Targeting stromal remodeling and cancer stem cell plasticity overcomes chemoresistance in triple negative breast cancer

Abstract

The cellular and molecular basis of stromal cell recruitment, activation and crosstalk in carcinomas is poorly understood, limiting the development of targeted anti-stromal therapies. In mouse models of triple negative breast cancer (TNBC), Hedgehog ligand produced by neoplastic cells reprograms cancer-associated fibroblasts (CAFs) to provide a supportive niche for the acquisition of a chemo-resistant, cancer stem cell (CSC) phenotype via FGF5 expression and production of fibrillar collagen. Stromal treatment of patient-derived xenografts with smoothened inhibitors (SMOi) downregulates CSC markers expression and sensitizes tumors to docetaxel, leading to markedly improved survival and reduced metastatic burden. In the phase I clinical trial EDALINE, 3 of 12 patients with metastatic TNBC derived clinical benefit from combination therapy with the SMOi Sonidegib and docetaxel chemotherapy, with one patient experiencing a complete response. These studies identify Hedgehog signaling to CAFs as a novel mediator of CSC plasticity and an exciting new therapeutic target in TNBC.

Conflict of interest statement

Novartis funded a part of the study. The authors declare no other competing interests.

Figures

Fig. 1

Malignant epithelial cells with increased self-renewal properties are located adjacent to the stroma of Hh-expressing cancers. a Scheme depicting the purification of epithelial and mixed stromal populations from disaggregated M6 murine tumor models. b Expression of genes significantly downregulated (blue) or upregulated (red) by RNA-Seq analysis (cut-off log2 (Fold Change) ≥ 2; vertical lines) in the epithelium of M6-Hh tumors compared to the epithelium of M6-Ctrl or M6-Hh tumors treated with the SMOi, GDC-0449 (100 mg/kg/bid), plotted against FDR values (horizontal lines indicate −log (FDR) > 2). Each symbol represents the transcriptome from five biological replicates per treatment group. c GSEA analysis reveals significant enrichment for genes encoding stemness and invasion in M6-Hh primary cells (FDR q value < 0.05). d Heat map showing relative CSC genes expression in the epithelium of M6-Hh tumors compared to M6-Ctrl and M6-Hh tumors + SMOi. Data show normalized row Z-score (n = 5 biological replicates for each treatment group). e Representative FACS dot plots showing the expression of CSC markers CD61 and CD29 within the EpCAM+/GFP+/CD24+ population of M6-Ctrl and M6-Hh tumors (n = 3 biological replicates per group). f Relative expression of key stemness genes in M6-Ctrl and M6-Hh whole tumors ± SMOi (100 mg/kg/bid). n = 3 biological replicates per treatment group; statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. g Kaplan−Meier and h tumor penetrance curves of mice injected with 250 primary M6-Ctrl (blue) or M6-Hh cells (violet); n = 10 biological replicates. Six tumors from the M6-Hh and two tumors from the M6-Ctrl models were detectable, respectively. Statistical significance was determined using log-rank test. i Primary M6-Ctrl and M6-Hh tumor cells were isolated by FACS and transplanted at various dilutions into recipient mice. Limiting dilution analysis demonstrating higher in vivo tumor-forming capacity in M6-Hh cells compared to M6-Ctrl cells. n = 10 mice per condition. j Representative images and quantification of CK6-progenitor and phospho-Histone H3-positive cancer cells at the tumor−stromal interface of M6 tumor models. Scale bars: 100 μm for CK6 and 200 μm for phospho-Histone H3. n = 3 biological replicates per treatment group. Statistical significance was determined using unpaired two-tailed Student’s t test; *P < 0.05; **P < 0.01. Bars represent mean ± s.e.m.

Fig. 2

Hh-activated stroma defines a novel transcriptional signature robustly associated with poor-prognosis in TNBC. a Expression of genes significantly downregulated (blue) or upregulated (red) (cut-off log2 (Fold Change) ≥ 2; vertical lines) in the stroma of M6-Hh tumors compared to the stroma of M6-Ctrl or M6-Hh tumors treated with SMOi (100 mg/kg/bid), plotted against FDR values (horizontal lines indicate −log (FDR) > 2). Each symbol represents the transcriptome of the stromal population from five biological replicates per treatment group. b RNA-Seq analysis reveals significant enrichment of GO groups related to ECM organization and production in M6-Hh tumor stroma (FDR q value < 0.05). c Heat maps demonstrating relative expression levels of gene-sets defined by RNA-Seq analysis. Genes highlighted in red represent those that are directly involved in ECM production. Data show normalized row Z-score (n = 5 biological replicates for each treatment group). d Kaplan−Meier curves of overall survival in unstratified breast cancer patients and in patients with basal, TNBC. Blue and red lines represent low and high Hh-stromal gene signature expression (HSGS), respectively. Statistical significance was determined using the Log-rank test; ***P < 0.001

Fig. 3

Paracrine Hedgehog signaling at cellular resolution. a Freshly isolated M6-Ctrl (blue), M6-Hh (magenta) and M6-Hh tumors + SMOi (100 mg/kg/bid; pink violin plot) were captured using 10x Chromium technology and the Cell Ranger Single Cell Software Suite 2.0 was used to perform demultiplexing, barcode processing, and single-cell 3′gene counting. t-SNE plot shows the subcellular clusters present in the breast TME of M6 tumors. Unless stated, P values are highly significant (Supplementary Data 2). b Single-cell RNA-Seq analysis reveals significant enrichment of GSEA groups related to ECM organization and production specifically in the CAF population of M6-Hh tumors (ECM-related processes highlighted in red; FDR q value < 0.05)

Fig. 4

High fibrillar collagen content resulting from Hh pathway activation promotes mechano-signaling and breast cancer stemness. a–d Concomitant expression analysis of collagen content and organization, integrin/focal adhesion activation and CSC-like characteristics at the tumor−stromal interface of M6 tumor models ± SMOi (100 mg/kg/bid; n = 3 biological replicates). a Representative multiphoton SHG imaging (scale bars: 100 μm) and quantitative analysis of collagen abundance. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. Corresponding graphs comparing fiber orientation (top right panel) and quantifying GLCM (bottom right panel). Unpaired two-tailed nonparametric Mann–Whitney U test was used for determining statistical significance across distributions. For GLCM analysis, statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. b Collagen I and III deposition detected and quantified by picrosirius red staining. Scale bars, 200 μm. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. c Representative immunofluorescence images and quantification of phospho-FAK and d CK6 expression. Scale bars, 100 μm. Statistical significance was determined using Kruskal−Wallis test. e M6-Ctrl and M6-Hh single cells were embedded and grown for 12 days in 3D Alginate IPNs containing increasing concentrations of collagen. Quantification of the number of colonies was normalized to the mean colony count in 0% collagen IPNs. n = 3 biological replicates with at least six technical replicates per condition. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. Representative phase contrast images of M6 colonies on polyacrylamide substrata after 12 days of culture. Scale bars: 100 μm. f Relative mRNA expression of the CSC markers Id3, Igtb3, and Krt6b in M6-Ctrl and M6-Hh cells cultured within 3D alginate-collagen I IPNs. n = 3 biological replicates with six technical replicates per experiment. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Bars represent mean ± s.e.m.

Fig 5

Hh-activated CAFs form a reversible, chemo-resistant CSC niche via FGF pathway activity. a RT-qPCR analysis of Fgf5 expression in M6 whole tumors. n = 3 biological replicates per treatment group with three technical replicates per assay. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. b Expression of Fgf5 at single-cell resolution in M6 tumor models. Freshly isolated M6-Ctrl (blue), M6-Hh (magenta), and M6-Hh tumors + SMOi (100 mg/kg/bid; violin dots) were captured using 10x Chromium technology. t-SNE plot represents the subcellular clusters present in the breast TME of M6 tumors. A sole subset of Hh-activated CAFs exhibits robust expression of Fgf5 in M6-Hh tumors. c Representative immunohistochemistry staining for phospho-FGFR on M6 tumors. Scale bars: 100 μm. Quantification of phospho-FGFR-positive cancer cells at the tumor−stromal interface. n = 4 biological replicates per treatment group; statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. d Relative mRNA expression of CSC markers Id3 and Sox10 in M6-Ctrl cells treated with DMSO (vehicle; blue) or recombinant FGF5 (red) in vitro. n = 3 biological replicates with three technical replicates per experiment; statistical significance was determined using unpaired two-tailed Student’s t test. e Primary and secondary tumorsphere formation of M6-Ctrl cells treated with DMSO (vehicle; blue) or recombinant FGF5 (red). Sphere Formation Efficiency (SFE) values in % are mean ± s.e.m.; n = 3 biological replicates with three technical replicates per tumorsphere assay. Representative phase contrast micrographs of M6-Ctrl spheres upon recombinant FGF5 stimulation. Scale bars: 100 μm. f Cell viability of M6-Ctrl and M6-Hh cells treated with indicated agents (n = 5 biological replicates with six technical replicates each). Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d.; *P < 0.05; **P < 0.01; ***P < 0.001. Bars represent mean ± s.e.m.

Fig. 6

Efficacy of long-term SMOi-combined therapy in preclinical models. a–c Concomitant expression analysis of collagen content and organization, integrin pathway activation and CSC-like characteristics at the tumor−stromal interface of the TNBC HCI-002 PDX model treated with vehicle (red) or the SMO inhibitor NVP-LDE225 (80 mg/kg/day; blue) (n = 5 biological replicates). a Representative imaging (scale bars: 100 μm) and quantitative analysis of collagen abundance (left panel). Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. Corresponding graphs comparing fiber orientation (middle) and quantifying GLCM analysis (right panel) in the vehicle and SMOi-treated models. Unpaired two-tailed nonparametric Mann–Whitney U test was used for determining statistical significance across distributions. b Collagen I and III deposition detected and quantified by picrosirius red staining. Scale bars, 200 μm. Statistical significance was determined using unpaired two-tailed Student’s t test with equal s.d. c Representative immunofluorescence images and quantification of concomitant phospho-FAK (green) and the human CSC marker ALDH1 (red) expression in the HCI-002 model. The white arrows illustrate examples of co-staining within the epithelial population. Scale bars, 100 μm. n = 3 biological replicates. Statistical significance was determined using Kruskal−Wallis test. d–f TNBC HCI-002 PDX model treated with vehicle (blue), SMO inhibitor (NVP-LDE225; 80 mg/kg/day; magenta), chemotherapy (docetaxel; 15 mg/kg/week; dark orange) or NVP-LDE225 (80 mg/kg/day) + docetaxel (15 mg/kg/week; orange line) (n = 7 mice per treatment group). d Tumor growth curves. Statistical significance was determined using unpaired Student’s t test. e Percentage of mice with detectable metastases in the lung, liver, and axillary lymph node in each treatment group. Statistical significance was determined using the Fisher’s exact test. f Kaplan−Meier curves of mice overall survival of each treatment group. Statistical significance was determined using Log-rank test of NVP-LDE225 + docetaxel vs. docetaxel; *P < 0.05; ***P < 0.001; ****P < 0.0001. Bars on the graphs represent mean ± s.e.m.

Fig. 7

Phase I clinical trial of docetaxel and SMO inhibitor, NVP-LDE225 (EDALINE) in patients with advanced TNBC. a Representative computed tomography (CT) images from a patient with complete radiological response. The 54-year-old postmenopausal woman was diagnosed with recurrent metastatic TNBC in the lungs with one measurable lesion on the right upper lobe (red arrow) and several nonmeasurable lesions (blue arrows). Therapeutic response was evaluated according to RECIST criteria version 1.1. The remaining structure seen in the right upper lobe corresponds to the azygos vein (violet arrow). b Representative HH and GLI1 immunostaining of treatment-naive tumor specimens from patients enrolled in the EDALINE trial. The left panel is representative of patients with high HPAS (characterized by high epithelial HH ligand and high stromal GLI1 expression) while the right panel represents low/intermediate epithelial HH and low stromal GLI1 expression. Scale bars: 100 μm. c Representative immunohistochemistry staining for phospho-FGFR, collagen deposition depicted by SHG imaging and concomitant phospho-FAK and ALDH1 stem cell marker expression in treatment-naive tumor specimens with high HPAS from the EDALINE trial. The left panel represents tissue derived from the patient who experienced a complete clinical response, the middle is from the patient with stable disease and the right panel represents tissue from the patient with high HPAS who progressed on the prescribed regimen. Scale bars: 100 μm. The white arrows illustrate examples of co-staining. d Graphical summary: Paracrine Hh signaling in TNBC drives a reversible stem-like, chemo-resistant phenotype via FGF signaling and ECM remodeling. CAFs represent the primary cells of the breast TME that respond to Hh ligand stimulation. Hh-activated CAFs enhance ECM collagen deposition and express FGF5 to establish a supportive niche for chemo-resistant cancer stem cell maintenance. This study strongly highlights a novel rational approach targeting both the tumor cells and their surrounding signaling support using SMO inhibitors to overcome chemoresistance in patients with TNBC