Thymic stromal lymphopoietin fosters human breast tumor growth by promoting type 2 inflammation

Abstract

The human breast tumor microenvironment can display features of T helper type 2 (Th2) inflammation, and Th2 inflammation can promote tumor development. However, the molecular and cellular mechanisms contributing to Th2 inflammation in breast tumors remain unclear. Here, we show that human breast cancer cells produce thymic stromal lymphopoietin (TSLP). Breast tumor supernatants, in a TSLP-dependent manner, induce expression of OX40L on dendritic cells (DCs). OX40L(+) DCs are found in primary breast tumor infiltrates. OX40L(+) DCs drive development of inflammatory Th2 cells producing interleukin-13 and tumor necrosis factor in vitro. Antibodies neutralizing TSLP or OX40L inhibit breast tumor growth and interleukin-13 production in a xenograft model. Thus, breast cancer cell-derived TSLP contributes to the inflammatory Th2 microenvironment conducive to breast tumor development by inducing OX40L expression on DCs.

Figure 1.

Inflammatory Th2 in breast cancer immune environment. (A) Cytokine profiles as determined by Luminex in supernatants of human breast tumor fragments stimulated for 16 h with PMA and ionomycin. Numbers on the x-axis indicate the number of tissue samples from different patients tested. (B) Cytokine profiles as determined by Luminex in supernatants of tumor fragments (T) and surrounding tissue (ST) from the same patient after PMA and ionomycin stimulation. Cytokine concentration values of IL-2, IL-4, and IL-13 from T and ST samples were plotted and analyzed using linear regression to determine the level of correlation between cytokine concentration in T and ST samples. (C) Cytokine profiles as determined by Luminex in supernatants of tumor fragments after PMA and ionomycin stimulation. Cytokine concentration values of TNF and IL-13 and of TNF and IL-4 were plotted and analyzed using nonparametric Spearman correlation to determine the level of correlation of two cytokines concentration in tumor samples. (D, top) Single-cell suspensions from tumor samples were stimulated for 5 h with PMA and ionomycin. Cytokine production was measured by flow cytometry. Dot plots are gated on CD3⁺CD4⁺ T cells. (top right dot plot) Blue indicates gate on CD3⁺CD4⁺IL-13⁺ T cells that coexpress IFN-γ and TNF. Representative of four different patients from whom we have been able to obtain sufficient numbers of cells for 10-color analysis (patient nos. 148, 155, 164, and 169). Bottom, percentages of CD4⁺ T cells expressing IL-4 and IL-13 in tumor infiltrates and surrounding tissue (ST) were analyzed by flow cytometry. Dotted lines indicate paired samples from the same patient (n = 7, Wilcoxon matched-pairs ranked test). Single points indicate the percentage of cytokine expressing cells in tumor samples analyzed by flow cytometry for which we did not obtain sufficient number of cells from surrounding tissue to allow the analysis. (E) Frozen tissue sections from the same patient as in D were analyzed by immunofluorescence. Triple staining with anti-CD3-FITC (green), anti–IL-13–Texas red (red), and DAPI nuclear staining (blue). Bar, 90 µm.

Figure 2.

OX40L in breast cancer immune environment. (A) Immunofluorescence of primary breast tumor with indicated antibodies. Bar, 180 µm. Representative of 57/60 tumors analyzed. (B) Flow cytometry analysis of single-cell suspensions of primary breast tumors and surrounding tissue. Dot plots are gated on CD14^neg nonlymphocytes. OX40L expression is analyzed on HLA-DR^highCD11c^high DCs. Graph summarizes percentages of OX40L-expressing DCs in tumor infiltrates and surrounding tissue (ST) analyzed by flow cytometry. Dotted lines indicate paired samples from the same patient (Wilcoxon matched-pairs ranked test). Single points indicate the percentage of OX40L⁺ DCs in tumor samples for which we did not obtain sufficient number of cells from surrounding tissue to allow the analysis. (C and D) mDCs were exposed to media alone, to supernatant of breast cancer cell lines (1806 or Hs587T), or to sonicate of primary breast cancer tissue from patients (tumor 43). OX40L and CD83 were measured by flow cytometry. FMO, fluorescence minus one indicates controls where one staining fluorescence is omitted to set negative gate. (E and F) mDCs were exposed for 48 h to supernatants of breast cancer cells Hs578T, and then co-cultured with allogeneic naive CD4⁺ T cells in the presence of 40 µg/ml of anti-OX40L (Ik-5 clone) or isotype control antibody. After 1 wk, cells were collected and restimulated for 5 h with PMA/ionomycin for intracellular cytokine staining. Data in E are representative of four experiments. (F) Summary of the effect of blocking OX40L during T cell stimulation by tumor-activated DCs. Graph shows the proportion of IL-13–secreting cells induced by DCs activated with supernatants from breast cancer cell line Hs578T (left) or primary breast tumors (right, T15, T29, and T53).

Figure 3.

TSLP in breast cancer environment. (A) Luminex analysis of TSLP in supernatants of breast cancer cell lines after 24 h of culture in the presence of PMA and ionomycin. (B) Luminex analysis of TSLP levels in supernatants of primary breast tumors (from 44 patients) activated with PMA and ionomycin. (C) NOD/SCID/β2m^−/− mice were irradiated the day before tumor implantation and 10 × 10⁶ MDA-MB-231 cells were implanted by subcutaneous injection. Tumors were harvested at 4 wk after implant. Frozen tissue sections were analyzed by immunofluorescence for expression of TSLP (red). Actively dividing cells were identified by expression of Ki67 (green). Bar, 45 µm. (D and E) Frozen tissue sections from primary breast tumors from patients (38 patient samples) were analyzed by immunofluorescence for expression of TSLP. Tissues were also stained for the expression of IL-13 and cytokeratin 19, as indicated, to confirm TSLP expression by cancer cells. Staining pattern is representative of 35 out of 38 analyzed tumor samples from different patients. Bars: (D) 180 µm; (E) 15 µm. (F) NOD/SCID/β2m^−/− mice were sublethally irradiated and transplanted with human CD34⁺ HPCs by intravenous injection. 4 wk after HPC transplant, 5 × 10⁶ MDA-MB-231 breast cancer cells were implanted subcutaneously. Tumors at the site of implantation, as well as lungs and kidneys, were harvested at 3 mo after implant. Frozen tissue sections were analyzed by immunofluorescence for expression of TSLP (green) and cytokeratin (red). Staining pattern is representative of tumors from three different mice. Bar, 90 µm.

Figure 4.

Blocking TSLP in vitro. (A) mDCs were incubated with supernatant of breast cancer cell line Hs578T in the presence or absence of 20 µg/ml of anti-TSLP (AB 19024; rabbit IgG). OX40L expression was measured by flow cytometry after 48 h of incubation. (B) mDCs treated as in A were co-cultured with naive allogeneic CD4⁺ T cells for 7 d. IL-13 production was measured by intracellular cytokine staining and flow cytometry after cells were restimulated for 5 h with PMA and ionomycin. Data are representative of three experiments. (C) mDCs were incubated with soluble factors from sonicated human breast tumors (T53, T60, and T97) in the presence or absence of 20 µg/ml of anti-TSLP (AB 19024; rabbit IgG). OX40L expression was measured by flow cytometry after 48 h of incubation. (D) mDCs treated as in C were co-cultured with naive allogeneic CD4⁺ T cells for 7 d. IL-13 production was measured by intracellular cytokine staining and flow cytometry after cells were restimulated for 5 h with PMA and ionomycin. Representative of three patients tested. Blue dots represent IL-13⁺ T cells gated in the same sample. (E) Graph shows data from three independent experiments as described in A–D.

Figure 5.

Blocking TSLP-R in vitro. (A) mDCs were treated with anti-TSLP-R (clone AB81_85.1F11, mouse IgG1), media or control antibody during activation with TSLP or with supernatant of one of the three different breast cancer cell lines (Hs578T, MDA-MB-231, and MCF7). mDCs were then co-cultured with allogeneic naive CD4⁺ T cells. After 1 wk, cells were collected, restimulated for 5 h with PMA and ionomycin, and analyzed by flow cytometry. (B) Analysis of different experiments showing the effect of blocking TSLP-R on the induction of IL-13 secreting cells as described in A.

Figure 6.

Blocking OX40L-TSLP in vivo. (A) NOD/SCID/β2m^−/− mice were sublethally irradiated and transplanted with human CD34⁺ HPCs by intravenous injection. 4 wk after HPC transplant, 10 × 10⁶ Hs578T breast cancer cells were implanted subcutaneously. 3, 6, and 9 d after, mice were reconstituted with autologous total T cells together with 200 µg per injection of blocking anti-OX40L or isotype control antibody (mouse Ig; red arrows). PBS group was injected with tumor cells but not with T cells (gray line). Tumor size was measured at indicated times. Mean values from three experiments representing nine mice per each condition. Anti-OX40L and isotype-treated cohorts were compared statistically. (B) NOD/SCID/β2m^−/− mice were irradiated and implanted with 10 × 10⁶ Hs578T breast cancer cells together with 200 µg per injection of neutralizing anti-TSLP (rabbit), rabbit isotype control antibody, or PBS. 3, 6, and 9 d after, mice were reconstituted with immature DCs and autologous total T cells together with 200 µg per injection of neutralizing anti-TSLP (rabbit), rabbit isotype control antibody or PBS. Representative of three independent experiments with a total of nine mice in TSLP blockade group. (C) Cytokine secretion in single-cell suspensions from tumors after 16-h restimulation with PMA and ionomycin. (D) Same as in B, but mice were injected with anti-TSLPR or isotype control at days 3, 6, and 9. Representative of two independent experiments. n indicates number of mice per cohort in this representative experiment.