumor growth in mice that received anti–PD-L1 (blue), anti-FBG (10mg/kg, green), or isotype control (beige) antibodies following grafting of TNC⁺ tumor cells and using treatment regimes described in the material and methods (n=14-16/group). (B-G) Analysis of tumor-infiltrating immune cells in mice grafted with TNC⁺ tumor cells that received anti-FBG treatment at 10mg/kg (green) or isotype control (beige); n=6-9/group. (H) Representative images of the localization of tenascin-C, F4/80, and CD206 in anti-FBG– or isotype control–treated mice 21 days following grafting of TNC⁺ tumors; n=4/group. Scale bar: 200μm. (I-J) Tumor growth was assessed every 3 days in MMTV-NeuNT mice that were treated with either anti-FBG (10mg/kg, green), or anti–PD-L1 (blue), the two antibodies together (pink), or isotype control (beige); n=10-12/group. (K) Lung histology representative of metastasis incidence of each group of MMTV-NeuNT mice treated as above. The dashed line delimits the metastasis area. (L) Quantification of number and size of lung metastasis from MMTV-NeuNT mice treated as above. Scale bar: 200μm. Data are representative of at least two independent experiments and are represented as the mean +/- SEM, except for the box and whiskers graphs, representing the median, and minimum to maximum values. Mann-Whitney non-parametric T and parametric T tests were used to compare data sets. *P<0.05; **P<0.01; ***P<0.001.

Figure 7. Tenascin-C expression is associated with a pathogenic TAM phenotype and disease prognosis in human breast cancer.

(A) Pairwise Spearman correlations of tenascin-C (TNC) with macrophage-related genes in The Cancer Genome Atlas (TCGA) bulk RNAseq dataset for human invasive breast carcinoma. Analysis was performed on 1045 tumor samples from individual donors, and p values were adjusted with the Benjamini-Hochberg correction (false discovery rate, FDR). (B) Estimation of tumor-infiltrating immune cell proportions by CIBERSORT analysis, visualized by heatmap and hierarchical clustering, n=1021. (C) Stratification of tumors according to TNC expression and estimated M2 macrophage abundance (left). Subsets in the upper or lower 20% of cases for both conditions were selected. Kaplan-Meier survival analysis of stratified TNC and M2 subsets at 10 years post-diagnosis (right). P values were calculated using the log-rank test and values < 0.05 were considered significant.