Activating a collaborative innate-adaptive immune response to control metastasis

Summary

Tumor-associated macrophages (TAMs) promote metastasis and inhibit T cells, yet macrophages can be polarized to kill cancer cells. Macrophage polarization could thus be a strategy for controlling cancer. We show that macrophages from metastatic pleural effusions of breast cancer patients can be polarized to kill cancer cells with monophosphoryl lipid A (MPLA) and interferon γ (IFNγ). MPLA+IFNγ injected intratumorally or intraperitoneally reduces primary tumor growth and metastasis in breast cancer mouse models and suppresses metastasis and enhances chemotherapy response in an ovarian cancer model. Both macrophages and T cells are critical for the treatment’s anti-metastatic effects. MPLA+IFNγ stimulates type I IFN signaling, reprograms CD206⁺ TAMs to inducible NO synthase (iNOS)⁺ macrophages, and activates cytotoxic T cells through macrophage-secreted interleukin 12 (IL-12) and tumor necrosis factor α (TNFα). MPLA and IFNγ are used individually in clinical practice and together represent a previously unexplored approach for engaging a systemic anti-tumor immune response.

Graphical Abstract

eTOC Blurb

Sun et al. combine monophosphoryl lipid A (MPLA) with interferon γ (IFN γ) to reprogram macrophages, activate cytotoxic T cells, suppress tumor growth and metastasis, and enhance chemotherapy response in mice. MPLA and IFNγ are used individually in clinical practice and together represents a previously unexplored approach for eliciting anti-metastatic responses.

Introduction

Cancer elicits an immune response that solid tumors can evade, in part by establishing an immunosuppressive microenvironment. The tumor microenvironment (TME) comprises tissue-resident cells (e.g., fibroblasts, endothelial cells), innate immune cells (e.g., macrophages), and adaptive immune cells (e.g., T cells) (Lim et al., 2018; Nakasone et al., 2012; Pogge von Strandmann et al., 2017). It was discovered over a century ago that intratumoral injection of dead bacteria, named “Coley’s toxin” after the physician who devised the treatment, led to durable anti-tumor responses in some patients (Wiemann and Starnes, 1994). These responses were likely caused by bacterial components activating immune cells, particularly macrophages. Yet, tumor-associated macrophages (TAMs) usually promote tumor growth and metastasis (Arwert et al., 2018; Kitamura et al., 2017; Yin et al., 2016). They also contribute to the immunosuppressive microenvironment, allowing tumors to evade intrinsic anti-tumor immune response and therapies (Cassetta and Kitamura, 2018; Peranzoni et al., 2018; Ruffell et al., 2014).

Macrophages exist on a continuum of phenotypes between classically and alternatively activated (Wynn et al., 2013). Classically activated macrophages are stimulated during acute inflammation by interferon γ (IFNγ) and toll-like receptor (TLR) ligands, including bacterially derived lipopolysaccharide (LPS). These macrophages secrete pro-inflammatory cytokines (tumor necrosis factor α [TNFα] and interleukins IL-1β, IL-2, IL-6, IL-12, and IL-23) and metabolize arginine to the “killer” molecule nitric oxide (NO) by upregulating inducible nitric oxide synthase (iNOS). Stimulating macrophages with IL-4/IL-13 leads to the alternatively activated phenotype and production of anti-inflammatory cytokines (IL-10, transforming growth factor [TGFβ]). TAMs usually exhibit traits of the alternatively activated macrophage subtype and have been clinically targeted by ablating strategies, such as anti-colony stimulating factor 1 receptor (CSF1R) therapeutics (Cannarile et al., 2017; Cassetta and Pollard, 2018). CSF1R inhibitors, e.g., PLX3397 (pexidartinib) and LY3022855, have been tested clinically. In patients with advanced tenosynovial giant cell tumor, PLX3397 is well tolerated and causes an anti-tumor response (Tap et al., 2015). However, PLX3397 is ineffective against recurrent glioblastoma (Butowski et al., 2015), and LY3022855 is ineffective against advanced, treatment-refractory breast or prostate cancer (Autio et al., 2020). Depending on context, macrophages can also eliminate cancer cells with NO, reactive oxygen species, or by phagocytosis, and they can stimulate cytotoxic T lymphocytes to kill tumors by presenting antigens and producing pro-inflammatory cytokines (Pozzi et al., 2005; Wynn et al., 2013). Since macrophages exhibit a high degree of plasticity, another strategy than macrophage ablation may be to “reprogram” TAMs to kill cancer cells and support cytotoxic T cells (Kowal et al., 2019; Pathria et al., 2019). Class IIa histone deacetylase (HDAC) inhibitors reprogram TAMs to be anti-tumor and increase chemotherapy and immunotherapy efficacy in breast cancer models (Guerriero et al., 2017). Similarly, inactivating macrophage phosphatidylinositol-3-Kinase γ (PI3Kγ) promotes an immunostimulatory transcriptional program, in turn restoring CD8⁺ T cell activation and increasing survival in mouse models of head and neck squamous cell carcinoma and breast carcinoma (Kaneda et al., 2016).

Reprogramming macrophages and/or directly stimulating adaptive immunity have been achieved using e.g., TLR agonists. TLRs are pattern recognition receptors that play fundamental roles in activating innate immunity (Adams, 2009). Nanoparticles loaded with the TLR7/8 agonist R848 accumulate in TAMs, polarize them to an inflammatory phenotype, and suppress colon cancer in a mouse model (Rodell et al., 2018). A TLR7/8 dual agonist conjugated to tumor-targeting antibodies elicits robust localized activation of macrophages and dendritic cells, resulting in tumor clearance and immunological memory in breast tumor models (Ackerman et al., 2021). Moreover, combining motolimod (a TLR8 agonist) with the epidermal growth factor receptor inhibitor cetuximab increases IL-6, CCL2, and CCL4 secretion and increases natural killer (NK) cell circulation and activation (Chow et al., 2017). Similarly, the TLR7 ligand imiquimod applied as a cream to breast cancer skin metastases shows anti-tumor activity and triggers a strong T-helper-1 (Th-1)/cytotoxic immune response in responding patients (Adams et al., 2012; Rozenblit et al., 2019). Yet, the therapeutic efficacy of motolimod and imiquimod is low: only 2 of 11 squamous cell carcinoma patients and 2 of 10 breast cancer patients achieved a partial response.

To improve the TLR agonists’ anti-tumor effects, we combined the TLR4 agonist monophosphoryl lipid A (MPLA) with IFNγ. MPLA is an LPS derivative; it can be chemically synthesized and displays reduced toxicity while maintaining most of LPS’s immunostimulatory activity (Mata-Haro et al., 2007). MPLA is used as an adjuvant in vaccines against cervical cancer and shingles (Lai et al., 2015; Monie et al., 2008). IFNγ is a cytokine essential for innate and adaptive immunity. It can promote macrophage activation, enhance antigen presentation, and regulate Th1/Th2 balance (Boehm et al., 1997; Schroder et al., 2004). Reprogramming TAMs in murine breast tumors depends on IFNγ (Clark et al., 2020). Furthermore, IFNγ 1b (a variant of IFNγ) is FDA-approved to treat chronic granulomatous disease and osteopetrosis (Key et al., 1992; Todd and Goa, 1992). Here, we tested if injecting MPLA with IFNγ (hereafter, “MPLA+IFNγ”) could provide a collaborative stimulus that would reprogram TAMs to become tumoricidal and activate T cells to elicit local and systemic immune responses to control metastatic breast cancer and ovarian cancer in mice.

Results

MPLA+IFNγ reprograms TAMs to be tumoricidal

We first evaluated whether TAMs could be reprogrammed to become tumoricidal. TAMs and tumor cells were isolated from the primary tumors of MMTV-PyMT mice (“PyMT” for short, a mouse model of luminal B breast cancer) (Figure S1A, S1B), co-cultured, and treated with a TLR agonist (LPS, MPLA, or polyI:C) alone or in combination with IFNγ. None of the TLR agonists alone or combined with IFNγ affected PyMT cancer cell viability in the absence of macrophages (Figure S1C). However, IFNγ together with either LPS or MPLA—both TLR4 agonists—activated TAMs, resulting in killing ~90% of cancer cells in 48 hours in the co-cultures (Figure 1A, 1B, Movies S1, S2). PolyI:C (a TLR3 agonist) combined with IFNγ, or LPS, MPLA, polyI:C, or IFNγ alone did not induce the tumoricidal activity of the macrophages (Figure 1A, 1B).

Figure 1. MPLA+IFNγ induces tumoricidal effects of breast TAMs from mice and patients.

(A) Macrophages and PyMT tumor cells were isolated from primary tumors of MMTV-PyMT (BL/6) mice, fluorescently labeled, and co-cultured for 48 hours as indicated. (B) Statistical analysis of relative PyMT cancer cell numbers for panel A. (C) The effect of macrophages on PyMT tumor cells were studied using: 1) CM: conditioned medium of vehicle- or MPLA+IFNγ-treated macrophages was added to the cancer cells; 2) Transwell: macrophages and cancer cells were seeded in upper or lower chambers, respectively; 3) Direct contact: macrophages and cancer cells were co-cultured in direct contact. B, C: Graph depicts the average results of three independent experiments. (D) Macrophages and EpCAM⁺ cancer cells were isolated from pleural effusions of breast cancer patients and co-cultured as indicated. EpCAM⁺ cancer cells did not proliferate during co-culture. For A–D, macrophages were stained with Deep Red Dye (falsely colored green); cancer cells were stained with CMFDA Dye. LPS, MPLA, polyI:C, mouse IFNγ, or human IFNγ was added at 0 hour. (E) Gene expressions in pleural effusion-derived macrophages (cultured without cancer cells) was determined by RT-qPCR 24 hours after indicated treatments were initiated. NOS2, CD40, TNF, and IL12B (p40 subunit of IL-12) are markers of tumoricidal macrophages. IL10 and MRC1 are markers of TAMs. The relative expression to ACTB was normalized to the relative expression in vehicle-treated samples, which was set to 1. D, E: Graph depicts the average results from three patients. One-way ANOVA was used for all panels. Data are represented as mean ± SD (standard deviation). See also Figure S1, Movies S1 and S2.

iNOS is required for macrophages’ tumoricidal activity (Dinapoli et al., 1996). Consistently, only the combination treatments—IFNγ + LPS or MPLA—induced iNOS expression (Figure S1D). Moreover, inhibiting iNOS activity with L-NIL blocked the tumoricidal effect of MPLA+IFNγ-activated macrophages (Figure S1E). Thus, TAMs can be reprogrammed to become tumoricidal using IFNγ with LPS or MPLA, and iNOS activity is required for their tumoricidal effects. iNOS metabolizes arginine to generate NO, which diffuse into adjacent cells and cause cell death (Kelm, 1999; Stuehr and Nathan, 1989). However, NO has a half-life of just 4-6 seconds (Kelm, 1999). Consistently, only when the MPLA+IFNγ-treated macrophages were directly cultured with cancer cells—not when the cells were separated by transwells (slowing diffusion) or when conditioned medium from MPLA+IFNγ-treated macrophages was added—could the macrophages kill the cancer cells (Figure 1C).

Nos2 (the gene coding for iNOS) expression is regulated by four transcription factors (NFκB, AP-1, STAT1, IRF1) (Gao et al., 1997; Martin et al., 1994; Salim et al., 2016; Xie et al., 1993). Signaling from TLR4, the MPLA receptor, can activate NFκB and AP-1 via MyD88 (Lu et al., 2008), and IFNγ activates the JAK/STAT1 pathway, leading to phosphorylated STAT1 dimer formation and IRF1 activation (Castro et al., 2018). Consistently, MPLA+IFNγ did not induce Nos2 expression or induce tumoricidal activity in Tlr4 knockout bone marrow-derived macrophages (BMDMs, Figure S1F–H). Furthermore, MPLA alone induced NFκB phosphorylation, while IFNγ alone induced STAT1 phosphorylation, and both transcription factors were phosphorylated by MPLA+IFNγ, as determined by Western blot of BMDMs (Figure S1I). Finally, inhibiting NFγB, STAT1/IRF1, or AP-1 was sufficient to prevent MPLA+IFNγ–induced Nos2 expression (Figure S1J). Together, these data suggest that MPLA and IFNγ induce separate, complementary signaling pathways that are required for maximal iNOS expression and thus tumoricidal activity.

To formally test whether patient-derived TAMs could be activated to kill patient-derived cancer cells, we isolated macrophages and cancer cells from the metastatic pleural effusions of breast cancer patients (Figure S1K–M). MPLA+IFNγ could reprogram these macrophages, killing 80–90% of the cancer cells isolated from the corresponding patient in vitro (Figure 1D). We next used RT-qPCR to identify TAM reprogramming markers. MPLA+IFNγ increased the expression of tumoricidal macrophage markers (NOS2, CD40, TNF, IL12B) and reduced the expression of TAM markers (IL10, MRC1) in the patient-derived macrophages (Figure 1E). Tnf and Il12b upregulation and Il10 and CD206 downregulation were also observed in primary PyMT tumor-derived macrophages after MPLA+IFNγ treatment (Figure S1N, S1O).

The class IIa HDAC inhibitor TMP195 and PI3Kγ inhibitor IPI-549 can reprogram TAMs to control tumor growth and metastasis. Using TAMs from PyMT tumors and drug concentrations from prior publications (Guerriero et al., 2017; Kaneda et al., 2016), we found that TMP195 did not increase Nos2 or Il12b or inhibit Arg1, Mrc1, or Il10 expression, but did increase Cd86 and H2-Q2 expression. Similarly, IPI-549 only increased H2-Q2 and inhibited Il10 expression (Figure S1P). Furthermore, neither TMP195 nor IPI-549 activated TAMs to kill cancer cells (Figure S1Q). Thus, MPLA+IFNγ is specifically effective at inducing Nos2 expression and reprogramming TAMs to be tumoricidal in vitro.

MPLA+IFNγ suppresses breast tumor growth and lung metastasis

We next asked: Could MPLA+IFNγ control cancer cells in vivo, in the context of a complex TME? We used two metastatic breast cancer models: transplanted PyMT (luminal B subtype) or 4T1 (triple negative subtype) breast tumors. After determining optimal doses (1 μg MPLA + 1 μg IFNγ per PyMT tumor; 1 μg MPLA + 3 μg IFNγ per 4T1 tumor; Figure S2A, S2B), we found that intratumorally injected MPLA+IFNγ suppressed primary tumor growth in both models (Figure 2A, 2B). LPS can promote metastasis by inducing neutrophil extracellular traps (NETs) (Albrengues et al., 2018; Yang et al., 2020). Yet, MPLA+IFNγ did not increase NET levels in plasma of treated mice (Figure S2C, S2D). Importantly, MPLA+IFNγ did not enhance metastasis: it instead significantly inhibited lung metastasis. All vehicle-treated mice had detectable metastasis, while only 2 out of 5 PyMT and 4 out of 9 4T1 tumor-bearing mice had metastases after MPLA+IFNγ treatment (Figure 2C, 2D). IFNγ alone also inhibited tumor growth and lung metastasis, but less than the combination treatment, and MPLA alone had no effect. We also employed the parental MMTV-PyMT mouse model (on both the C57BL/6 and FVB/N genetic background). In this model, intratumoral injection of MPLA+IFNγ also suppressed primary tumor growth and lung metastasis (Figure 2E–J). Finally, since intratumoral injection is not the standard of care for any cancer type and not all tumors can be intratumorally injected, we tested intraperitoneal administration of MPLA+IFNγ: higher doses of MPLA (30 μg/mouse) and IFNγ (30 μg/mouse) were needed to reduce tumor growth (Figure S2E, S2F). Importantly, these higher doses did not change the body weight of C57BL/6 mice (Figure 2K), nor did it cause elevated serum levels of aspartate aminotransferase, alkaline phosphatase, albumin, or creatine kinase (markers of liver, heart, or kidney damage) (Figure 2L). Moreover, histological evaluation of liver, lung, kidney, bone marrow, heart, and spleen from the treated mice revealed no signs of toxicity (Figure 2M).

Figure 2. MPLA+IFNγ suppresses breast tumor growth and lung metastasis in mice.

(A, B) Tumor growth curve of (A) transplanted PyMT or (B) 4T1 tumors, arrows indicate time of treatments; 1 μg MPLA, 1 μg (A) or 3 μg (B) mouse IFNγ, or both (MPLA and IFNγ were mixed, then injected) per tumor. N=8 tumors/group. (C, D) Lung metastatic burden of (C) PyMT (N=4–6 mice/group) and (D) 4T1 tumor-bearing mice (N=5–9 mice/group) determined from histology. (E, H) Tumor growth curves, (F, I) representative H&E staining of lungs (scale bar: 4 mm), and (G, J) lung metastatic burden determined from histology of spontaneous C57BL/6-MMTV-PyMT (E–G, N=6-7 mice/group) and FVB/N-MMTV-PyMT (H–J, N=6-10 mice/group) mice (1 μg MPLA+1 μg IFNγ per tumor). Lungs were collected 3 days after last treatment. One-way ANOVA (2A–D) or t-test (2E, 2G, 2H, and 2J) was performed for tumor volume analysis at the end point and for lung metastasis analysis, and due to unequal group variances, Welch’s ANOVA was used for panels 2A, 2C, 2D, and Welch’s t-test was used for panels 2E, 2G, 2J. ***p<0.001, compared to vehicle. Data are represented as mean ± SD. (K) Body weight and (L) serum levels of AST (aspartate aminotransferase), ALP (alkaline phosphatase), albumin, and creatine kinase of tumor-free C57BL/6 mice treated with vehicle (N=9) or MPLA+IFNγ, treated as in Figure S2F (N=10). T-test was performed for panels 2K and 2L. Data are represented as mean ± SD. n.s.: not significant. (M) Representative H&E stained sections of liver, lung, kidney, bone marrow, heart, or spleen. Scale bar: 100 μm. Serum and tissues were collected 2 days after the 6^th treatment. See also Figure S2.

We next benchmarked MPLA+IFNγ against other macrophage polarization approaches (TMP195 and IPI-549), as well as PLX3397 with anti-PD1 checkpoint immunotherapy (Wainberg et al., 2016). MPLA+IFNγ was more effective at reducing PyMT tumor growth and lung metastasis than the other treatments at previously published doses (DeNardo et al., 2011; Guerriero et al., 2017; Kaneda et al., 2016) (Figure S2G–I). Together, these results suggest that a safe and effective dose of MPLA+IFNγ could be used in future clinical trials.

MPLA+IFNγ stimulates chemokine secretion and type I IFN signaling

To explore how MPLA+IFNγ controls tumor growth and metastasis, we first focused on changes to the TME induced during early treatment phases. PyMT tumors were collected 24 hours after the 2^nd intratumoral injection and digested. The cytokines released to the supernatant were analyzed by cytokine arrays (Figure 3A). We found that MPLA+IFNγ induced the secretion of a large number of chemokines responsible for immune cell recruitment, including CCL5 (leukocytes, including monocytes), CXCL9 (lymphocytes, particularly T cells), CXCL13 (B cells), CCL12 (monocytes), and CXCL5 (neutrophils) (Figure 3B). We also detected increased CD40 (a costimulatory protein on antigen-presenting cells), IL-12 (which enhances the cytotoxic activity of NK cells and CD8⁺ T cells), and BAFF (B-cell activating factor) (Figure 3B). Increased CXCL9 and CCL5 secretion after MPLA+IFNγ treatment was confirmed by enzyme-linked immunosorbent assay (ELISA) (Figure S3A, S3B). Thus, the combined MPLA+IFNγ treatment dramatically altered the cytokine microenvironment.

Figure 3. MPLA+IFNγ enhances chemokine secretion and stimulates the type I IFN signaling pathway in breast tumors.

(A) Experimental design for obtaining cytokine array and RNA-seq data; 1 μg MPLA, 1 μg mouse IFNγ, or both were used per PyMT tumor. (B) Heatmap of the secreted cytokines that were changed by MPLA+IFNγ compared to vehicle. (C) Venn diagrams of upregulated and downregulated genes in tumors after the indicated treatments. MPLA+IFNγ uniquely upregulated 156 genes and downregulated 32 genes that did not show altered transcription when tumors were treated by MPLA or IFNγ alone. (D) Heatmap of the top genes that were uniquely changed in the MPLA+IFNγ group. (E) Gene ontology analysis of the genes that were uniquely up- or down-regulated in the MPLA+IFNγ group. *FDR<0.05, **FDR<0.01, ***FDR<0.001, ****FDR<0.0001, compared to vehicle. See also Figure S3 and Data file S1.

We next explored MPLA+IFNγ’s effect on gene regulation by RNA-seq of bulk tumors. There were only 20 upregulated and 5 downregulated genes after MPLA treatment alone; 33 upregulated and 15 downregulated genes after IFNγ treatment alone, but 179 upregulated and 36 downregulated genes after the combined MPLA+IFNγ treatment (Figure 3C, Data file S1). Both IFNγ alone and MPLA+IFNγ increased the expression of IFNγ-inducible genes (Data file S1). In addition, MPLA+IFNγ uniquely enhanced the expression of the type I IFN (IFNα/β)-inducible genes Stat1, Stat2, Irgm2, Gbp2, and Cxcl9 (Figure 3D). MPLA+IFNγ also uniquely decreased the expression of extracellular matrix (ECM)-associated genes (Col3a1, Col4a1, Col12a1, and Igfbp3) (Figure 3D). The altered transcription of select type I IFN-inducible and ECM-associated genes was verified in PyMT tumors, and was also shown to occur in MPLA+IFNγ-treated 4T1 tumors, by RT-qPCR (Figure S3C–F). These results indicate that MPLA+IFNγ activated immune cells and also had wider effects on the TME.

To investigate the implications of the altered gene programs, we performed gene ontology (GO) analysis (Figure 3E), which revealed enhanced immunostimulatory responses (such as type I IFN signaling, antigen processing and presentation, and defense response to virus) and inhibited immunosuppressive responses (such as collagen-activated signaling and ECM organization). Overall, these data showed that only the combination MPLA+IFNγ treatment promoted the expression of genes involved in effector immune cell recruitment and stimulated the type I IFN signaling pathway, thus inducing changes associated with a strong anti-tumor response.

MPLA+IFNγ reprograms TAMs and activates cytotoxic T cells

We next elucidated the contributions of immune cells to the anti-tumor effects. By flow cytometry (Figure S4A–K), we found that as a percentage, there were almost three times as many immune (CD45⁺) cells in PyMT tumors after intratumoral injection of MPLA+IFNγ as in vehicle-treated tumors (Figure S4B). Among immune cells (Figure S4C–K), the proportion of B cells was decreased (Figure S4G), while the proportions of CD4⁺ T cells (Figure S4E) and inflammatory monocytes (Ly6C⁺Ly6G⁻, Figure 4A) were increased. Macrophage infiltration in PyMT tumors (F4/80⁺, Figures 4B, S4I) was also enhanced. Not only did the tumors contain more macrophages, but the proportion of macrophages that expressed iNOS (Figure 4C) or CD40 (Figure 4D) was increased, while the proportion of macrophages expressing CD206 was reduced (Figure 4E). MPLA+IFNγ also increased the numbers of iNOS⁺ or CD40⁺ cells, as determined by immunofluorescence (IF) staining (Figure 4F, 4G). Furthermore, IF staining showed STAT1 and NFκB p65 phosphorylation in macrophages (and other cells) when IFNγ was given with MPLA (Figure S4L), consistent with our in vitro data (Figure S1I, S1J). Together, these results suggest that MPLA+IFNγ had reprogrammed TAMs to tumoricidal macrophages in the mouse breast tumors.

Figure 4. MPLA+IFNγ reprograms TAMs to be tumoricidal in breast tumors.

(A) Flow cytometry of inflammatory monocytes in PyMT tumors. N=5 mice/group. (B) IF staining of macrophages (F4/80⁺) in PyMT tumors. In total, 6–14 fields of view (FOVs) were evaluated from 4 mice/group. Scale bar: 50 μm. (C–E) The percentages of tumoricidal macrophages (iNOS⁺ or CD40⁺) and TAMs (CD206⁺) of total macrophages were identified by flow cytometry. N=3–6 mice/group. (F, G) IF staining of iNOS⁺ (F) and CD40⁺ cells (G) in PyMT tumors. In total, 6–14 FOVs were evaluated from 4 mice/group. Scale bar: 50 μm. (H) Antigen-presenting activities of macrophages determined by the percentage of H-2Kb OVA⁺ cells in PyMT-chOVA tumors. N=5 mice/group. For all figure panels, the analysis was performed 2 days after the last (6^th) intratumoral injection. One-way ANOVA was used for panels 4A, 4C–E, and 4H, and Welch’s ANOVA was used for panels 4B, 4F, and 4G. Data are represented as mean ± SD. See also Figure S4.

We next used a variant of the MMTV-PyMT model, the MMTV-PyMT-chOVA model (Engelhardt et al., 2012; Fein et al., 2020) to explore the effect of MPLA+IFNγ on MHC class I antigen-presentation. Cancer cells in the MMTV-PyMT-chOVA model express the ovalbumin (OVA) peptide SIINFEKL, allowing measurement of MHC class I presentation of OVA (H-2Kb OVA⁺). In the tumors of these mice, there was an almost twofold increase in the percentage of H-2Kb OVA⁺ macrophages (Figure 4H), H-2Kb OVA⁺ dendritic cells (Figure S4M), and CD103⁺CD11c^HighCD11b^High cells (Figure S4N) after MPLA+IFNγ treatment. In the PyMT tumors, MPLA+IFNγ significantly increased the infiltration of total (CD8a⁺, Figure 5A, S4D) and activated (CD107a⁺ or PD1⁺) (Figure 5B, 5C) cytotoxic T cells. Importantly, the combined treatment also promoted the production of effector memory T cells (CD44⁺CD62L⁻) in the tumors (Figure 5D). We also observed increased T cell accumulation in the lungs of tumor-bearing mice (which at the analyzed time point contain small metastasis) (Figure S5A), suggesting that disseminated cancer cells elicited a stronger T cell response in MPLA+IFNγ-treated mice. Together, these data suggest MPLA+IFNγ reverse immune suppression in the TME and activate lasting adaptive T cell immune responses to protect against future metastases.

Figure 5. MPLA+IFNγ can activate cytotoxic T cells through macrophage-secreted IL-12 and TNFα and antigen presentation.

(A) IF staining of CD8⁺ T cells in PyMT tumors. In total, 5–9 FOVs were evaluated from 4 mice/group. Scale bar: 50 μm. (B–D) The percentages of cytotoxic T cells (CD8⁺CD107a⁺, B), PD1⁺CD8⁺ T cells (C), and effector memory CD8⁺ T cells (CD44⁺CD62L⁻, D) in PyMT tumors were identified by flow cytometry. For panels A–D, the analysis was performed 2 days after the last (6^th) intratumoral injection. (E–G) CD69 expression of CD8⁺ T cells in in vitro co-culture systems was identified by flow cytometry. (E) CD8⁺ T cells isolated from the spleens of PyMT tumor-bearing mice were cultured with or without macrophages derived from the bone marrow of the same mice, and MPLA or IFNγ or both were added to the cell cultures immediately after T cell seeding. (F) CD8⁺ T cells were cultured with macrophages similarly to (E), and neutralized antibodies against IL-12, TNFα, or CXCL9 were added 30 minutes before adding MPLA+IFNγ. (G) Macrophages derived from the bone marrow of PyMT tumor-bearing mice were activated with MPLA (or IFNγ or both) for 16 hours, then incubated with OVA-Q4H7 peptide (10⁻⁹ M) for 2 hours, washed with PBS and added to CD8⁺ T cells isolated from the spleens of OT1 mice. One-way ANOVA was used for panels 5C, 5E, and 5F, and Welch’s ANOVA was used for panels 5A, 5B, 5D, and 5G. Data are represented as mean ± SD. See also Figure S5.

Intratumoral injection of MPLA+IFNγ in tumors of the parental MMTV-PyMT mouse models also increased infiltration of immune cells, macrophages, CD8⁺ T cells; activated cytotoxic T cells (CD107a⁺ or PD1⁺); decreased the percentage of CD206⁺ macrophages; and increased the percentage of iNOS⁺ macrophages (Figure S5B). We also found more activated myeloid cells (CD40⁺) and cytotoxic, activated T cells (CD107a⁺ or PD1⁺) in the lungs (Figure S5C, S5D)—cells that may have been recruited and/or activated in response to the spontaneous dissemination of cancer cells. We next tested whether systemic MPLA+IFNγ treatment would also alter the PyMT tumor immune microenvironment. As we had found after intratumoral injection of MPLA+IFNγ, the total numbers of immune cells, macrophages, CD8⁺ T cells, activated CD107a⁺ CD8⁺ T cells, and PD1⁺ CD8⁺ T cells were increased after intraperitoneal injection, and there was a switch in macrophage marker expression, with decreased CD206 and increased iNOS expression (Figure S5E). Taken together, both local and systemic administration of MPLA+IFNγ can alter the immune microenvironment of primary tumors and lung metastases.

MPLA+IFNγ can activate cytotoxic T cells through macrophage-secreted cytokines and antigen presentation

Our in vivo data strongly suggested that MPLA+IFNγ activated cytotoxic T cells either directly or indirectly via macrophages. We incubated CD8⁺ T cells isolated from the spleens of PyMT tumor-bearing mice with MPLA+IFNγ in vitro, but found that the treatment did not directly stimulate T cell activation (Figure 5E). In contrast, when these CD8⁺ T cells were co-cultured with macrophages derived from the bone marrow of the same mice, MPLA+IFNγ increased the expression of CD69 on CD8⁺ T cells (Figure 5E). These results suggest that MPLA+IFNγ-activated macrophages stimulate cytotoxic T cell function. MPLA+IFNγ promoted the expression of IL-12, TNFα, and CXCL9 in macrophages and PyMT tumors (Figures 1E, S1N, 3B, S3A). Thus, we explored the possible roles of these three cytokines in macrophage-T cell crosstalk in vitro (Figure 5F). The increased expression of CD69 by T cells cultured with macrophages was partially reduced by incubation with the IL-12 or TNFα neutralizing antibody, indicating that MPLA+IFNγ activates cytotoxic T cells through the macrophage-secreted cytokines IL-12 and TNFα. Neutralizing CXCL9 had no effect on CD69 expression, consistent with a likely role in T cell recruitment rather than activation. Blocking IL-12 and TNFα together did not completely prevent the increased CD69 expression, suggesting that factors besides IL-12 and TNFα contribute to macrophage-T cell crosstalk.

We next challenged CD8⁺ T cells from OT1 mice with OVA peptide-incubated macrophages. OT1-derived T cells recognize cells with the MHC class I-bound OVA peptide. We found that the OT1 CD8⁺ T cells were activated, as determined indirectly by CD69 expression, by the combination of OVA and MPLA+IFNγ–treated macrophages (Figure 5G), while MPLA or IFNγ alone had no effect. Taken together, our in vitro results show that MPLA+IFNγ not only activates macrophages’ tumoricidal effects but also enhances their capacity to activate cytotoxic T cells.

Both macrophages and T cells are essential for MPLA+IFNα’s anti-tumor effects

We next performed in vivo antibody depletion experiments to test which immune cell populations are necessary for MPLA+IFNγ to eliminate metastatic breast cancer. Cell depletion was confirmed by flow cytometry (Figure S6A, S6B). In PyMT tumors, T cell depletion reduced the anti-tumor efficacy of MPLA+IFNγ (Figure S6C, S6D), but we were unable to determine the contribution of macrophages, as the anti-CSF1R antibody was ineffective in depleting them (Figure S6A). In 4T1 tumors, depleting either macrophages or T cells was effective (Figure S6B), and depleting either cell type abolished the suppressive effects of MPLA+IFNγ on both tumor growth and lung metastasis (Figure 6A, 6B). Of the two T cell populations, CD8⁺ T cells were required for MPLA+IFNγ to inhibit tumor growth and lung metastasis, while CD4⁺ T cells were required for tumor growth inhibition, but did not significantly affect lung metastasis (p=0.071). Thus, both the innate and adaptive immune systems are required for the anti-tumor effects of MPLA+IFNγ.

Figure 6. Both macrophages and T cells are essential for MPLA+IFNγ’s anti-tumor effects.

(A) Tumor growth curve of 4T1 tumors after T cell depletion (anti-CD4, anti-CD8, or anti-CD4/CD8) or macrophage depletion (anti-CSF1R). N=8 tumors/group. Immune cell-depleting antibodies were injected 1 day before intratumoral injection of MPLA+IFNγ. Welch’s ANOVA was performed for the tumor volume analysis at the end point. *p<0.05, **p<0.01, ***p<0.001, compared to MPLA+IFNγ+IgG. (B) Lung metastatic burden of 4T1 tumor-bearing mice was determined by H&E staining. Lung tissues were collected 3 days after the last MPLA+IFNγ treatment. N=5–7 mice/group. (C) Procedure for tumor re-challenge assay with or without T cell depletion (anti-CD4/CD8). i.t.: intratumoral; i.v.: intravenous. (D–E) Lung metastasis formed by newly injected breast cancer cells (ECFP⁺) without (D) or with (E) T cell depletion was determined by IF staining of lung sections for ECFP. N=4–5 mice/group. Welch’s ANOVA was used to analyze the data in panels 6B, 6D, and 6E. Data are represented as mean ± SD. See also Figure S6.

To test if MPLA+IFNγ induces a long-lasting immune response, tumor-bearing mice had their tumors surgically removed, recovered for 2 weeks, and were then re-challenged with intravenously injected cancer cells from MMTV-PyMT; ACTB-ECFP mice (Figure 6C), allowing us to identify the new cancer cells by their expression of enhanced cyan fluorescent protein (ECFP). Only mice treated with MPLA+IFNγ had significantly reduced ECFP⁺ metastatic burden, not mice treated with MPLA or IFNγ alone (Figure 6D). When T cells were depleted during MPLA+IFNγ treatment (Figure 6C), the ability of MPLA+IFNγ to inhibit new metastasis was abolished (Figure 6E). These data, together with enhanced memory T cell generation (Figure 5D), indicate that the combination treatment not only reverses immune suppression but also triggers anti-tumor memory, which may have a long-lasting effect on metastasis.

MPLA+IFNγ suppresses metastatic ovarian cancer

Intraperitoneal MPLA+IFNγ’s effectiveness in our breast cancer model (Figure S2F, S5E) motivated us to explore this treatment for ovarian cancer, for which intraperitoneal chemotherapy is used and for which there are few effective treatments for late-stage disease. Furthermore, TAMs contribute to the immunosuppressive microenvironment in ovarian cancer (Rodriguez et al., 2018; Yigit et al., 2010). At 7 weeks after ID8-p53^−/− cancer cell transplantation (Figure 7A), the mice were visibly sick (Figure S7A) and had bloody ascites in their abdominal cavity (Figure 7B). All mice had an elevated body weight (Figure S7B) and many metastases in the peritoneum, pancreas, mesentery, and diaphragm (Table S1). Yet, when MPLA+IFNγ was initiated 3 weeks after cancer cell transplantation (Figure 7A), the mice at the 7-week time point looked healthy (Figure S7A), and only 2 of 5 mice had visible metastases in the peritoneum (Table S1). Moreover, MPLA+IFNγ decreased the total cell number (Figure 7C), increased the percentage of CD45⁺ immune cells (Figure S7C), and reduced the proportion of EpCAM⁺ cancer cells in the ascites (Figure 7D). Strikingly, MPLA+IFNγ prolonged the median survival of nD8-p53^−/− tumor-bearing mice from ~63 to ~100 days regardless of whether treatment was started 3 weeks (Figure 7E) or 6 weeks (Figure 7F) after tumor inoculation, when many mice already had a small amount of bloody ascites. Together, these findings indicate that MPLA+IFNγ suppressed the metastatic progression of ovarian cancer.

Figure 7. MPLA+IFNγ suppresses metastatic ovarian cancer in mice.

(A) Experimental design: three weeks after intraperitoneal ID8-p53^−/− ovarian cancer cell inoculation, MPLA mixed with IFNγ was injected to the mice every 5 days, for six times. Two days after the last treatment, mice were euthanized and ascites/peritoneal lavage was collected for flow cytometry, i.p.: intraperitoneal. (B) Metastasis was evident from the appearance of the ascites/peritoneal lavage of the treated mice. (C, D) Metastasis was determined by the total cell number (C) and percentage of epithelial cancer cells (EpCAM⁺, D) in the ascites/peritoneal lavage. (E) Kaplan-Meier survival curves of DD8-p53^−/− tumor-bearing mice treated with MPLA+IFNγ. Log-rank test, N=10 mice/group. Median survival times: Vehicle: 64 days; MPLA+IFNγ: 103 days. (F) Kaplan-Meier survival curves of DD8-p53^−/− tumor-bearing mice treated with MPLA+IFNγ alone or in combination with cisplatin as indicated. Log-rank test, N=10 mice/group. Median survival times: Vehicle: 63 days; MPLA+IFNγ: 98.5 days; cisplatin: 105 days; and cisplatin followed by MPLA+IFNγ: undefined. (G–I) The proportions of monocytes (Ly6C⁺Ly6G⁻, G), tumoricidal macrophages (iNOS⁺, H), and TAMs (CD206⁺, I) in ascites/peritoneal lavage were determined by flow cytometry. N=5 mice/group. T-test was used for panels 7G and 7I, and Welch’s t-test was used for 7C, 7D, and 7H. Data are represented as mean ± SD. See also Figure S7 and Table S1.

Most patients with ovarian cancer are treated with chemotherapy, but the cancer will often recur (Herzog, 2006). We thus compared MPLA+IFNγ to cisplatin, a widely used chemotherapy for ovarian cancer. Both treatments increased the median survival of the mice, from 63 days for vehicle to 98.5 and 105 days for MPLA+IFNγ and cisplatin, respectively (Figure 7F). When MPLA+IFNγ was given after four cycles of cisplatin, the median survival time could not be determined (Figure 7F), as 70% of the mice were still alive at the endpoint of 150 days.

We next investigated the effect of MPLA+IFNγ on the ovarian cancer immune microenvironment. MPLA+IFNγ increased the percentage of monocytes in the ascites (Figure 7G), and similar to the breast cancer model, it altered the balance between iNOS⁺and CD206⁺ macrophages (Figure 7H, 7I): the percentage of iNOS⁺ macrophages increased, while that of CD206⁺ macrophages decreased. To determine if macrophages and T cells were both important for the anti-metastatic effects of MPLA+IFNγ, we depleted macrophages using an anti-CSF1 antibody (DeNardo et al., 2011; Guerriero et al., 2017) (Figure S7D). Macrophages were essential for the anti-metastatic effects of MPLA+IFNγ in the ovarian cancer model, measured by the total cell number in the ascites of the mice (Figure S7E) or the proportion of EpCAM⁺ cancer cells (Figure S7F). Furthermore, both CD4⁺ and CD8⁺ T cells contributed to inhibiting metastasis in this model (Figure S7E, S7F). The median survival of ID8-p53^−/− tumor-bearing mice treated with MPLA+IFNγ (102 days) was longer than that of mice treated with TMP195 (51 days), IPI-549 (68 days), or PLX3397+anti-PD1 (70 days) (Figure S7G–I).

Altogether, our data show that MPLA+IFNγ dramatically suppresses metastatic breast and ovarian cancers by reprogramming the tumor-immunosuppressive microenvironment and stimulating the collaborative innate-adaptive anti-tumor immune response.

Discussion

Breast and ovarian tumors are infiltrated by large numbers of macrophages and T cells. Though these cells can kill cancer cells, they must be activated properly. With this study, we developed a strategy to reprogram the immune microenvironment of cancer. We revisited the approach used for Coley’s toxin and exploited the combined power of the TLR4 agonist MPLA with the cytokine IFNγ. The combination of MPLA and IFNγ reversed immune suppression in the TME and induced a systemic anti-tumor response. This treatment harnessed the power of macrophages and T cells, so that together, they eliminated metastatic breast and ovarian cancer. This result is in line with previous reports using other approaches to reprogram macrophages (Guerriero et al., 2017; Kaneda et al., 2016), strongly supporting the argument for engaging both innate and adaptive immunity for cancer immunotherapy.

MPLA+IFNγ stimulated innate immunity, but the adaptive immune response was also critical for the treatment’s in vivo effects. We found that MPLA+IFNγ did not stimulate T cells directly in vitro and that CD8⁺ T cell activation depended on macrophage-secreted cytokines, including IL-12 and TNFα. It is likely that MPLA+IFNγ in vivo also induces dendritic cells to produce IL-12 to activate cytotoxic T cell responses, as such effects have been reported (ten Brinke et al., 2010). Our data are consistent with reports that IL-12 bridges innate and adaptive immunity by enhancing the cytotoxic activity of CD8⁺ T cells (Trinchieri, 2003); that intratumor adoptive transfer of tumor-specific CD8⁺ T cells engineered with transiently expressed IL-12 mRNA eradicates tumors in melanoma mouse models (Etxeberria et al., 2019); and that genetically engineered myeloid cells expressing IL-12 activate antigen presentation and T cell response and reverse immune suppression to reduce metastasis in rhabdomyosarcoma and pancreatic tumor mouse models (Kaczanowska et al., 2021). TNFα can promote the activation and proliferation of naïve and effector T cells (Croft, 2009; Mehta et al., 2018). Macrophages from murine tumors and from malignant pleural effusions of breast cancer patients could be reprogrammed to be tumoricidal by MPLA+IFNγ ex vivo, suggesting that TAMs in the metastatic TME are sufficiently plastic to be reprogrammed to become tumoricidal. Moreover, the reprogrammed patient-derived macrophages expressed IL12B and TNF, suggesting that it will also be possible to activate both macrophage and T cell responses in cancer patients.

The collaborative therapeutic effect between MPLA and IFNγ in vivo is likely also related to MPLA+IFNγ’s induction of a robust type I IFN response in the tumors. Type I IFNs play critical roles in anticancer immunity [reviewed in (Parker et al., 2016; Vacchelli et al., 2015; Zitvogel et al., 2015)], and they can promote antigen presentation. This finding is consistent with our data showing increased antigen-presenting activities of macrophages and dendritic cells after combined MPLA+IFNγ treatment, but not by either treatment alone. Type I IFNs also skew the adaptive immune system to promote antigen-specific T cell responses and immunological memory. Indeed, MPLA+IFNγ significantly increased the abundance of activated cytotoxic T cells and effector memory T cells in tumors. In the metastasis re-challenge assay, MPLA+IFNγ–treated mice developed the fewest new metastatic lesions, and T cells were required for this effect, suggesting that MPLA+IFNγ triggers long-term anti-tumor T cell memory. Similar to our results, MPLA alone could not suppress murine melanoma or colon tumor growth, though MPLA does increase the anti-tumor responses of anti-PDL1 and anti-CTLA4 antibodies in the models (Jallad et al., 2020; Zhang et al., 2021). In situ anti-CD40+MPLA treatment activates local antigen-presenting cells, overcomes T cell exhaustion, and restores PD1 sensitivity in mouse models of melanoma, colorectal, hepatocellular, and bladder cancer (Khalil et al., 2019). Thus, MPLA combined with various other immunomodulatory drugs is a promising approach to improve anti-tumor immunity.

In our experiments, in situ injection of 1 μg MPLA (~50 μg/kg body weight) with 1–3 μg IFNγ (~50–150 μg/kg body weight) was sufficient to induce local immune modulation in breast tumors and to invoke a systemic anti-tumor immune response that protected against metastatic disease, either from spontaneously disseminated tumor cells or from cells injected as a re-challenge of the immune system. Intriguingly, under our experimental conditions, MPLA+IFNγ was more effective at suppressing metastasis in both breast and ovarian cancer models than other treatments that alter macrophage activation, namely TMP195, IPI-549, and PLX3397+anti-PD1. Along with our strategy, there are now several different approaches to activating the innate immune response to treat cancer. Potential drawbacks of invoking local immune modulation include the reliance on adequate immune cell infiltration available for reprogramming and the requirement of suitable injection sites in the tumor. As an alternative, we administrated MPLA+IFNγ intraperitoneally. Though this approach required higher doses (10–30 μg/ ~0.5–1.5 mg/kg body weight), we were reassured by the lack of overt toxicity.

MPLA and IFNγ are already FDA-approved biologies. If MPLA+IFNγ’s elimination of metastasis can be translated from our preclinical models to patients, then the approach may significantly improve the survival rates of late-stage breast and ovarian cancer patients.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Mikala Egeblad (egeblad@cshl.edu).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

RNA-seq data have been deposited in the Sequence Read Archive (SRA) with the accession number GSE146211 (also listed in the Key Resources Table).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-mouse iNOS	BD Biosciences	Cat #610431
Goat polyclonal anti-mouse CD206	R&D Systems	Cat #AF2535
Mouse monoclonal anti-β-actin	Santa Cruz Biotechnology	Cat #sc-47778
Rabbit monoclonal anti-NFκB p65	Cell Signaling Technology	Cat #8242
Rabbit monoclonal anti-Phospho-NFκB p65 (Ser536) (93H1)	Cell Signaling Technology	Cat #3033
Rabbit monoclonal anti-STAT1	Cell Signaling Technology	Cat #14994
Rabbit monoclonal anti-Phospho-STAT1 (Tyr701) (58D6)	Cell Signaling Technology	Cat #9167
Mouse monoclonal anti-GAPDH	Santa Cruz Biotechnology	Cat #sc-32233
IRDye 800CW Goat anti-Mouse IgG Secondary Antibody	LI-COR	Cat #926-32210
IRDye 800CW Donkey anti-Goat IgG Secondary Antibody	LI-COR	Cat #926-32214
Goat polyclonal anti-Neutrophil Elastase	Santa Cruz Biotechnology	Cat #sc-9521
BV785™ anti-mouse CD45 Antibody	BioLegend	Cat #103149
Alexa Fluor® 647 anti-mouse CD3 Antibody	BioLegend	Cat #100209
APC/Cy7 anti-mouse CD3 Antibody	BioLegend	Cat #100221
APC anti-mouse/human CD11b Antibody	BioLegend	Cat #101212
FITC anti-mouse CD4 Antibody	BioLegend	Cat #100510
PE anti-mouse CD8a Antibody	BioLegend	Cat #100708
APC/Cy7 anti-mouse CD8a Antibody	BioLegend	Cat #100714
PE/Cy5 anti-mouse CD19 Antibody	BioLegend	Cat #115509
PE/Cy7 anti-mouse CD11c Antibody	BioLegend	Cat #117318
APC/Cy7 anti-mouse CD11c Antibody	BioLegend	Cat #117324
APC/Cy7 anti-mouse Ly6C Antibody	BioLegend	Cat #128026
FITC anti-mouse Ly6G Antibody	BioLegend	Cat #127605
PerCP/Cy5.5 anti-mouse F4/80 Antibody	BioLegend	Cat #123127
Alexa Fluor® 488 anti-mouse F4/80 Antibody	BioLegend	Cat #123120
APC anti-mouse CD326 (EpCAM) Antibody	BioLegend	Cat #118214
PE/Cy7 anti-mouse CD40 Antibody	BioLegend	Cat #124621
FITC anti-mouse CD206 (MMR) Antibody	BioLegend	Cat #141704
FITC anti-mouse CD69 Antibody	BioLegend	Cat #104505
PE anti-mouse CD107a (LAMP-1) Antibody	BioLegend	Cat #121612
FITC anti-mouse CD279 (PD-1) Antibody	BioLegend	Cat #135213
PerCP/Cy5.5 anti-mouse/human CD44 Antibody	BioLegend	Cat #103032
APC anti-mouse CD62L Antibody	BioLegend	Cat #104412
APC anti-mouse H-2Kb bound to SIINFEKL Antibody	BioLegend	Cat #141605
Alexa Fluor® 647 anti-human CD326 (EpCAM) Antibody	BioLegend	Cat #324212
BV510™ anti-human CD45 Antibody	BioLegend	Cat #368525
APC anti-human CD68 Antibody	BioLegend	Cat #333810
PE Rat Anti-CD11b Clone M1/70 (RUO)	BD Biosciences	Cat #557397
CD335 (NKp46) Monoclonal Antibody (29A1.4), eFluor 450	Invitrogen	Cat #48335180
iNOS Monoclonal Antibody (CXNFT), PE	Invitrogen	Cat #12592082
CD103 (Integrin alpha E) Monoclonal Antibody (2E7), FITC	Invitrogen	Cat #11103182
PE anti-mouse CD40 Antibody	BioLegend	Cat #124610
FITC anti-mouse CD8a Antibody	BioLegend	Cat #100706
Rat monoclonal anti-PyMT antigen	Abcam	Cat #ab15085
Alexa Fluor 488 goat anti-Rat IgG secondary antibody	Invitrogen	Cat #A11006
Alexa Fluor® 647 anti-mouse CD4 Antibody	BioLegend	Cat #100424
Rat monoclonal anti-Mouse F4/80	Bio-Rad	Cat #MCA497R
Rabbit anti-Phospho-NFkB p65 (Ser276) Antibody	Invitrogen	Cat #PA5-37718
Rabbit polyclonal anti-CFP	Bio-Rad	Cat #AHP2986
Alexa Fluor 488 goat anti-rabbit IgG secondary antibody	Invitrogen	Cat #A11070
Goat polyclonal anti-Mouse IL-12	R&D Systems	Cat #AF-419-NA
Rat monoclonal anti-Mouse TNF-alpha	R&D Systems	Cat #MAB4101
Goat polyclonal anti-Mouse CXCL9/MIG	R&D Systems	Cat #AF-492-NA
InVivoMAb anti-mouse CD8α	BioXCell	Cat #BE0004-1
InVivoMAb anti-mouse CD4	BioXCell	Cat #BE0003-1
InVivoMAb anti-mouse CSF1R (CD115)	BioXCell	Cat #BE0213
InVivoMAb anti-mouse CSF1	BioXCell	Cat #BE0204
InVivoPlus anti-mouse PD1	BioXCell	Cat #BP0273
Biological Samples
Metastatic pleural effusions from breast cancer patients	New York University	IRB #i17-01382
Chemicals, Peptides, and Recombinant Proteins
Matrigel	Corning	Cat #356231
Human M-CSF	ProSpec	Cat #CYT-308
L-glutamine	Gibco	Cat #25030081
Sodium pyruvate	Gibco	Cat #11360070
2-mercaptoethanol	Gibco	Cat #31350010
B27 supplement	Gibco	Cat #17504044
Human EGF	ProSpec	Cat #CYT-217
Insulin	Gibco	Cat #12585014
Collagenase/hyaluronidase	STEMCELL Technologies	Cat #07912
DNase I	Roche	Cat #4716728001
Dispase	STEMCELL Technologies	Cat #07913
Collagenase D	Sigma	Cat #11088866001
DNase I	Sigma	Cat #045362820
Liberase DL	Sigma	Cat #05466202001
TrypLE Express Enzyme	Gibco	Cat #12605010
Mouse M-CSF	ProSpec	Cat #CYT-439
Fibronectin	Sigma Aldrich	Cat #F1141
CellTracker™ Deep Red Dye	Invitrogen	Cat #C34565
CellTracker™ Green CMFDA Dye	Invitrogen	Cat #C7025
LPS	Sigma Aldrich	Cat #L4391
MPLA for in vitro experiments	InvivoGen	Cat #tlrl-mpls
MPLA for in vivo experiments	InvivoGen	Cat #vac-mpls
PolyI:C	InvivoGen	Cat #tlrl-picw
Mouse IFNγ	R&D Systems	Cat #485-MI/CF
Human IFNγ	R&D Systems	Cat #285-IF/CF
TMP195	MedChemExpress	Cat #HY-18361
IPI-549	MedChemExpress	Cat #HY-100716
PLX3397 (Pexidartinib)	MedChemExpress	Cat #HY-16749
L-NIL (N6-(1-Iminoethyl)-L-lysine hydrochloride)	TOCRIS	Cat #1139
Creatine	Selleckchem	Cat #S5588
JSH-23	Selleckchem	Cat #S7351
SR11302	TOCRIS	Cat #2476
CellTiter 96® AQueous One Solution reagent	Promega	Cat #G3582
PowerUp™ SYBR™ Green Master Mix	Applied Biosystems	Cat #A25780
Taqman™ Universal Master Mix II, no UNG	Applied Biosystems	Cat #4440040
RIPA lysis buffer	Thermo Fisher Scientific	Cat #89990
Protease and phosphatase inhibitor	Thermo Fisher Scientific	Cat #78440
Cisplatin	MBL	Cat #JM-1550-1000
Sodium citrate solution	Sigma	Cat #C3821
ABTS Substrate Solution	Thermo Fisher Scientific	Cat #37615
Fc receptor blocker	Innovex	Cat #NB309
Goat serum	Dako	Cat #X0907
Mounting media	Electron Microscopy Sciences	Cat #17985-16
OVA-Q4H7 peptide	AnaSpec	Cat #AS-64405
Critical Commercial Assays
RNeasy Mini kit	Qiagen	Cat #74104
TaqMan™ Reverse Transcription Reagents	Invitrogen	Cat #N8080234
Cell Death Detection ELISA Kit	Roche	Cat #11774425001
Proteome Profiler Mouse XL Cytokine Array Kit	R&D Systems	Cat #ARY028
Mouse CXCL9/MIG DuoSet ELISA	R&D Systems	Cat #DY492
Mouse CCL5/RANTES DuoSet ELISA	R&D Systems	Cat #DY478
DuoSet ELISA Ancillary Reagent Kit 2	R&D Systems	Cat #DY008
TruSeq Stranded mRNA Library Prep Kit	Illumina	Cat #20020594
Fixation/Permeabilization Solution Kit	BD Biosciences	Cat #554714
Comprehensive Chemistry test of serum	IDEXX BioAnalytics	Cat #6006
CD8a+ T Cell Isolation Kit, mouse	Miltenyi Biotec	Cat #130-104-075
Deposited Data
RNA-seq data	This paper	GEO: GSE146211
Experimental Models: Cell Lines
Mouse: 4T1 cells	ATCC	Cat #CRL-2539
Mouse: ID8-p53−/− cells	Walton et al., 2016	N/A
Experimental Models: Organisms/strains
MMTV-PyMT mice (on C57BL/6 background)	Jackson Laboratory	Cat #022974
MMTV-PyMT mice (on FVB/N background)	Jackson Laboratory	Cat #002374
MMTV-PyMT-chOVA mice (on C57BL/6 background)	Engelhardt et al., 2012	N/A
ACTB-ECFP mice	Jackson Laboratory	Cat #004218
OT1 mice	Jackson Laboratory	Cat #003831
Tlr4 KO mice (on C57BL/6 background)	Jackson Laboratory	Cat #029015
MMTV-PyMT; ACTB-ECFP mice (on C57BL/6 background)	This paper	N/A
Oligonucleotides
Primers for human CD40, ACTB, NOS2, TNF, MRC1, IL12B, IL10, mouse Stat1, Stat2, Ifit1, Irgm2, Gbp2, Cxcl9, Igfbp3, Col4a1, Actb, Tnf, Il12b, Il10, Nos2, Mrc1, Arg1, see Table S2	This paper	N/A
Taqman Probe for mouse Tbp	Thermo Fisher Scientific	Mm00446973-m1
Taqman Probe for mouse Cd86	Thermo Fisher Scientific	Mm00444543-m1
Taqman Probe for mouse H2-Q2	Thermo Fisher Scientific	Mm04213368-g1
Software and Algorithms
STAR	Dobin et al., 2012	N/A
FeatureCounts	Liao et al., 2013	N/A
DESeq2	Love et al., 2014	N/A
Other
CD326 (EpCAM) microbeads, human	Miltenyi Biotec	Cat #130-061-101
CD11b MicroBeads, human and mouse	Miltenyi Biotec	Cat #130-049-601
24-well glass bottom plate	MatTek	Cat #P24G-1.5-13-F
PVDF membrane	Bio-Rad	Cat #1620177
Relfex 7 mm wound clips	CellPoint Scientific	Cat #203-1000
96-well Enzyme ImmunoAssay/Radio Immuno-Assay (EIA/RIA) plates	Costar	Cat #3590
Not TC-treated culture dishes	Corning	Cat #430597
Cell culture insert for 24-well Plate with 1.0 μm Transparent PET Membrane	Corning	Cat #353104
Rodent Chow 5053 With PLX3397 Diet	Research Diets, Inc.	N/A
Rodent Chow 5053 With IPI-549 Diet	Research Diets, Inc.	N/A

EXPERIMENTAL MODEL AND SUBJECT DETAIL

Patient samples

Metastatic pleural effusions from three female breast cancer patients were collected during a standard therapeutic procedure used to palliate symptoms. Patient 1 was 34 years old, and the tumor was estrogen-receptor (ER) positive (⁺), human epidermal growth factor receptor (HER)-2 negative (⁻); patient 2 was 35 years old, and the tumor was ER⁺/HER2⁺; and patient 3 was 53 years old, and the tumor was ER⁺/HER2⁻. All research was performed in accordance with New York University and Cold Spring Harbor Laboratory Institutional Review Board guidelines and regulations, and written informed consent was obtained from all patients.

Animals

OT1 mice and TLR4 KO (knockout) mice (on C57BL/6 [referred to hereafter as BL/6] background) were purchased from Jackson Laboratory. MMTV-PyMT mice (on BL/6 or FVB/N background) were bred at CSHL. To obtain MMTV-PyMT; ACTB-ECFP mice, ACTB-ECFP mice (on BL/6 background, originally obtained from Jackson Laboratory) were intercrossed with MMTV-PyMT mice. MMTV-PyMT-chOVA mice (Engelhardt et al., 2012) were a kind gift from Dr. Mathew Krummel (University of California, San Francisco).

Six- to 8-week-old female BL/6 and BALB/c host mice were purchased from Jackson Laboratory and acclimated to the animal housing facility for one week prior to performing experiments. Animals were housed All animal experiments were conducted in accordance with procedures approved by the Institutional Animal Care and Use Committee at CSHL and the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.

Tumor mouse models

PyMT cancer cells for orthotopic transplantation were isolated from the primary tumors of MMTV-PyMT mice or MMTV-PyMT-chOVA mice (BL/6 background). The cancer cells were isolated and were injected immediately into the 4^th mammary fat pads of BL/6 mice (2×10⁵ cells/fat pad in 20 μl 1:1 PBS/Matrigel, 2 tumors per mouse) to obtain the transplanted PyMT tumor model. For 4T1 cells, 5×10⁴ cells/fat pad were transplanted into the 4^th mammary fat pads of BALB/c mice (2 tumors per mouse). ID8-p53^−/− ovarian cancer cells were cultured as previously described (Walton et al., 2016). A total of 5×10⁶ ID8-p53^−/− cells (in 200 μl PBS) were injected intraperitoneally into BL/6 mice.

METHOD DETAILS

Isolation of macrophages and tumor cells from pleural effusions of breast cancer patients

Pleural effusions were shipped on ice to CSHL, and cells were isolated less than 24 hours after sample collection, as follows: the effusions were spun down at 450 x g for 10 minutes to remove the liquid supernatant, and the cell pellet was washed with Dulbecco’s Phosphate-Buffered Saline (DPBS) twice, then resuspended in Ammonium-Chloride-Potassium (ACK) lysis buffer on ice for 5–10 minutes; an additional spin down at 300 x g for 10 minutes was performed to eliminate the lysed red blood cells. Then, CD326 (EpCAM) microbeads were applied to enrich for EpCAM⁺ and EpCAM⁻ cells, according to the manufacturer’s instructions. Macrophages were obtained from the EpCAM⁻ cells after 2 hours of adherence to the tissue culture dish and were cultured with 50 ng/ml human macrophage colony-stimulating factor (M-CSF) in RPMI-1640 containing 20% fetal bovine serum (FBS), 200 μM L-glutamine, 100 μM sodium pyruvate, and 50 μM 2-mercaptoethanol. EpCAM⁺ cells were cultured with Dulbecco’s Modified Eagle Medium (DMEM)-F12 containing 20% FBS, 2% B27, 20 ng/ml human EGF, and 4 μg/ml insulin.

Isolation and culturing of macrophages and cancer cells from primary breast tumors of mice

The primary tumors (6–8 mm in diameter) of MMTV-PyMT mice (on BL/6 background) were mechanically dissociated and digested with commercial collagenase/hyaluronidase solution containing DNase I (4 U/ml) at 37 °C for 45 minutes. After pulse centrifugation (450 x g), the supernatant (mostly immune cells) and cell pellets (mostly cancer cells) were collected. CD11b MicroBeads were used to enrich CD11b⁺ cells from the supernatant, and macrophages were obtained by allowing the CD11b⁺ cells to adhere to the tissue culture dish for 2 hours (non-adherent cells were removed by washing with PBS). Cancer cells were isolated from the tumor cell pellets and dissociated into single tumor cells with TrypLE Express Enzyme containing DNase I (4 U/ml) at 37 °C for 10 minutes.

Mouse macrophages were cultured with RPMI-1640 containing 20% FBS, 200 μM L-glutamine, 100 μM sodium pyruvate, 50 μM 2-mercaptoethanol, and 10 ng/ml mouse M-CSF. Cancer cells were cultured with DMEM + 10% FBS.

Isolation of bone marrow-derived macrophages

Macrophages were differentiated from the bone marrow of wild-type (WT) BL/6 mice and TLR4 KO mice. The bone marrow was cultured for 6 days in the mouse macrophage culture medium on Not TC-treated culture dishes. The attached cells at the end of this culture period are unpolarized macrophages (Marim et al., 2010).

In vitro macrophage and tumor cell co-culture assay

A 24-well glass bottom plate was coated with fibronectin (1:100). For the direct contact condition, macrophages were seeded (1.5×10⁵/well) to the plate and 2 hours later, labeled with CellTracker™ Deep Red Dye (1 μM). Tumor cells were labeled with CellTracker™ Green CMFDA (5-chloromethylfluorescein diacetate) Dye (1 μM) and seeded (5×10⁴/well) to the plate. From that point forward, the culture medium was changed from a macrophage culture medium to cancer cell medium. Two hours later, LPS (100 ng/ml), MPLA (a low-toxicity derivative of LPS; 100 ng/ml; this product is produced by chemical synthesis and contains 6 fatty acyl groups), polyI:C (1 μg/ml), mouse IFNγ (33 ng/ml) or human IFNγ (50 ng/ml), TMP195 (a class IIa HDAC inhibitor, 500nM), or IPI-549 (a PI3Kγ inhibitor, 1 μM) was added to the medium (time point: 0 hour), alone or in indicated combinations, and live cell imaging was performed with a Yokagawa spinning disk confocal microscope (Solamere Technology Group) and LiveCell Stage Top Incubation System (Pathology Devices) for 48 hours. To inhibit iNOS activity, L-NIL (N⁶-(1-Iminoethyl)-L-lysine hydrochloride, 200 μM) was added to the medium before MPLA+IFNγ. For the conditioned medium (CM) condition, macrophages were treated with MPLA+IFNγ for 24 hours before collection of the CM. Tumor cells were fluorescently labeled and seeded to the plate in DMEM + 10% FBS. Two hours later, the medium was replaced with the CM from MPLA+IFNγ-treated macrophages (time point: 0 hour). For the transwell condition, macrophages were seeded to the cell culture insert (pore size: 1 μm), and tumor cells were seeded to the lower glass bottom plate. Two hours later, MPLA+IFNγ was added to the medium of macrophages (time point: 0 hour). In all the conditions, cancer cell numbers at 0 hour and 48 hours were determined by counting the cells in 5 fields of view, and cell numbers at 48 hours relative to those at 0 hour were then calculated and reported.

MTS assay

PyMT tumor cells were seeded to a 96-well plate (5×10³/well) and treated with LPS (100 ng/ml), MPLA (100 ng/ml), or polyI:C (1 μg/ml) with or without mouse IFNγ (33 ng/ml) for 48 hours. Next, 20 μl CellTiter 96® AQueous One Solution reagent was added to 100 μl medium and incubated for 2 hours at 37 °C. Absorbance at 490 nm was measured using a SpectraMax MiniMax 300 Imaging Cytometer (Molecular Devices).

RNA extraction and quantitative real-time PCR

Total RNA from macrophages (cultured alone) or primary tumors was extracted using an RNeasy Mini kit, following the manufacturer’s instructions. Reverse transcription was performed with TaqMan™ Reverse Transcription Reagents. Quantitative real-time PCR was performed in a 384-well format on a QuantStudio 6-flex Instrument (Applied Biosystems) using PowerUp™ SYBR™ Green Master Mix or Taqman™ Universal Master Mix II. Relative quantitation was performed with the 2^(−ΔΔCT) method using ACTB, Actb, or Tbp expression for normalization. The PCR primers are listed in the Key Resources Table and Table S2.

Western blot

Macrophages were lysed by RIPA lysis buffer with protease and phosphatase inhibitors. Protein extracts were resolved through 8% SDS-PAGE, transferred to a polyvinylidene difluoride (PVDF) membrane, probed with antibodies against iNOS (1:1000), CD206 (1:1000), STAT1 (1:1000), p-STAT1 (Tyr701, 1:1000), NFκB p65 (1:1000), p-p65 (Ser536, 1:1000), or β-actin (1:2000), GAPDH (1:2000), and then with fluorescent secondary antibody (1:2000). Binding was visualized by an Odyssey® Classic Imaging System.

In vivo tumor experiments

For PyMT and 4T1 tumors, tumor size was measured by caliper, and tumor volume was calculated as volume = (length×width²)/2. Treatment was initiated when tumors reached 5–6 mm in size and was given twice a week: 1 μg MPLA/tumor and indicated doses of mouse IFNγ were injected into the tumors to optimize the doses by comparing their inhibitory effects on tumor growth. Because the growth of PyMT tumors treated with MPLA+0.33, 1, or 3 μg IFNγ was almost equally suppressed compared to controls (Figure S2A) and the growth of 4T1 tumors treated with MPLA+3 or 9 μg IFNγ was inhibited similarly (Figure S2B), we chose the following doses: 1 μg MPLA with 1 μg IFNγ per tumor (or 1 μg MPLA alone or 1 μg IFNγ alone) was used for the intratumoral injection of PyMT tumors, and 1 μg MPLA with 3 μg IFNγ per tumor (or 1 μg MPLA alone or 3 μg IFNγ alone) was used for the intratumoral injection of 4T1 tumors. For the spontaneous MMTV-PyMT tumors (BL/6 background and FVB/N background), treatment was initiated when the total tumor volume of each mouse was approximately 150 mm³, and tumors that had reached 5–6 mm were intratumorally injected with 1 μg MPLA+1 μg IFNγ twice a week. Total volumes of all tumors were calculated for each mouse and reported. For intraperitoneal administration to mice with PyMT tumors, 30 μg MPLA with 30 μg IFNγ per mouse were used. At the indicated time points (tumor size<2 cm), mice were euthanized in a CO₂ chamber before performing a cardiac perfusion with PBS, and tumors and lungs were removed for analysis.

For the toxicology analysis experiments (shown in Figure 2K–M), tumor-free C57BL/6 mice were intraperitoneally injected with 30 μg MPLA+30 μg IFNγ every 3 or 4 days, for six times total. The body weight of the mice was measured before each injection. Two days after the last treatment, the mice were euthanized. The serum was collected for Comprehensive Chemistry Test, and liver, lung, kidney, heart, spleen, and femur were collected for hematoxylin and eosin (H&E) staining.

For the tumor re-challenge experiments (shown in Figure 6C), transplanted PyMT tumors were treated with MPLA+IFNγ (intratumoral injection) four times. Two days after the last treatment, breast tumors were surgically removed by making a small incision at the skin near the site of the tumor. The tumor along with 4^th mammary fat pad was excised and the wound was closed using Relfex 7 mm wound clips. Two weeks after tumor removal, ECFP⁺ PyMT cancer cells were injected into the tail vein of the mice. The ECFP⁺ PyMT cancer cells were isolated from the primary tumors of MMTV-PyMT; ACTB-ECFP mice as described above, and 5×10⁵ cells (in 100 μl PBS) were injected into the tail vein of the mice whose tumors had been surgically removed. Seventeen days later, the lungs were removed for analysis.

At indicated times, 10 μg MPLA with 20 μg mouse IFNγ was administered intraperitoneally to the ID8-p53^−/− tumor-bearing mice (Figure 7A–D, every 5 days), for six times total. At the end of the experiment (2 days after last treatment), the mice were weighed and intraperitoneally injected with PBS, and the ascites/peritoneal lavage was collected (~12 ml collected per mouse) for cell counting and flow cytometry. Ascites/peritoneal lavage was spun down at 300 x g for 10 minutes, and the cell pellet was resuspended in ACK lysis buffer on ice for 5–10 minutes, then spun down at 300 x g for an additional 10 minutes to eliminate the red blood cells. The total cell numbers were determined by a CellDrop cell counter. For the survival experiments, cisplatin (10 μg/g body weight) was administrated intraperitoneally to the mice every week, for a total of four times. MPLA (10 μg) with mouse IFNγ (20 μg) was administered as indicated (Figure 7E, every 5 days; Figure 7F, every 3–4 days). Mice were euthanized when the presence of ascites was evident and affected the mobility of the mouse.

For the comparison experiments of different approaches (shown in Figure S2G–I and Figure S7G–I), MPLA+IFNγ or anti-PD 1 antibody (200 μg/mouse, intraperitoneal injection every 3 days, three times in total) was administered on days indicated by arrows in figures showing the experimental design (Figure S2G and Figure S7G). During the indicated period, TMP195 (50 mg/kg body weight, in 10% DMSO + 90% sesame oil) was administrated intraperitoneally to the mice every day; IPI-549 and PLX3397 (pexidartinib, a CSF1R inhibitor) were administered in the mouse chow. The chow was custom-made. To obtain estimated daily doses (IPI-549: 15 mg/kg body weight/day and PLX3399: 40 mg/kg body weight/day), we assumed a female mouse body weight of 20 g and a daily chow intake of 3.5 g.

Lung metastasis analysis

Lungs were placed in PBS in a vacuum desiccator for 2 hours, then fixed in 4% paraformaldehyde (PFA) for 4 hours, infiltrated with 30% sucrose in PBS for 48 hours, and embedded in OCT compound. The blocks were then placed in a custom-made cutting chamber with razor blade inserts every 2 mm, and the block was cut into 2 mm sections. These sectioned lung pieces were then re-embedded in fresh OCT and sent to the Histology Core at CSHL for cross-sectional cuts from the entire lung tissue and for H&E staining. The metastatic burden was calculated as the percentage of metastasis/lung area, determined using Aperio eSlide Capture Devices software (Leica Biosystems).

ELISA for NETs

Plasma samples were obtained by cardiac blood collection using a syringe with a 25 G needle containing 25 μl sodium citrate solution. Whole blood was centrifuged at 1,300 x g for 10 minutes at 4 °C, and the top plasma layer was collected. As in a previous study (Albrengues et al., 2018), 96-well Enzyme ImmunoAssay/Radio Immuno-Assay (EIA/RIA) plates were coated overnight at 4 °C with an anti-elastase antibody (1:250) in 15 mM Na₂CO₃ and 35 mM NaHCO₃ at pH 9.6. The next day, the wells were washed three times with PBS, blocked in 5% bovine serum albumin (BSA) for 2 hours at room temperature, and washed three times with PBS. Then, 50 μl plasma samples were added to the wells and incubated for 2 hours at room temperature on a shaker, and the plates were washed three times with a wash buffer (1% BSA, 0.05% Tween 20 in PBS). Next, anti-DNA-peroxidase conjugated antibody (1:50, part of Cell Death Detection ELISA Kit) in 1% BSA in PBS was added to the wells for 2 hours at room temperature, and the wells were washed five times with wash buffer before the addition of 2,2’-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) (ABTS). Optical density was read 40 minutes later at 405 nm using a SpectraMax MiniMax 300 Imaging Cytometer (Molecular Devices).

Cytokine array

Twenty-four hours after the second intratumoral injection, transplanted PyMT tumors were collected, weighed, and digested with commercial collagenase/hyaluronidase (25 mg tumor/1 ml enzyme solution) containing DNase I at 37 °C for 30–45 minutes. The supernatant that contained secreted cytokines was collected and spun down to remove cells and cell debris (12,000 x g, 30 minutes). Supernatant samples (1 ml) and a Proteome Profiler Mouse XL Cytokine Array Kit were used according to the manufacturer’s instructions. Films were scanned, and spots were analyzed using ImageJ software. The heatmap was generated with http://www2.heatmapper.ca/expression/.

ELISA of cytokines

The samples were collected in the same manner as for the cytokine array analysis. CXCL9 and CCL5 secreted by tumors were assayed using a DuoSet ELISA and ancillary reagent kit according to the manufacturer’s guidelines. Absorbance at 450 nm was measured using a SpectraMax MiniMax 300 Imaging Cytometer (Molecular Devices).

Sample and library preparation for RNA-seq

Twenty-four hours after the second intratumoral injection, transplanted PyMT tumors were collected and total RNA was extracted using an RNeasy Mini Kit. RNA-seq libraries were prepared using a TruSeq Stranded mRNA Library Prep Kit and sequenced using aNextSeq500 High Output Illumina platform with 76 base single end reads at the CSHL Next Generation Sequencing Core Facility. This procedure yielded an average of 33.3 million filtered reads per sample.

Flow cytometry

Two days after the last intratumoral or intraperitoneal injection, tumors were harvested and processed as previously described (Fein et al., 2020). Briefly, tumors were mechanically dissociated and digested with collagenase/hyaluronidase containing DNase I (4 U/ml) at 37 °C for 30–45 minutes, lysed with ACK lysis buffer, and filtered with a 40 μm cell strainer. Lungs were digested with Dispase (2.5 U/ml)/Collagenase D (0.1 mg/ml)/Liberase DL (0.2 mg/ml) containing DNase I (25 U/ml) at 37 °C for 30 minutes. For flow cytometry staining, cells (1×10⁶) were incubated with mouse Fc receptor blocker at 4 °C for 10 minutes, then stained with the appropriate antibodies to surface markers at 4 °C for 30 minutes in the dark, and/or fixed/permeabilized (Fixation/Permeabilization Solution Kit) and stained with intracellular antibodies at 4 °C for 30 minutes. The stained populations were analyzed using a Fortessa flow cytometer (BD Biosciences) and FlowJo software (BD Biosciences). The antibodies used are listed in detail in the Key Resources Table. The PyMT tumors in our model are derived from transplanted primary cells, and the percentages of CD8/CD4 and F4/80 out of CD45⁺ cells varied between different experiments, likely due to the diversity of the PyMT tumors used to generate donor cells in individual experiments. We compared treatments only between groups of mice transplanted with the same primary cell population.

Immunofluorescence staining

Tumors were fixed in 4% PFA, infiltrated with 30% sucrose in PBS, and embedded in OCT compound, except the samples for staining p-STAT1 and p-p65, which were embedded in paraffin. The lungs were embedded in OCT compound. Tissues were cryosectioned at the Histology Core at CSHL. OCT compound-embedded tissue sections were post-fixed with ice-cold methanol and acetone (1:1 ratio), rinsed with PBS, blocked with Fc receptor blocker, and incubated with 1 × blocking buffer (5% goat serum, 2.5% BSA, 0.1% Triton X-100 in PBS) (Albrengues et al., 2018). Tumor sections were then incubated with fluorochrome-conjugated antibody (1:100, Alexa Fluor 488 anti-mouse F4/80; or PE anti-mouse CD40; or PE anti-iNOS; or FITC anti-mouse CD8a) in 0.5 × blocking buffer overnight at 4 °C. Lung sections were incubated with anti-PyMT antigen antibody (1:100) overnight. The second day, tissue sections were counterstained with DAPI and mounted with mounting media, while lung sections were incubated with Alexa Fluor 488 goat anti-Rat secondary antibody (1:1500) for 45 minutes, then incubated with Alexa Fluor 647 anti-mouse CD4 (1:100) and PE anti-mouse CD8a (1:100) overnight. The third day, lung sections were counterstained with DAPI and mounted with mounting media. Images were obtained with a Leica SP8 confocal microscope. The percentage of F4/80, CD8a, CD40, or iNOS positive cells in tumor sections was determined using ImageJ.

For the staining of F4/80 with p-STAT1 or p-p65, paraffin sections were deparaffinized and rehydrated and antigen retrieval was carried out by boiling slides in Tris EDTA buffer (10 mM Tris Base, 1 mM EDTA, pH 9.0) for 10 min in a pressure cooker. The slides were blocked and incubated with anti-F4/80 antibody (1:100), and anti-p-STAT1 antibody (1:100) or anti-p-p65 antibody (1:50) overnight at 4°C. The second day, the sections were incubated with Alexa Fluor 488 goat anti-Rat secondary antibody (1:1500) and Alexa Fluor 568 goat anti-Rabbit secondary antibody (1:1500), and were counterstained with DAPT. The numbers of p-STAT1⁺F4/80⁺ cells and the numbers of p-p65⁺F4/80⁺ cells per field of view were determined.

For the tumor re-challenge assay, lung sections were stained with anti-ECFP antibody (1:100) and Alexa Fluor 488 goat anti-rabbit secondary antibody (1:1500). Images were scanned using an Aperio ScanScope® FL System (Leica Biosystems), and ECFP⁺ lung metastasis was determined using Aperio eSlide Capture Devices software (Leica Biosystems).

In vitro T cell activation experiments

CD8⁺ T cells were isolated from the spleens of transplanted PyMT tumor-bearing mice with a CD8a⁺ T Cell Isolation Kit, according to the manufacturer’s instructions. Macrophages were differentiated from the bone marrow of these mice. CD8⁺ T cells were cultured with or without macrophages under different experimental conditions, as indicated below and in the figure legends, in RPMI-1640 with 20% FBS, 200 μM L-glutamine, 100 μM sodium pyruvate, and 50 μM 2-mercaptoethanol. Macrophages were seeded 2 hours before T cells to allow for their attachment, and MPLA (100 ng/ml), mouse IFNγ (33 ng/ml), or the two together were added to the culture system immediately after the seeding of T cells. T cell activation was identified by CD69 expression using flow cytometry, similar to the experiments described above using cells isolated from tumors. To block the cytokines that activate CD8⁺ T cells, neutralizing antibodies against IL-12 (1 μg/ml), TNFα (5 μg/ml), or CXCL9 (6 μg/ml) were added to the co-culture system 30 minutes before MPLA+IFNγ.

For the antigen-presenting assay, CD8⁺ T cells were isolated from the spleens of OT1 mice with a CD8a⁺ T Cell Isolation Kit. Macrophages were derived from the bone marrow of PyMT tumor-bearing mice as above and treated with MPLA (100 ng/ml), IFNγ (33 ng/ml), or the two together for 16 hours, then incubated with OVA-Q4H7 peptide (10⁻⁹ M) for 2 hours. The pre-treated macrophages were washed with PBS and finally added to CD8⁺ T cells. CD69 expression of CD8⁺ T cells was identified by flow cytometry.

Immune cell depletion/neutralization experiments

Anti-CD4, anti-CD8, anti-CSFIR, or anti-CSF1 antibodies were injected intraperitoneally one day before treatment with MPLA+IFNγ. For the first depletion, 0.5 mg of anti-CD4, 0.5 mg of anti-CD8, 1 mg of anti-CSF1R, or 1 mg of anti-CSF1 were injected per mouse. For the subsequent depletions, 0.2 mg of anti-CD4, 0.2 mg of anti-CD8, 0.5 mg of anti-CSF1R, or 0.5 mg of anti-CSF1 were injected per mouse.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical methods

Prism software (version 8.4.1, GraphPad) was used to analyze the data. In survival experiments, Kaplan-Meier survival curves were plotted and a log rank test was performed. In all other experiments with only two groups, the unpaired t-test (equal variances) or unpaired Welch’s t-test (unequal variances) was used. One-way ANOVA (with equal variances) or Welch’s ANOVA (unequal variances) was performed for experiments with more than two groups. Bartlett’s test or Brown-Forsythe test was used to assess equality of variances among groups. When the ANOVA was significant, group means were compared as indicated in the figures, and p-values adjusted for multiple comparisons were obtained using the Holm-šidák procedure (equal variances) or Dunnett T3 procedure (unequal variances). All graphs present the mean, the error bars represent the standard deviation (SD), and p-values are presented in the figures. All p-values<0.05 were considered statistically significant. In all in vitro experiments, the average results were taken from at least three independent experiments or three patients. In all in vivo experiments, the results are representative of at least two independent experiments (the numbers of mice and statistical tests are listed in the figure legends). Mice were randomized before the administration of therapy.

Analysis of RNA-seq data

FASTQ sequencing files were processed and analyzed using the Galaxy server hosted by the CSHL Bioinformatics Core Facility. Reads were aligned to the GRCm38/mm10 reference mouse genome using STAR (Dobin et al., 2012). Gene read counts were determined using FeatureCounts (Liao et al., 2013). Differential gene expression analysis was performed using DESeq2 (Love et al., 2014), with a 0.1 false discovery rate (FDR)-adjusted p-value cutoff for significance. DESeq2-normalized read counts of the top genes that were uniquely changed in the MPLA+IFNγ group were visualized as a heatmap using http://www2.heatmapper.ca/expression/. Genes that were uniquely upregulated or downregulated in the MPLA+IFNγ group were used as the input for biological process gene ontology analysis (http://geneontology.org/).

Supplementary Material

1

Data file S1. List of upregulated and downregulated genes in the MPLA+IFNγ, MPLA, and IFNγ treatment groups, related to Figure 3. RNA-seq data were collected following the experimental design in Figure 3A. After the combined MPLA+INFγ treatment, 179 genes were upregulated and 36 genes were downregulated in PyMT tumors, compared to vehicle. After treatment with MPLA alone, 20 genes were upregulated and 5 genes were downregulated. After treatment with INFγ alone, 33 genes were upregulated and 15 genes were downregulated.

2

Movie S1. Un-activated breast tumor-derived macrophages do not kill cancer cells, related to Figure 1. Macrophages and PyMT tumor cells were isolated from the primary tumors of MMTV-PyMT mice, and macrophages were stained with CellTracker™ Deep Red Dye, while tumor cells were stained with CellTracker™ CMFDA Dye. Deep Red dye is falsely colored green for clarity. Time indicated is time after imaging was initiated.

3

Movie S2. MPLA+IFNγ activates breast tumor-derived macrophages to kill cancer cells, related to Figure 1. Macrophages and PyMT tumor cells were isolated from the primary tumors of MMTV-PyMT mice, and macrophages were stained with CellTracker™ Deep Red Dye, while tumor cells were stained with CellTracker™ CMFDA Dye. Deep Red dye is falsely colored green for clarity. MPLA with mouse IFNγ was added before imaging was initiated. Time indicated is time after imaging was initiated.

4

Supplemental Figures and Tables

Highlights.

1. MPLA+IFNγ repolarizes TAMs from mice and patients to tumoricidal macrophages.

2. MPLA+IFNγ activates cytotoxic T cells through macrophage-secreted IL-12 and TNFα.

3. MPLA+IFNγ suppresses tumor growth & metastasis in breast and ovarian cancer models.

4. MPLA+IFNγ enhances the response to chemotherapy in ovarian cancer.