A synthetic metastatic niche reveals antitumor neutrophils drive breast cancer metastatic dormancy in the lungs

Abstract

Biomaterial scaffolds mimicking the environment in metastatic organs can deconstruct complex signals and facilitate the study of cancer progression and metastasis. Here we report that a subcutaneous scaffold implant in mouse models of metastatic breast cancer in female mice recruits lung-tropic circulating tumor cells yet suppresses their growth through potent in situ antitumor immunity. In contrast, the lung, the endogenous metastatic organ for these models, develops lethal metastases in aggressive breast cancer, with less aggressive tumor models developing dormant lungs suppressing tumor growth. Our study reveals multifaceted roles of neutrophils in regulating metastasis. Breast cancer-educated neutrophils infiltrate the scaffold implants and lungs, secreting the same signal to attract lung-tropic circulating tumor cells. Second, antitumor and pro-tumor neutrophils are selectively recruited to the dormant scaffolds and lungs, respectively, responding to distinct groups of chemoattractants to establish activated or suppressive immune environments that direct different fates of cancer cells.

Fig. 1. Subcutaneous scaffold implants and lungs in BALB/c mice with 4T1 metastatic breast cancer are infiltrated with Gr1⁺CD11b⁺Ly6G⁺ neutrophils that overproduce S100A8/A9 cytokines.

a The frequencies of immune cell subsets in scaffolds and lungs derived from tumor-free BALB/c mice and mice bearing 4T1 tumor. H healthy, D diseased. b Tumor cells spontaneously migrate from the primary tumor to scaffolds of mice bearing 4T1-Luc2-tdTomato tumor. Scaffold implants were analyzed in the middle stage of cancer. c In vivo extravasation of intracardially injected 4T1-Luc2-tdTomato cells to scaffolds and lungs of tumor-free mice and mice bearing non-fluorescent 4T1 tumor at different cancer stages. The humane endpoint of BALB/c mice bearing 4T1 tumor without treatment was ~28 days after orthotopic tumor inoculation due to lethal lung metastases. TF tumor-free, TB tumor-bearing. d In vitro extravasation of 4T1 cells to conditioned media prepared by cells derived from tumor-free mice or mice bearing 4T1 tumor. Experiments were performed in the Boyden chamber. 4T1 cells were stained with a green dye and then placed in the top chamber while the conditioned media was placed in the lower chamber. Bar = 20 µm. CM conditioned media, TEM trans-endothelial migration. e, f S100a8 and S100a9 genes are overexpressed in diseased tissues relative to healthy tissues. Gene expression in these tissues were analyzed by single-cell RNA sequencing (e) and OpenArray (f). FC fold change. g The abundance of S100A8/A9 cytokines in media conditioned by healthy or diseased tissues. h The abundance of S100A8/A9 cytokines in media conditioned by cell subsets derived from diseased scaffolds (top panel) and in vitro extravasation of 4T1 cells to these conditioned media (lower panel). Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 3 for (a), (g), (h, top panel), and n = 6 for (b), (c), (h, lower panel). Source data are provided as a Source Data file.

Fig. 2. Subcutaneous scaffold implants and lungs in BALB/c mice with 4T1 metastatic breast cancer employ S100A8/A9 cytokines to recruit tumor cells with lung metastasis gene signature from the circulation.

a Neutralizing S100A8/A9 in the conditioned media with antibodies decreased in vitro extravasation of 4T1 tumor cells to the conditioned media. Ab antibody, D-sca diseased scaffold. b, c Administration of anti-Gr1 antibodies to BALB/c mice bearing 4T1 tumor depleted the Gr1⁺CD11b⁺ myeloid cells (b), decreased the abundance of S100A8/A9 cytokines (b), and decreased the in vivo extravasation of intracardially injected 4T1-Luc2-tdTomato cells to diseased scaffolds and lungs (c). Antibodies were administrated at the intermediate cancer stage (marked red in the timeline). The abundance of 4T1-Luc2-tdTomato tumor cells in the tissues was measured by flow cytometry and IVIS the second day after fluorescent 4T1 tumor cells were administrated to BALB/c mice bearing non-fluorescent 4T1 tumor through intracardiac injection. d Expression of signature genes mediating breast cancer metastasis to lungs or bones in tumor cells transmigrating towards different cancer cell chemoattractants and conditioned media in vitro, and those colonizing diseased scaffolds and lungs in vivo. Gene signature names are to the left. Red text is overexpressed in the signature while blue text is under-expressed in the signature. 231BR MDA-MB-231BR, BM bone marrow. e Schematic illustration of how scaffolds and lungs recruit the same subpopulation of tumor cells from the circulation in mouse models with 4T1 metastatic breast cancer. Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 3 for (a), (b), and n = 6 for (c). Source data are provided as a Source Data file.

Fig. 3. The synthetic MNs (scaffold implants) eradicate tumor cells and drive metastatic dormancy, while the endogenous lungs promote metastatic outgrowth in BALB/c mice with 4T1 metastatic breast cancer.

a, b Different fates of 4T1-Luc2-tdTomato cells recruited to diseased scaffolds and lungs of BALB/c mice bearing 4T1 tumor in experimental (a) and spontaneous (b) metastasis models. In the experimental metastasis model, 4T1-Luc2-tdTomato cells were transfused to BALB/c mice bearing non-fluorescent 4T1 tumor through intracardiac injection, and then tdTomato⁺ cells in MNs were measured by flow cytometry at day 1 and day 12 after injection. In the spontaneous metastasis model, 4T1-Luc2-tdTomato cells were orthotopically inoculated to BALB/c mice, and then tdTomato⁺ cells in the MNs were quantified at days 14, 21, and 28 after tumor inoculation. c Antitumor and pro-tumor transcription factors are differentially overexpressed in 4T1 cells co-cultured with diseased scaffold and lung cells. Activities of transcription factors were measured by TRACER (TRanscriptional Activity CEll aRrays). d Apoptosis and in vitro proliferation of 4T1-Luc2-tdTomato cells co-cultured with cells derived from diseased scaffolds or lungs. e Lung metastases of BALB/c mice without or with scaffold implants at days 10, 15, and 20 after orthotopic inoculation of 4T1-Luc2-tdTomato, representing early, intermediate, and late stages of cancer, respectively. Mice were immediately imaged after surgically resecting the primary tumor at the desired time. Mice bearing scaffold implants were less likely to have detectable metastases than mice without implants (top panel, p value is from Fisher’s exact test). Mice bearing scaffolds had reduced luminescence localized to the lung in the day 15 and day 20 groups (lower panel, p values are from non-parametric Mann–Whitney test). Data are shown as Mean ± SEM. p values are from student’s two-tailed t test when not otherwise specified. n = 6 for (a), (b), n = 3 for (d, top and lower panels), and n = 8 for (e, lower panel). Source data are provided as a Source Data file.

Fig. 4. The synthetic MNs (scaffold implants) develop potent in situ antitumor immunity through biomaterial-induced inflammation in 4T1 metastatic breast cancer.

a–c Phenotypic characterization of immune cells in diseased scaffolds and lungs of BALB/c mice bearing 4T1 tumor on activation and polarization markers by flow cytometric (a), single-cell RNA sequencing (b), and qRT-PCR analyses (c). In single-cell RNA sequencing analysis, Cd86 (cyan) and Cd206 (red) are the markers for N1 and N2, respectively. Ngp and Mmp8 (dark green) are makers for immature neutrophils, while Cd24a and Il1b (blue) are makers for mature neutrophils^,. N neutrophil, M macrophage, PRF1 perforin, GzmB granzyme B. d–f Expression of the cellular senescence marker p16 proteins (d, Bar = 200 µm), production of cytokines and chemokines (e), and expression of genes associated with the senescence-associated secretory phenotype (SASP) (f) in diseased scaffolds and lungs of BALB/c mice bearing 4T1 tumor. Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 3 for (a), (e). Source data are provided as a Source Data file.

Fig. 5. CXCL1, CXCL2, and CXCL5 recruit N1-polarized neutrophils to the dormant synthetic MNs, while S100A8/A9 recruits N2-polarized neutrophils to the immunosuppressive lungs in BALB/c mice with 4T1 metastatic breast cancer.

a Phenotypic change of T cells in diseased scaffolds and lungs after lung- or scaffold-derived Gr1⁻ cells were separately incubated with niche-specific Gr1⁺ cells in vitro. b, c Differential gene expression and protein production in neutrophils derived from diseased scaffolds and lungs. d Signature genes for neutrophils derived from diseased scaffolds or lungs, and their expression in circulating neutrophils transmigrating towards conditioned media and neutrophil chemoattractants. TEM⁺ cells were isolated from Boyden chambers in vitro. e Expression of genes for neutrophil chemoattractants, osteopontin (OPN, Spp1 gene), and decorin (DCN, Dcn gene) proteins in cell subsets derived from diseased scaffolds and lungs. N neutrophil, M macrophage, EC endothelial cell, Fibro fibroblast. f, g Immune cell phenotypes (f) and protein production (g) in diseased lung cells after incubating with osteopontin and decorin proteins in vitro. h, i Differential gene expression and protein production in diseased scaffold and lung cells. j Expression of genes for type 1 interferon in diseased scaffolds compared to diseased lungs. k N1 neutrophil chemoattractants secreted by diseased lungs compared to those secreted by healthy lungs (top panel), and these chemoattractants secreted by diseased lungs of 4T1-bearing mice receiving anti-Gr1 antibodies compared to those secreted by the lungs of tumor-bearing mice without receiving antibodies (lower panel). l Schematic illustration of the selective recruitment of N1 and N2 neutrophils to scaffolds and lungs, respectively, in animal models of metastatic breast cancer responding to distinct groups of chemoattractants. Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 3 for (a), (f). Source data are provided as a Source Data file.

Fig. 6. N1 neutrophils and their chemoattractants (CXCL1, CXCL2, and CXCL5) are abundant in the synthetic MNs and endogenous lungs of BALB/c mice with 67NR or 4T07 non-metastatic breast cancer.

a The frequencies of Gr1⁺CD11b⁺Ly6G⁺ neutrophils and the expression of S100a8 and S100a9 genes in healthy tissues of tumor-free BALB/c mice and diseased tissues of mice bearing 67NR or 4T07 non-metastatic or 4T1 metastatic breast cancer tumor. Healthy tissues were the control. b In vivo extravasation of intracardially injected 4T1-Luc2-tdTomato cells to scaffolds and lungs of tumor-free mice and mice bearing non-fluorescent 67NR, 4T07, or 4T1 tumor. c, d The frequencies of IFNγ⁺ and IL-17A⁺ CD4⁺ T cells (c) and immune cells with pro-inflammatory and antitumor phenotypes (d) in diseased scaffolds and lungs of mice bearing 67NR, 4T07, or 4T1 tumors. Diseased tissues from 4T1-bearing mice were the control. e, f The proteins secreted by diseased tissues of mice bearing 67NR (e) or 4T07 (f) tumor compared to those secreted by diseased tissues of mice bearing 4T1 tumor. g The proteins secreted by diseased lungs of mice bearing 67NR, 4T07, or 4T1 tumor compared to those secreted by healthy lungs of tumor-free BALB/c mice. h 4T1-Luc2-tdTomato cells that were transfused to BALB/c mice bearing 67NR tumor through intracardiac injection were attracted to the lungs but failed to grow to a secondary tumor. i In vivo extravasation of intracardially injected N1 neutrophils to scaffolds and lungs of tumor-free BALB/c mice and mice bearing 67NR, 4T07, or 4T1 tumor. Gr1⁺ splenocytes subpopulation migrating towards CXCL1 in the Boyden chamber were isolated, and these neutrophils with verified N1 phenotypes were fluorescently labeled with DiI dye and then transfused into mice by intracardiac injection. Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 3 for (a), (c), (d), n = 6 for (b), (i), and n = 4 for (h). Source data are provided as a Source Data file.

Fig. 7. N1 neutrophils and their chemoattractants (CXCL1, CXCL2, and CXCL5) are abundant in the synthetic MNs and the endogenous lungs of C57 mice with EO771 weakly metastatic breast cancer.

a EO771-GFP tumor cells migrated from the primary site to the diseased scaffolds and lungs but failed to form macroscopic metastases in C57 mice. The humane endpoint of C57 mice bearing EO771-GFP tumor without treatment was ~40 days after orthotopic tumor inoculation due to the large size of the primary tumor. b, c The frequencies of Gr1⁺CD11b⁺Ly6G⁺ neutrophils (b) and the expression of S100a8 and S100a9 genes (c) in diseased scaffolds and lungs of C57 mice bearing EO771-GFP tumor compared to healthy tissues of tumor-free C57 mice. d The frequencies of immune cells with pro-inflammatory and antitumor phenotypes in scaffolds and lungs derived from mice bearing EO771-GFP tumor. e The proteins secreted by diseased lungs of C57 mice bearing EO771 tumor compared to those secreted by diseased lungs of BALB/c mice bearing 67NR, 4T07, or 4T1 tumor or those secreted by healthy lungs of tumor-free C57 mice. f The proteins secreted by healthy lungs of tumor-free C57 mice compared to those secreted by healthy lungs of tumor-free BALB/c mice. g Size of the EO771-GFP primary tumor in wild type (WT), Ifng^−/−, and Il17a/Il17f^−/− C57 mice at day 28 after orthotopic tumor inoculation. h, i The abundance of CXCL1, CXCL2, and CXCL5 proteins (h) and the frequencies of immune cells with pro-inflammatory and antitumor phenotypes (i) in the diseased lungs of wild type (WT), Ifng⁻, and Il17a/Il17f^−/− C57 mice bearing EO771-GFP tumor. j Schematic illustration of the selective recruitment of N1 and N2 neutrophils to the lungs of animal models of breast cancers with varying tumor aggressiveness responding to distinct groups of neutrophil chemoattractants, followed by creating an immunosuppressive or dormant environment in the lungs. Data are shown as Mean ± SEM and p values are from student’s two-tailed t test. n = 6 for (a), n = 3 for (b), (c), (d), (i), and n = 4 for (g). Source data are provided as a Source Data file.

Fig. 8. The N1-to-N2 chemoattractant ratio is a predictive biomarker of metastatic dormancy, low tumor aggressiveness, and survival advantage for breast cancer.

a A positive correlation of the CXCL1-to-S100A8 ratio in MNs with the potential of metastatic dormancy in animal models of breast cancer. Scaffolds and lungs derived from animal models of different breast cancers were compared in terms of the expression of Cxcl1 and S100a8 genes in unsorted niche cells and their ratios (left panel), the abundance of CXCL1 and S100A8 proteins in the conditioned media and their ratios (middle panel), and the frequencies of CXCL1⁺ and S100A8⁺ cells in unsorted niche cells and their ratios (right panel). b Kaplan–Meier curves displaying the survival probability of patients with grade 1, 2, or 3 breast cancers. p values are from Mantel-Cox test. c The values of the Cxcl1-to-S100a8 gene expression ratio in breast cancers with different grades. p values are from student’s two-tailed t test. d Kaplan–Meier curves displaying the survival probability of breast cancer patients with a low (black) or high (red) value of the expression ratio of genes for N1 chemoattractants (Cxcl1 or Cxcl2 or Cxcl5) to genes for N2 chemoattractants (S100a8 or S100a9) at the primary tumor site. HR: hazard ratio; n = number of patients with available clinical data. p values are from Mantel-Cox test. Data for (b), (c), and (d) was retrieved from an online database. Data are shown as Mean ± SEM. n = 3 for (a). In (b) and (c), n = 396 for grade 1, n = 1174 for grade 2, and n = 1286 for grade 3. Source data are provided as a Source Data file.