Inhibition of interferon-signalling halts cancer-associated fibroblast-dependent protection of breast cancer cells from chemotherapy

Abstract

Background: Triple negative breast cancers (TNBC) have poor prognoses despite aggressive treatment with cytotoxic chemotherapy. Cancer-associated fibroblasts (CAFs) are prominent in tumour stroma. Our hypothesis was that CAFs modulate chemotherapy sensitivity.

Methods: TNBC cells and breast fibroblasts were cultured; survival after chemotherapeutics was assessed using luciferase or clonogenic assays. Signalling was investigated using transcriptomics, reporters, recombinant proteins and blocking antibodies. Clinical relevance was investigated using immunohistochemistry.

Results: Breast CAFs dose-dependently protected TNBC cell lines MDA-MB-231 and MDA-MB-157, but not MDA-MB-468s, from chemotherapy. CAF-induced protection was associated with interferon (IFN) activation. CAFs were induced to express IFNβ1 by chemotherapy and TNBC co-culture, leading to paracrine activation in cancer cells. Recombinant IFNs were sufficient to protect MDA-MB-231 and MDA-MB-157 but not MDA-MB-468 cells. In TNBC patients, IFNβ1 expression in CAFs correlated with cancer cell expression of MX1, a marker of activated IFN signalling. High expression of IFNβ1 (CAFs) or MX1 (tumour cells) correlated with reduced survival after chemotherapy, especially in claudin-low tumours (which MDA-MB-231 and MDA-MB-157 cells represent). Antibodies that block IFN receptors reduced CAF-dependent chemoprotection.

Conclusions: CAF-induced activation of IFN signalling in claudin-low TNBCs results in chemoresistance. Inhibition of this pathway represents a novel method to improve breast cancer outcomes.

Fig. 1. Breast CAFs, but not NFs, protect some TNBC lines from chemotherapy.

a MDA-MB-231-luc cells were cultured alone (0% fibroblasts) or with increasing proportions of immortalised breast NFs (left panel) or CAFs (right panel). Cultures were treated with three different doses of epirubicin as shown, or with vehicle control for 24 h. Cultures were incubated for a further 48 h in fresh medium, before survival of MDA-MB-231 was assessed using luciferase assays. Data represent survival after epirubicin relative to matched vehicle control cultures, and are means (±SE) of three independent experimental replicates. b MDA-MB-231-GFP/luc (top panels) or MDA-MB-468-GFP cells (bottom panels) were cultured alone (0% fibroblasts) or with increasing proportions of immortalised breast NFs or CAFs. Cultures were treated with 10 nM epirubicin or vehicle control for 24 h. Epithelial cells were then collected by FACS and clonogenic survival was determined. Data are presented as colony counts (left panels) or relative survival after epirubicin (colony counts after epirubicin relative to matched untreated cultures; right panels). Data represent means (±SE) of three independent experimental repeats. c MDA-MB-231-GFP/luc cells were cultured alone (0% fibroblasts) or with increasing proportions of primary (p) breast NFs or CAFs cultured from a triple negative breast cancer resection. Cultures, cells and data were treated as for part B. Data represent means (±SD) of technical duplicates from one experimental repeat. Statistics: linear regression was carried out for analyses in A and B, with selected significant differences in the overall trend across the fibroblast proportions shown (ns not significant). ANOVA tests were performed in addition; these also demonstrated that CAFs provided significant protection from epirubicin in MDA-MB-231 cells (p < 0.01; lowest dose Fig. 1a and Fig. 1b right panel) and not in MDA-MB-468 cells.

Fig. 2. CAFs stimulate IFN signalling in some co-cultured breast epithelial cell lines.

a MDA-MB-231-GFP/luc cells were cultured alone (0%) or with CAFs (20%) and were treated with 10 nM epirubicin. Epithelial cells were then collected by FACS and RNA was prepared. Three separate biological repeats were performed giving three pairs of samples. Gene expression was assessed using Affymetrix Clariom D microarrays, and comparisons were made between 0% and 20% groups using hierarchical clustering. b MDA-MB-231-GFP/luc or MDA-MB-468-GFP cells were cultured on their own (0%) or in combination with CAFs (20%), with or without 10 nM epirubicin. Epithelial cells were collected by FACS and RNA was prepared. Relative expression of interferon response genes OAS1, MX1 and miR-155 was determined using qPCR. c MDA-MB-157 cells were cultured on their own (0%) or in combination with CAF-GFP cells (20%), with or without 10 nM epirubicin. Epithelial cells were collected by FACS and RNA was prepared. Relative expression of interferon response genes OAS1, MX1 and miR-155 was determined using qPCR. b, c Data represent the mean of technical triplicates (±SD) from one biological experiment, apart from miR-155 analysis in MDA-MB-231 cells, which is from three biological experiments (±SE) and is analysed using two-tailed Mann–Whitney U tests (selected significant difference shown).

Fig. 3. Epithelial:fibroblast crosstalk induces IFNβ1 expression in CAFs and IFN signalling in epithelial cells.

a MDA-MB-231-GFP/luc cells were cultured alone, breast CAFs were cultured alone, or co-cultures of MDA-MB-231-GFP/luc cells and CAFs were established (80% epithelial cells with 20% fibroblasts: “20%”). CAFs used were either immortalised, left, or primary, right. Cultures were treated with or without 10 nM epirubicin for 24 h. All cultures were processed for cell sorting, allowing separation of CAFs and MDA-MB-231-GFP/luc cells from the co-cultures on the basis of GFP fluorescence in the CAFs. RNA was extracted, and qPCR used to determine relative expression of IFNβ1. Data represent the mean of duplicate culture wells (±SD) for one biological experiment. ND not detected. b MDA-MB-231 cells were transfected with ISRE or GAS reporter plasmids driving firefly luciferase expression, and a control plasmid (pRL-TK; HSV thymidine kinase promoter driving renilla luciferase). Transfected MDA-MB-231 cells were then cultured on their own or with different proportions of immortalised NFs or CAFs for 24 h with 10 nM epirubicin. Dual luciferase assays were performed, with firefly readings normalised to renilla readings. Data represent means (±SD) for triplicate wells, for one biological experiment.

Fig. 4. Recombinant IFNs are sufficient to stimulate chemoresistance in MDA-MB-231 and MDA-MB-157 cells.

a MDA-MB-231-GFP/luc cells were cultured with a range of doses of IFNα or IFNγ for 24 h. Cells were then treated with 10 nM epirubicin, or control, and redosed with IFNs. Clongenic survival was determined as previously. b MDA-MB-231-GFP/luc, MDA-468-GFP or MDA-MB-157 cells were cultured with IFNα or IFNγ (2500 pg/ml and 1500 pg/ml, respectively) separately or combination or with appropriate isotype control antibodies for 24 h. Cells were then treated with epirubicin for 24 h (10 nM or, for MDA-MB-157s, 25 nM). Clongenic survival was determined as previously. a, b Data are shown as either colony counts (left), or survival relative to untreated (right) and represent means (±SE) of either 3 (a) or 2 (b) independent experiments. ANOVA tests were performed and selected differences are shown (ns not significant).

Fig. 5. In primary cancers, IFNβ1 in CAFs and MX1 in cancer cells correlate with each other and with poor survival.

TMAs of tissue from 109 TNBC resections were assembled and expression of IFNβ1 in fibroblasts, and MX1 and claudin-3 in tumour cells was determined using immunohistochemistry. a Representative images of immunohistochemistry, showing tissue scored ‘3’ for IFNβ in fibroblasts (left), ‘3’ for MX1, and ‘positive’ for claudin-3. b The cohort was split into groups with high or low expression of IFNβ1 in fibroblasts (left) or MX1 in tumour cells (right) using ROC analyses. Cumulative disease-free survival in the groups was compared using Kaplan-Meier analyses and log rank tests. c The cohort was split into claudin-low or claudin-high groups, based on expression levels of claudin-3 (positive or negative). The claudin-low group (n = 49) were analysed as in b.

Fig. 6. Blocking antibodies inhibit CAF-induced chemoprotection.

a, b MDA-MB-231-GFP/luc (left) or MDA-MB-157 (right) cells were cultured alone, or with breast CAFs or CAF-GFP cells, respectively. Cultures were treated with type I (1 μg/ml) or type II (5 μg/ml) interferon-signalling blocking antibodies or appropriate isotype controls for 24 h. Cultures were then treated with 10 nM (MDA-MB-231 cells) or 25 nM (MDA-MB-157 cells) epirubicin or vehicle control, and were redosed with antibodies for a further 24 h. Epithelial cells were then collected by FACS. a Clonogenic survival was determined. Data are presented as colony counts or relative survival after epirubicin (colony counts relative to untreated). b Relative expression of the marker of IFN-signalling activity MX1 was determined. c MDA-MB-231-GFP/luc cells were cultured alone, or with primary breast CAFs and were treated with antibodies and epirubicin/control exactly as above. Clonogenic survival was determined (left): data are presented as colony counts or relative survival after epirubicin (colony counts relative to untreated). Relative expression of the marker of IFN-signalling activity MX1 was also determined (right). a, b, c Data represent means (±SE) of three independent experimental repeats. Two-tailed Mann–Whitney U tests were carried out and selected differences are shown (ns not significant).