Remodeling the tumor microenvironment via blockade of LAIR-1 and TGF-β signaling enables PD-L1-mediated tumor eradication

Abstract

Collagens in the extracellular matrix (ECM) provide a physical barrier to tumor immune infiltration, while also acting as a ligand for immune inhibitory receptors. Transforming growth factor-β (TGF-β) is a key contributor to shaping the ECM by stimulating the production and remodeling of collagens. TGF-β activation signatures and collagen-rich environments have both been associated with T cell exclusion and lack of responses to immunotherapy. Here, we describe the effect of targeting collagens that signal through the inhibitory leukocyte-associated immunoglobulin-like receptor-1 (LAIR-1) in combination with blockade of TGF-β and programmed cell death ligand 1 (PD-L1). This approach remodeled the tumor collagenous matrix, enhanced tumor infiltration and activation of CD8+ T cells, and repolarized suppressive macrophage populations, resulting in high cure rates and long-term tumor-specific protection across murine models of colon and mammary carcinoma. The results highlight the advantage of direct targeting of ECM components in combination with immune checkpoint blockade therapy.

Trial registration: ClinicalTrials.gov NCT00001846.

Keywords: Cancer immunotherapy; Immunology.

Conflict of interest statement

Conflict of interest: Authors from NextCure, Inc. are employees or officers of the company. CP discloses spouse’s employment and holdings in MacroGenics, Inc. All other authors from the NCI do not have any competing interests to disclose. The NCI has ongoing Cooperative Research and Development Agreements with NextCure, Inc. and EMD Serono (CrossRef Funder ID: 10.13039/100004755).

Figures

Figure 1. NC410 and bintrafusp alfa synergize for effective tumor control.

(A and F) Representative images of MC38 and EMT6 tumors analyzed for collagen (trichrome staining), NC410-biotin, and control IgG-biotin staining. Scale bars: 50 μm (A) and 100 μm (F). (B) Treatment schedule for mice bearing MC38 tumors. Individual tumor growth and number of cures (C) and average tumor growth (D) are shown; n = 5 mice/group (control) or n = 6 (all other groups). Data are representative of 1 of 2 independent experiments. (E) Left: Cured mice were rechallenged s.c. with MC38 and LLC tumor cells. Right: Control C57BL/6 mice were injected s.c. with either MC38 or LLC tumor cells. Graphs show individual tumor growth and number of mice free of tumor. (G) Treatment schedule for mice bearing s.c. EMT6 tumors. Individual tumor growth and number of cures (H) and average tumor growth (I) are shown; n = 8 mice/group (control) or n = 9 (all other groups). (J) Left: Cured mice from the combination group were rechallenged s.c. with EMT6 and 4T1 tumor cells. Right: Control BALB/c mice were injected s.c. with either EMT6 or 4T1 tumor cells. Graph shows individual tumor growth. Error bars indicate SEM of biological replicates. *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001 by 2-way ANOVA (D and I).

Figure 2. Expression of target molecules and treatment effect on tumor immune infiltrates.

(A) Treatment schedule for mice bearing MC38 tumors. (B) Immunofluorescence-based analysis of harvested tumors for human IgG, indicating presence of therapeutic agents. Graph shows the fluorescence signal across 15 regions of interest (ROIs) randomly selected within each tumor section, n = 2–3 tumors per treatment group, normalized to the average signal in the control group. (C) Representative images of MC38 tumors. Scale bar: 50 μm. (D) scRNA-seq profiling of tumor-infiltrating CD45+ cells isolated from tumors treated as indicated in A. UMAP plots for all treatment groups combined and analyzed as described in the Methods section, showing selected identified immune cell subset clusters with events colored according to cell type. (E) Frequency of selected cell subsets identified by scRNA-seq analysis from MC38 tumors treated as indicated in A. (F and G) Expression of selected genes of relevance by scRNA-seq in single-color UMAP plots (F) or in bubble plot representation across selected immune cell subset clusters. Bubble size shows percentage of cells expressing the indicated gene; color intensity represents scaled expression levels. Data from scRNA-seq analysis are from a single experiment. (H) Flow cytometry analysis of LAIR-1 expression on indicated immune cell subsets in the blood, spleen, and tumors from MC38 and EMT6 tumor–bearing mice; n = 5 mice/group (MC38), n = 4/group (EMT6). Tissues for analysis were collected on day 12 prior to any treatment. For violin plots, dashed line displays the median and dotted lines display quartiles. Data from MC38 spleen and tumor flow cytometry are representative of 1 of 2 independent experiments.

Figure 3. Combination therapy increases infiltration with activated CD8+ T cells.

(A) UMAP plots showing CD4+ and CD8+ T cells, T regulatory (Treg) cells, NK and NKT cell clusters as identified by scRNA-seq analysis from MC38 tumors treated as in Figure 2A. (B) Frequency of indicated immune cell subsets, as a percentage of total CD45+ cells. (C) Flow cytometry analysis of indicated immune infiltrating cells or (E) analysis of CD8+ T cells for expression of Ki67 or granzyme B (GzmB) in MC38 tumors collected on day 17 following treatment with NC410 (250 μg) and/or bintrafusp alfa (250 μg) on days 9, 11, and 13. Graphs show the number of cells per mg tumor weight; n = 7 (control, NC410, bintrafusp alfa), n = 5 (NC410 + bintrafusp alfa). (D) UMAP plots showing expression of selected genes by scRNA-seq. (F) Representative images of CD8+ T cell infiltrates (magenta) in MC38 tumors treated as indicated in Figure 2A. DAPI (cyan) was used as a nuclear stain. Scale bars: 100 μm and 10 μm (insets). (G) IFN-γ ELISPOT analysis of spleens from MC38 tumor–bearing mice treated as indicated, against the p15E tumor antigen; n = 6/group. Representative images of well signals from 2 individuals per group are displayed. (H) Average tumor growth of MC38 tumors untreated or treated with NC410 plus bintrafusp alfa with or without depleting antibodies for CD4+, CD8+, or NK cells; n = 6 in the NK depletion group; n = 7 in all other groups. For violin plots, dashed line displays the median and dotted lines display quartiles. Error bars indicate SEM of biological replicates. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001 by 1-way ANOVA followed by Tukey’s post hoc test in C, E, and G and 2-way ANOVA in H.

Figure 4. NC410 plus bintrafusp alfa reduces tumor infiltration with tumor-associated M2 macrophages.

(A) Flow cytometry analysis of total macrophages and CD38+ macrophages in MC38 tumors collected on day 17 following treatment with NC410 (250 μg) and/or bintrafusp alfa (250 μg) on days 9, 11, and 13. Graphs show the number of cells per mg tumor weight; n = 7 (control, NC410, bintrafusp alfa), n = 5 (NC410 + bintrafusp alfa). For violin plots, dashed line displays the median and dotted lines display quartiles. (B) Top 10 activated GO gene pathways in M1 clusters identified by scRNA-seq in the NC410 plus bintrafusp alfa versus the control group. UMAP plots showing (C) expression of Mrc1 and Cd163 genes used to identify M2 cell clusters by scRNA-seq, and (D) variations in their expression across treatment groups. (E) Frequency of subpopulations of M2 macrophages according to their expression of Cd163 and Mrc1. (F) Selected activated GO/REACTOME/KEGG/HALLMARK gene pathways in Cd163negMrc1pos M2 clusters identified by scRNA-seq in the NC410 plus bintrafusp alfa versus the control group. (G) Bubble plot representation of the top 30 upregulated and top 20 downregulated genes (logFC ≥ 0.25 or ≤ –0.25 and P value ≤ 0.05) in Cd163negMrc1pos M2 clusters from the NC410 plus bintrafusp alfa group versus the control group. Bubble size shows percentage of cells expressing the indicated gene; color intensity represents scaled expression levels. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001 by 1-way ANOVA followed by Tukey’s post hoc test in A.

Figure 5. Remodeling of collagen in tumors treated with NC410 plus bintrafusp alfa therapy.

(A) Representative images of immunofluorescence-based staining of denatured collagen utilizing a linearized collagen hybridizing peptide (CHP, magenta) in MC38 tumors collected as indicated in Figure 2A. White dash–outlined squares identify magnified regions in bottom images. (B) Mean fluorescence intensity value of denatured collagen across treatment groups; 10 regions of interest (ROIs) randomly selected within each tumor section; n = 3 tumors per treatment group. For violin plots, dashed line displays the median and dotted lines display quartiles. (C) Staining of denatured collagen in control MC38 tumors utilizing linearized CHP peptide or nonlinearized CHP as a negative control. (D) Staining of denatured collagen in control MC38 tumors in the absence or presence of NC410 to rule out competition for binding to collagens. (E) Total collagen content measured with CHP staining after heat retrieval in MC38 tumors collected 1 day following the second dose of treatments, as indicated in Figure 2A. DAPI staining of nuclei (cyan) is shown. (F) Representative images of MC38 tumors treated with control, NC410, bintrafusp alfa, and NC410 plus bintrafusp alfa analyzed for collagen expression by trichrome staining. Scale bars: 20 μm (A, bottom panels, and D), 50 μm (A, C, and E), and 100 μm (F). ****P ≤ 0.0001 by 1-way ANOVA followed by Tukey’s post hoc test in B.

Figure 6. Inhibition of TGF-β and PD-L1 are both required for optimal tumor control in combination with NC410.

MC38 tumor–bearing mice were administered indicated doses of NC410, bintrafusp alfa, anti–PD-L1, or TGF-β trap control on days 9, 11, and 13 after tumor injection. Graphs show (A) individual tumor growth and number of cures in each group, and (B) average tumor growth; n = 6 mice/group (bintrafusp alfa, NC410 + bintrafusp alfa) or n = 7 (control, anti–PD-L1, TGF-β trap control, NC410, NC410 + anti–PD-L1, NC410 + TGF-β trap control). Data are representative of 1 of 2 independent experiments. Error bars indicate SEM of biological replicates. *P ≤ 0.05; ***P ≤ 0.001; ****P ≤ 0.0001 by 2-way ANOVA. (C) Treatment schedule of indicated therapeutic agents; CD45+ cells isolated from MC38 tumors collected on day 12 were used for scRNA-seq analysis. (D) Frequency of effector CD4+, CD8+, T regulatory (Treg) cells, NK and NKT cell clusters as determined by scRNA-seq, shown as a percentage of total CD45+ cells. (E) Frequency of subpopulations of M2 macrophages according to their expression of Cd163 and Mrc1. (F). Bubble plot representation of all genes differentially expressed (logFC ≥ 0.25 or ≤ –0.25 and P value ≤ 0.05) in total CD45+ cells from the NC410 plus bintrafusp alfa group versus all other groups. Bubble size shows percentage of cells expressing the indicated gene; color intensity represents scaled expression levels. (G) UMAP plots showing expression of selected genes by scRNA-seq analysis on CD45+ cells in each treatment group. Data from scRNA-seq analysis are from a single experiment; control and NC410 plus bintrafusp alfa groups from Figures 2–4 are shown for comparison.

Figure 7. NC410 alters the M2 polarization phenotype of human macrophages in vitro.

(A) Schematic detailing human macrophage polarization process and purity as determined by flow cytometry via CD68+/CD11b+ staining. (B) Gene expression heatmap based on RNA-seq analysis depicting the top 500 genes differentially expressed by variance in M2-like polarized macrophages across all treatment groups; (n = 3 donors/group). Volcano plots of differentially expressed genes between (C) NC410-treated and control groups, (D) bintrafusp alfa versus control and NC410 versus NC410 plus bintrafusp alfa groups; red dots indicate genes with an adjusted P value ≤ 0.05; genes related to M1/M2 macrophage polarization are indicated. (E) Gene expression heatmap depicting selected M1/M2 macrophage polarization genes in control and NC410-treated M2-like human macrophages; shown at the bottom is the z-score scale. (F) Top 10 significantly activated (left panel) and deactivated (right panel) GO/REACTOME/KEGG/HALLMARK gene pathways in NC410-treated versus control M2-like human macrophages. (G) Flow cytometry data depicting the mean fluorescence intensity (MFI) of CD163 expression on M2-like macrophages prepared from PBMCs from healthy donors via culture in the presence of IL-4, IL-13, and collagen (n = 12 donors) or a mix of tumor-conditioned media (TCM) and collagen (n = 9 donors), left untreated or treated with NC410 for 48 hours, as indicated in panel A. *P ≤ 0.05; **P ≤ 0.01 by 2-tailed Student’s t test. (H) Flow cytometry histograms of representative donors in G, showing both CD206 and CD163 expression with indicated percentage positive cells and MFI (in parentheses) of total cells. Data from RNA-seq analysis are from a single experiment.

Figure 8. Expression of collagens, PD-L1, and TGF-β1 in colon carcinoma tissues.

(A) Representative images of tissues from a colon cancer tumor microarray (TMA) stained for collagen content via Masson’s trichrome stain (upper row), NC410-biotin (center row), and control IgG-biotin (bottom row). Shown are a representative case each with low, intermediate, and high collagen content and corresponding low, intermediate, and high binding of NC410. Black dash–outlined squares identify magnified regions in adjacent images. Scale bars: 200 μm (whole sections) and 50 μm (zoomed images). Representative images of (B) a primary colon tumor, (C) a metastatic lymph node (LN), (D) a liver metastasis, and (E) a lung metastasis from colon cancer stained for binding of NC410 (NC410-biotin, brown), cytokeratin to identify tumor cells (CK, green), total leukocyte infiltration (CD45, white), PD-L1 (red), CD163 to identify M2-like macrophages (white), and TGF-β1 mRNA by RNA in situ hybridization (red). DAPI was used to stain nuclei (blue). Scale bars: 50 μm (B and C–E).