Ferroptotic Neutrophils Induce Immunosuppression and Chemoresistance in Breast Cancer

Abstract

Inducing ferroptosis in tumor cells is emerging as a strategy for treating malignancies that are refractory to traditional treatment modalities. However, the consequences of ferroptosis of immune cells in the tumor microenvironment need to be better understood in order to realize the potential of this approach. In this study, we discovered that neutrophils in chemoresistant breast cancer are highly sensitive to ferroptosis. Reduction of the acyltransferase MOAT1 in chemoresistance-associated neutrophils induced phospholipid reprogramming, switching the preference from monounsaturated fatty acids to polyunsaturated fatty acids, which increased their susceptibility to ferroptosis. Ferroptotic neutrophils secreted PGE2, IDO, and oxidized lipids that suppressed the proliferation and cytotoxicity of antitumor CD8+ T cells. Furthermore, neutrophil ferroptosis was closely related to a distinct subset of IL1β+CXCL3+CD4+ (Fer-CD4) T lymphocytes, which were enriched in chemoresistant tumors. Fer-CD4 T cells orchestrated neutrophil ferroptosis by modulating MOAT1 expression via IL1β/IL1R1/NF-κB signaling. Moreover, Fer-CD4 T cells secreted CXCL3, IL8, and S100A9 to replenish the neutrophil pool in the tumor microenvironment. Ferroptotic neutrophils in turn fostered Fer-CD4 T-cell differentiation. In spontaneous tumorigenesis mouse models, targeting IL1β+ CD4+ T cells or IL1R1+ neutrophils broke the cross-talk, restraining neutrophil ferroptosis, enhancing antitumor immunity, and overcoming chemoresistance. Overall, these findings uncover the role of neutrophil ferroptosis in shaping the immune landscape and propose appealing targets for restoring immunosurveillance and chemosensitivity in breast cancer. Significance: In chemoresistant breast cancer, IL1β+CXCL3+CD4+ T cells mediate neutrophil ferroptosis that suppresses antitumor immunity, indicating that interfering with this intercellular cross-talk could be an attractive strategy to reverse chemoresistance.

Graphical abstract

Figure 1.

Neutrophils in chemoresistant breast cancer were prone to ferroptosis. A, Enriched signatures in upregulated and downregulated genes in chemoresistant vs. chemosensitive TINs. B, Relative cell viability of neutrophils from chemoresistant tumors in the presence of inhibitors for apoptosis, necrosis, or ferroptosis. zVAD, z-VAD-FMK; Nec-1, necrostatin-1; Fer-1, ferrostatin-1; Lip-1, liproxstatin-1; DFO, deferoxamine. C, Cell death of neutrophils from chemosensitive and chemoresistant patients in response to distinct dosages of RSL3. D, Representative electron microscopic images (left) and quantification (right) for mitochondria of chemosensitive and chemoresistant TINs. Top row scale bar, 2 μm. Bottom row scale bar, 200 nm. Arrows, shrunken mitochondria. E, Representative flow cytometric images of oxidized BODIPY dye (left) of chemosensitive and chemoresistant neutrophils and quantification (right) by the ratio of oxidized to nonoxidized BODIPY fluorescence. F, Representative fluorescent images for oxidized and nonoxidized lipids of chemosensitive and chemoresistant TINs. Scale bar, 20 μm. G and H, Representative immunofluorescent staining (H) and quantification (G) of MPO and 4-HNE in chemosensitive and chemoresistant breast cancer specimens. Arrows, 4-HNE⁺ neutrophils. Scale bar, 50 μm. CR, complete response; SD, stable disease. I, Kaplan–Meier survival curves for overall and disease-free survival of patients with breast cancer with low or high 4-HNE⁺ proportions of neutrophils. Results are represented as mean ± SD of n = 5 (A–C), n = 6 (D), n = 8 (E), or n = 468 (G) different patients. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by Student t test (C–E), two-sided one-way ANOVA with the Tukey test (B and G) or two-sided log-rank test (I). Independent experiments were performed on three different patients for F. CI, confidence interval.

Figure 2.

MBOAT1 reshaped the phospholipid profile and ferroptosis vulnerability of chemoresistant neutrophils. A and B, Relative abundance of various PE-MUFAs (A) and PE-AAs (B) of chemosensitive and chemoresistant TINs by lipidomic analysis normalized by different reference lipids in resistant (A) or sensitive patients (B). C and D, Expression of genes associated with lipid metabolism or ferroptosis defense by qRT-PCR normalized by different reference genes in sensitive patients (C) and heatmap for MBOAT1 by bulk RNA-seq of chemosensitive or chemoresistant TINs (D). E, Representative immunoblot for MBOAT1 level of chemosensitive and chemoresistant TINs. F, Cell death of progenitor-derived neutrophils conditioned by chemoresistant tumor with or without MBOAT1 OE in response to distinct dosages of RSL3. G, Lipid peroxidation of progenitor-derived neutrophils conditioned by sensitive or resistant tumors with or without MBOAT1 overexpression (OE) by BODIPY staining. Scale bar, 5 μm. H, Relative abundance of distinct PE-MUFAs (left) and PE-PUFAs (right) of chemoresistant tumor-conditioned progenitor-derived neutrophils with or without MBOAT1 OE normalized by different reference lipids. I, Cell death of progenitor-derived neutrophils conditioned by chemosensitive tumor with or without MBOAT1 knockout in response to distinct dosages of RSL3. Results are represented as mean ± SEM (A, B, and H) or SD (F and I) of n = 5 (A–D, F, H, and I) different patients; *, P < 0.05; **, P < 0.01; ***, P < 0.001 by Student t test (A–C, F, H, and I). Independent experiments were performed on three different patients (E and G) with quantification in Supplementary Fig. S2.

Figure 3.

Ferroptotic neutrophils were intense immune suppressors. A, Schematic of the experimental design where TINs were purified from chemosensitive and chemoresistant patients and cocultured with CTLs. B and C, Proliferation (B) and granzyme B production (C) of CTLs after coculturing with neutrophils from chemosensitive or chemoresistant tumors. D, Granzyme B production of CTLs after coculturing with chemoresistant neutrophils pretreated with distinct inhibitors. zVAD, z-VAD-FMK; Nec-1, necrostatin-1; Fer-1, ferrostatin-1. E, Representative immunofluorescent staining of CD8, granzyme B, MPO, and 4-HNE in chemosensitive and chemoresistant breast cancer specimens. Arrows, 4-HNE⁺ neutrophils. Scale bar, 50 μm. F, Correlation between granzyme B⁺ CD8 T cells and ferroptotic neutrophils in breast tumor specimens. G, Granzyme B production of CTLs upon exposure to different components of CM of chemosensitive and chemoresistant TINs. H, Granzyme B production of CTLs upon exposure to EV-depleted portion of chemoresistant-neutrophil CM with proteases, nucleases or delipidation, respectively. I, Heatmap showing the relative level of PGE₂, IDO, HGF, TGFβ, IL10, IL21, EBI3, IL6, IL17, and IL23 released by chemosensitive or chemoresistant TINs, relative to the medium concentration. J, Granzyme B production of CTLs after coculturing with chemoresistant tumor neutrophils pretreated with COXi and/or 1-MT. K–M, Proliferation (K), perforin and IFNγ production (L), and cytotoxicity against tumor cells (M) of CTLs after coculturing with chemoresistant neutrophils pretreated with COXi and/or 1-MT by flow cytometry. Results are mean ± SD of n = 5 (B–D and G–M) or n = 468 (F) different patients. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-sided one-way ANOVA with the Tukey test (B–D, H, and J–M), Student t test (G), or two-tailed Pearson correlation coefficient test (F). ns, nonsignificant.

Figure 4.

Tumor neutrophil ferroptosis was related to a distinctive subset of CD4 T cells enriched in chemoresistant tumors. A, Uniform Manifold Approximation and Projection (UMAP) visualization of tumor-infiltrating CD4 T cells from five chemosensitive and five chemoresistant breast cancers by scRNA-seq. B, Three-dimensional UMAP plot of CD4 T cells colored by chemosensitivity. C, The proportions of different subpopulations of CD4 T cells in sensitive and resistant tumors. D, Heatmap displaying scaled expression of discriminating genes for each cluster of CD4 T cells in scRNA-seq data. E, Heatmap for IL1B expression in CD4 TILs from nine chemosensitive and nine chemoresistant patients by bulk RNA-seq. F, Representative flow cytometry for IL1β and CXCL3 expression in chemosensitive and chemoresistant tumor-infiltrating CD4 T cells. G, Cell death ratio of peripheral neutrophils pretreated with the inhibitors for apoptosis, necrosis, or ferroptosis and cocultured with IL1β⁺CXCL3⁺-d or IL1β⁺CXCL3⁺ CD4 T cells. H, Representative flow cytometric images for cellular lipid peroxidation of peripheral neutrophils cocultured with different CD4 T cells. I, Proliferation of tumor-specific CTLs upon exposure to peripheral neutrophils pretreated with IL1β⁺CXCL3⁺ or IL1β⁺CXCL3⁺-d CD4 T cells. J, Representative immunofluorescence images (left) and quantification (right) for IL1β and CXCL3 expression in CD4 T cells in chemosensitive and chemoresistant breast cancer sections. Arrows, IL1β⁺CXCL3⁺ CD4 T cells. Scale bar, 50 μm. K, Correlation between IL1β⁺CXCL3⁺ CD4 T cells and ferroptotic neutrophils in breast tumor specimens. Results are represented as mean ± SD of n = 5 (A–D and G–I), n = 9 (E and F), or n = 468 (J) different patients; for K, n = 468 different patients; ***, P < 0.001 by Student t test (F), two-sided one-way ANOVA with the Tukey test (G, I, and J) or two-tailed Pearson correlation coefficient test (K). Quantification is shown in Supplementary Figs. S4 and S5 (F and H).

Figure 5.

Ferroptosis-inducing CD4 T cells predisposed tumor neutrophils to ferroptosis through IL1β/IL1R1/NF-κB signaling. A and B, Cell death ratio (A) and lipid peroxidation (B) of peripheral neutrophils in the noncontact system with Fer-CD4 T cells in the presence of distinct neutralizing antibodies. C, Representative immunofluorescent images for IL1R1 expression in neutrophils with or without different CD4 T-cell encounter. Scale bar, 3 μm. D, Cell death ratio of progenitor-derived neutrophils with or without IL1R1 silencing cocultured with Fer-CD4 T cells. E, Representative fluorescent images for oxidized and nonoxidized lipids of IL1R1^KD or vector-treated progenitor-derived neutrophils cocultured with Fer-CD4 T cells. Scale bar, 5 μm. F, Immunoblot for total and phosphorylated p65 in peripheral neutrophils cocultured with different CD4 T cells. G, Cell death ratio of peripheral neutrophils pretreated with inhibitors for p65, p38, or AKT signaling and cocultured with Fer-CD4 T cells. H, Representative immunoblot for total and phosphorylated p65, p38, and AKT of peripheral neutrophils cocultured with Fer-CD4 T cells with IL1β neutralizing antibody or IL1R1 antagonist (Anak) treatment. I and J, Representative fluorescent images for p65 distribution (I) and immunoblot for MBOAT1 (J) of peripheral neutrophils cocultured with Fer-CD4 T cells with IL1β neutralization or IL1R1 blockade. Anak, Anakinra. Scale bar, 5 μm. K, Kaplan–Meier survival curves for overall and disease-free survival of patients with breast cancer with low or high proportion of Fer-CD4 T cells. Results are represented as mean ± SD of n = 5 (A, B, D, and G) different patients; for K, n = 468 different patients. ***, P < 0.001 by two-sided one-way ANOVA with the Tukey test (A, D, and G) or two-sided log-rank test (K). For C, E, F, and H–J, independent experiments were performed on three different patients; for I, independent experiments were performed on five different patients, and quantification is shown in Supplementary Fig. S6.

Figure 6.

Cross-talk between neutrophils and Fer-CD4 T cells maintained extensive neutrophil ferroptosis. A, Gene Ontology (GO) terms associated with upregulated genes of C2_IL1B cluster in scRNA-seq. B, Gene set enrichment analysis of bulk RNA-seq revealed enrichment of neutrophil chemotaxis genes in chemoresistant CD4 TILs. C, Scheme of chemotaxis assays with Boyden transwell chambers. D, Representative images (left) and quantification (right) of chemotaxis assays for peripheral neutrophils toward CM of different CD4 TILs. Scale bar, 100 μm. E, Migration tracks of neutrophils in μ-slide chemotaxis experiments toward CM of non–Fer-CD4 or Fer-CD4 T cells. F, Representative fluorescent images of cytoskeleton staining (left) and quantification (right) of filopodium-like protrusions (FLP) of peripheral neutrophils in the presence of CM from non–Fer-CD4 or Fer-CD4 T cells. Scale bar, 5 μm. G, Ex vivo tumor slice migration assays for the recruitment of CFSE-labeled neutrophils into breast tumor slices with high or low density of Fer-CD4 T cells. Scale bar, 100 μm. H, Dot plot for the expression of various chemokines in C2_IL1B cells in scRNA-seq. I, ELISA for CXCL3, IL8, and S100A9 release of non–Fer-CD4 or Fer-CD4 T cells. J and K, Migration (J) and filopodium-like protrusions (K) of peripheral neutrophils in the presence of CM from Fer-CD4 T cells and different antibodies. L, Cell death of peripheral neutrophils in the presence of CM of Fer-CD4 T cells and different antibodies categorized by location (top or bottom chamber of transwell system). M, CXCL3 and IL1β protein level of naïve CD4 T cells by immunoblotting after coculturing with control or ferroptotic neutrophils, which were generated by engagement with non–Fer-CD4 or Fer-CD4 T cells, respectively. Results are represented as mean ± SD of n = 7 (D) or n = 5 (F and I–M) or n = 6 (G). *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-sided one-way ANOVA with the Tukey test (D, F, and J–L) or Student t test (G, I, and M). Representative images are shown in Supplementary Fig. S7 (J and K).

Figure 7.

Breaking the cross-talk between Fer-CD4 T cells and neutrophils boosted antitumor immunity and chemosensitivity. A, Tumor growth curves of wild-type or Cd4-Il1b^Δ MMTV-PyMT mice treated with PBS or ADM. B, Representative immunofluorescent images for TUNEL and CK in tumors with the indicated administrations. Scale bar, 50 μm. C, Representative flow cytometric plots for lipid peroxidation and CD71 expression of neutrophils from tumors of wild-type or Cd4-Il1b^Δ MMTV-PyMT mice with PBS or ADM treatment. D, Representative flow cytometric plots for granzyme B and perforin production of CD8⁺ T cells isolated from tumors of wild-type or Cd4-Il1b^Δ MMTV-PyMT mice with PBS or ADM treatment. E, Representative immunofluorescent images for CD8 and granzyme B in tumors with the indicated administrations. Arrows, CD8⁺ granzyme B⁺ cells. Scale bar, 50 μm. F, Tumor growth curves of wild-type or Mrp8-Il1r1^Δ MMTV-PyMT mice treated with ADM or PBS. G, Lipid peroxidation of neutrophils, perforin, and granzyme B production of CD8 TILs of wild-type or Mrp8-Il1r1^Δ MMTV-PyMT mice treated with ADM or PBS by flow cytometry. Results are represented as mean ± SD of n = 5 mice per group (A–G). *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-sided one-way ANOVA with Tukey test (A, D, F, and G). Quantification is shown in Supplementary Fig. S8 (B–E).