Invasive breast cancer reprograms early myeloid differentiation in the bone marrow to generate immunosuppressive neutrophils

Abstract

Expansion of myeloid cells associated with solid tumor development is a key contributor to neoplastic progression. Despite their clinical relevance, the mechanisms controlling myeloid cell production and activity in cancer remains poorly understood. Using a multistage mouse model of breast cancer, we show that production of atypical T cell-suppressive neutrophils occurs during early tumor progression, at the onset of malignant conversion, and that these cells preferentially accumulate in peripheral tissues but not in the primary tumor. Production of these cells results from activation of a myeloid differentiation program in bone marrow (BM) by a novel mechanism in which tumor-derived granulocyte-colony stimulating factor (G-CSF) directs expansion and differentiation of hematopoietic stem cells to skew hematopoiesis toward the myeloid lineage. Chronic skewing of myeloid production occurred in parallel to a decrease in erythropoiesis in BM in mice with progressive disease. Significantly, we reveal that prolonged G-CSF stimulation is both necessary and sufficient for the distinguishing characteristics of tumor-induced immunosuppressive neutrophils. These results demonstrate that prolonged G-CSF may be responsible for both the development and activity of immunosuppressive neutrophils in cancer.

Keywords: cancer; hematopoiesis; immunology; myeloid-derived suppressor cells; stem cell biology.

Fig. 1.

Activated, T cell-suppressive Ly6G⁺ neutrophils are the predominant myeloid cells to expand in peripheral tissues during early tumor progression. (A) Quantification of the frequency (% of total) of CD11b⁺Gr1⁺ cells in WT and PyMT mice (6–15 wk). P > 0.05 was not significant (ns). (B) Quantification of the frequency (% of total) of Ly6G⁺ or Ly6C^hi cells. Wright-Giemsa (WG) staining of FACS-sorted Ly6G⁺ cells from PyMT mice (Insets). (Scale bar, 10 μm.) (C) Representative histograms show CFSE fluorescence in unstimulated (gray-shaded) and CD3/CD28-stimulated (black line) CD4⁺ and CD8⁺ T cells. Values shown are the % of CD4⁺ or CD8⁺ T cells that proliferated (red bar) in the absence (served as the control) or presence of Ly6G⁺ cells from WT or PyMT spleen. Quantification of CD4⁺ and CD8⁺ T-cell proliferation in the absence (control) or presence of WT or PyMT Ly6G⁺ cells from the spleen. Values shown are the frequency of T cells that proliferated normalized to the control. (D) Superoxide production in Ly6G⁺ cells in blood was assessed using DHR and flow cytometry. FACS plots illustrate the frequency (% of total) of Ly6G⁺ cells and the histogram shows DHR fluorescence in Ly6G⁺ cells. Bar graphs summarize the median fluorescent intensity (MFI) of DHR in Ly6G⁺ cells. (E) Rb1 protein expression in total splenocytes (Left) and FACS-sorted Ly6G⁺ cells (Right) was assessed by Western blot. Data are representative of (A and B) six experiments at each time point (mean ± SEM, n = 8–15) or two experiments (WG staining, n = 2), (C) two experiments (mean ± SEM of four samples), (D) three experiments (mean ± SEM, n = 5–6), and (E) four experiments (blots for total splenocytes; n = 4–6 biological samples) and two experiments (blots for FACS-sorted Ly6G⁺ cells; n = 2 biological samples). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Fig. 2.

Breast cancer development results in profound remodeling of BM hematopoiesis. (A) Gating strategy to evaluate BrdU incorporation in Ly6G⁺ cells by flow cytometry. (B) Representative FACS plots of BrdU incorporation (2 and 24 h after i.p. injection) in Ly6G⁺ cells in BM and spleen of WT and PyMT FVB/n mice (14–15 wk). (C) Summary of the % of Ly6G⁺ cells that are BrdU⁺ in BM and spleen at 2 and 24 h. (D) Frequency (% of total) of CD11b⁺Gr1⁺ cells in BM of WT and PyMT FVB/n mice during tumor progression (8–15 wk). (E) Representative image of femurs and tibias from WT and PyMT FVB/n mice. (F) Schematic of myeloid differentiation. (G–I) The frequency and total cell numbers of MPs and HSPCs were evaluated in BM of WT and PyMT FVB/n mice (14–15 wk) by flow cytometry. (G) FACS plots illustrate the gating for MP, MPP, and HSC populations. Values shown are the frequency of cells within each parent gate. (H) Total cell numbers of MP subpopulations (CMP, GMP, and MEP). (I) Total cell numbers of HSCs and MPPs (MPP^F- and MPP^F+) in BM of WT and PyMT FVB/n mice (8–12 wk). Data are representative of (A–C) two experiments (mean ± SD, n = 3–4), (D) four experiments at each time point (mean ± SD, n = 10–12), (E) six experiments (n = 8–15), (H) three experiments (mean ± SD, n = 5–7), (I) three experiments (8 and 10 wk; mean ± SD, n = 5), two experiments (12 wk; mean ± SD, n = 6–7), and five experiments (14–15 weeks; mean ± SD, n = 5–9). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Fig. 3.

Tumor-derived G-CSF, not GM-CSF or M-CSF, increases systemically during early tumor progression. Concentrations of G-CSF, GM-CSF, M-CSF, CXCL1, CCL2, CCL3, and CCL4 in (A) serum from WT and PyMT FVB/n mice (6–15 wk) and conditioned medium collected from (B) primary PyMT C57BL/6 tumor cells and (C) FVB/n VO-PyMT cell line were determined using multiplex laser bead technology (Eve Technologies). Data are representative of (A) two or three experiments (mean ± SEM, n = 4–7) and (B and C) two experiments, each sample in duplicate (mean ± SD of four biological samples). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Fig. 4.

G-CSF is sufficient for the enhanced ROS activity and Rb1^low phenotype of tumor-induced Ly6G⁺ neutrophils. (A) Superoxide production was assessed by DHR fluorescence in Ly6G⁺ cells in blood from WT and isotype or anti–G-CSF–treated PyMT FVB/n mice (14–15 wk). Representative FACS plots illustrate the frequency (% of total) of Ly6G⁺ cells (Left), histogram illustrates DHR fluorescence in Ly6G⁺ cells (Center), and bar graphs summarize the MFI of DHR in Ly6G⁺ cells (Right). Multiplex laser capture technology (Eve Technologies) was used to measure the concentration of G-CSF in the serum from (B) WT, PyMT, and anti–G-CSF–treated PyMT FVB/n mice (14–15 wk) and (C) PBS control treated or G-CSF–stimulated (12 or 24 h) WT mice. (D) Superoxide production was assessed by DHR fluorescence in Ly6G⁺ cells in blood from WT, G-CSF–stimulated, and PyMT FVB/n mice (14–15 wk). Representative FACS plots illustrate the frequency (% of total) of Ly6G⁺ cells (Left), histogram illustrates DHR fluorescence in Ly6G⁺ cells (Center), and bar graphs summarize the MFI of DHR in Ly6G⁺ cells (Right). (E and F) Rb1 expression was assessed by Western blot in (E) total splenocytes and (F) total splenocytes (Upper) or FACS-sorted Ly6G⁺ cells from the spleen (Lower). WT or PyMT FVB/n mice (14–15 wk). (G) FACS plots illustrate the frequency (% of total) of Ly6G⁺ cells and histrograms illustrate intracellular Rb1 expression in Ly6G⁺ cells, assessed by flow cytometry. Data are representative of (A) two experiments (mean ± SEM, n = 3–4), (B) two experiments (mean ± SEM, n = 6), (C) one experiment (n = 2–4), (D) two experiments (mean ± SEM, n = 3–4), (E and F) two experiments (n = 2–3 biological samples), and (G) two experiments (n = 4). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Fig. 5.

Kinetics of G-CSF stimulation reveal G-CSF targets expansion of HSC and MPP populations in the BM. (A and B) BM, blood, spleen, lung, and MG (tumor) were harvested from WT FVB/n and PyMT mice (14–15 wk) or anti–G-CSF–treated mice. (A) Quantification of total BM cells (Left), cell numbers of Ly6G⁺ and Ly6C^hi cells (Center), and cell numbers of HSC, MPP^F+, and MPP^F- populations (Right) in BM. (B) Representative images of femurs and tibias (Left) and spleens (Right). (C–H) BM, spleen, and blood were harvested from WT and G-CSF–stimulated (0.5–5 consecutive days) FVB/n mice. (C) Summary of the frequency (% total) of CD11b⁺Gr1⁺ cells assessed by flow cytometry. (D) FACS plots illustrate the frequency (% of total) of Ly6G⁺ cells in BM, spleen, and blood (Left), histograms of Ly6G expression in Ly6G⁺ cells (Center), and total cell numbers of Ly6G⁺ cells (Right). (E) Cell numbers of CMPs, GMPs, and MEPs in BM. (F) Representative images of femurs and tibias. Cell numbers of (G) HSCs, MPPs^F+, and MPPs^F- in BM and (H) HSCs, MPPs^F+, and MPPs^F- in spleen. Data are representative of (A and B) three experiments (mean ± SEM, n = 3–7) and (C–H) two experiments (mean ± SEM, n = 3–10). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

Fig. 6.

Intrinsic G-CSF signaling is required in MPPs for altered myeloid differentiation in breast cancer. Mixed BM chimera (WT:G-CSF-R^−/− at 1:1) C57BL/6 mice were transplanted with C57BL/6 PyMT primary tumor cells to evaluate chimerism. (A) Schematic illustrating experimental approach. (B) Summary of the frequency (% of total) of Ly6G⁺ cells in lung, blood, spleen, and BM. (C) Cell numbers of CMP, GMP, MEP, HSC, MPP^F-, and MPP^F+ populations in BM. (D and E) Representative FACS plots illustrate chimerism (Left) and quantification of cell numbers (Right) of CMP, GMP, MEP, HSC, MPP^F-, and MPP^F+ populations in BM of tumor transplanted or sham-treated mixed BM chimera mice. (F) Model for the generation of activated, T cell-suppressive neutrophils in breast cancer. Tumor-derived G-CSF induces expansion of immature hematopoietic compartments, before the level of CMPs, resulting in reprogramming of the BM and expansion of activated, Rb1^low and T cell-suppressive neutrophils in peripheral tissues. Data are representative of (A–E) four experiments (mean ± SEM, n = 8). *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.