Tumor-derived miR-130b-3p induces cancer-associated fibroblast activation by targeting SPIN90 in luminal A breast cancer

Abstract

Cancer-associated fibroblasts (CAFs) in the tumor microenvironment (TME) interact closely with cancer cells to promote tumor development. Downregulation of SPIN90 in CAFs has been reported to facilitate breast cancer progression, but the underlying mechanism has not been elucidated. Here, we demonstrate that miR-130b-3p directly downregulates SPIN90 in stromal fibroblasts, leading to their differentiation into CAFs. As the decrease of SPIN90 in CAFs was shown to be more prominent in estrogen receptor (ER)-positive breast tumors in this study, miR-130b-3p was selected by bioinformatics analysis of data from patients with ER-positive breast cancer. Ectopic expression of miR-130b-3p in fibroblasts accelerated their differentiation to CAFs that promote cancer cell motility; this was associated with SPIN90 downregulation. We also found that miR-130b-3p was generated in luminal A-type cancer cells and activated fibroblasts after being secreted via exosomes from cancer cells. Finally, miR-130b-3p increased in SPIN90-downregulated tumor stroma of luminal A breast cancer patients and MCF7 cell-xenograft model mice. Our data demonstrate that miR-130b-3p is a key modulator that downregulates SPIN90 in breast CAFs. The inverse correlation between miR-130b-3p and SPIN90 in tumor stroma suggests that the miR-130b-3p/SPIN90 axis is clinically significant for CAF activation during breast cancer progression.

Fig. 1. SPIN90 is significantly downregulated in CAFs of ER-positive breast cancer patients.

A KM plots of recurrence-free survival in ER-positive (n = 2354) and ER-negative (n = 1080) breast cancer patients according to Spin90 level. Outlier data were removed for quality control. HR hazard ratio. B Representative IHC images showing expression of SPIN90 (green) and the epithelial cell marker E-cadherin (red) in breast cancers and paired normal stroma from ER-positive (n = 29) and ER-negative (n = 11) samples in a tumor microarray (BR804b). The white arrows indicate stromal fibroblasts. The graph shows the relative intensity score of SPIN90 in stroma as follows: 0 = negative, 1 = weak, 2 = moderate, 3 = strong. The images were scanned with an Olympus research slide scanner and analyzed using the Olympus cellSens and ImageJ software packages. Scale bar, 50 μm. All data are presented as mean ± standard deviation. *p ≤ 0.05; ***p ≤ 0.001 (Student’s t-test).

Fig. 2. miR-130b-3p inhibits SPIN90 expression by complementary binding.

A Venn diagram showing the process used to narrow down candidate SPIN90-targeting miRNAs through in silico and in vitro analyses. B Western blot indicating the expression of SPIN90 in HEK-293T cells transfected with miR-130b-3p mimic or negative control (n = 3). Relative protein levels were normalized to the expression of α-tub. C The Spin90 mRNA level was quantified by RT-qPCR assay in HEK-293T cells transfected with miR-130b-3p mimic or negative control (n = 3). The relative mRNA level was normalized to the expression of Gapdh. D Sequence information of the Spin90 mRNA indicating the predicted binding site (bolded sequences) for miR-130b-3p. E Reporter gene constructs. GFP-3U indicates a GFP vector inserted with the 3’UTR of the Spin90 mRNA, and GFP-3UM refers to a mutant construct lacking the binding site for miR-130b-3p. Bolded sequences, seed region. Red sequences, mutated region. F Reporter gene assay examined by Western blotting in HEK-293T cells co-transfected with reporter constructs and miRNAs (n = 3). GFP protein size depends on the location of the stop codon in each reporter gene construct. G Immunofluorescence assays of reporter gene expression in HEK-293T cells (n = 3). Scale bar, 50 μm. All data are reported as mean ± standard deviation. ***p ≤ 0.001 by Student’s t-tests. NC negative control of the miRNA mimic; 130b and 130b-3p, miR-130b-3p mimic.

Fig. 3. miR-130b-3p activates fibroblasts to exhibit CAF properties.

A Fibroblast activation ability of miR-130b-3p, as assessed by Western blot analysis. miR-130b-3p was transfected to HBFs and MRC5 cells at 25, 50, and 100 nM for 48 h, and histograms were normalized to the expression in the normal control (NC; black bar) of each group (n = 3). B Relative mRNA expression of Spin90 and myofibroblast markers, as examined by RT-qPCR (n = 3). C Recovery of SPIN90 expression in miR-130b-3p mimic transfected HBFs (n = 3). miRNAs were applied for 24 h, cells were transfected with HA or HA-SPIN90 DNA vector for 24 h, and lysates were collected. D Inhibition of the effect of miR-130b-3p on fibroblasts by transient expression of QIAGEN LNA miR-130b-3p power inhibitor. The oligonucleotides were co-transfected to HBFs with miRNA mimics for 48 h (n = 3). E Migration and invasion assay of MCF7 cells. CM from miRNA mimic-treated fibroblasts was added to the lower chamber. MCF7 cells were seeded in matrigel-coated or -uncoated upper chambers. Scale bar, 100 μm. All data are presented as mean ± standard deviation. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (Student’s t-test). NC negative control of miRNA mimic; 130b and 130b-3p, miR-130b-3p mimic; NC-i, inhibitor of negative control; 130b-i, inhibitor of miR-130b-3p.

Fig. 4. miR-130b-3p, which is secreted mostly from luminal A breast cancer, acts as a SPIN90 repressor in fibroblasts.

A Dicer knockdown in HBF. Dicer and SPIN90 expression were detected by Western blotting. B Representative images showing expression of miR-130b-3p (green) and the epithelial cell marker E-cadherin (red) in tumor and paired normal tissues from a commercially available tumor microarray (BR804b, Luminal A (n = 16), Luminal B (n = 13), Her2-enriched (n = 6), TNBC (n = 5)). Graph indicates log ratio for the intensity of miR-130b-3p in tumor tissues relative to normal tissues. Scale bar, 50 μm. C RT-qPCR analysis quantified the expression of mature miR-130b-3p in cancer cell lines of various luminal types, compared to that in normal MCF10A cells (n = 3). The level of each miRNA was normalized to that of RNU6. D Relative expression of miR-130b-3p in CM enriched from luminal A cancer and normal cell lines, normalized by the average Ct level of miR-23a-3p, miR-191-5p, miR-425-5p, and miR-451a (n = 3). E Western blotting analysis for SPIN90 and α-SMA in HBFs incubated for 48 h in the indicated CM (n = 3). F Relative amounts of miR-130b-3p in MCF7 CM treated with RNase in the presence or absence of Triton X-100 (n = 3 each). MCF7-derived CM was incubated with 10 μg/mL RNase or 1% (v/v) Triton X-100 at 37 °C for 30 min. G Western blot analysis of HBF cells incubated with MCF7 CM in the presence or absence of RNase and/or Triton X-100. The expression of each protein was normalized relative to that of α-tub. Buffer in MCF7 CM was replaced using a centrifugal filter to minimize Triton X-100 cytotoxicity. H HBFs were treated with CM from MCF7 cells stably expressing an oligonucleotide inhibitor for miR-130b-3p (130b-i) or a negative control oligonucleotide (NC-i) (n = 3). All data are presented as mean ± standard deviation. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (Student’s t-test).

Fig. 5. Expression pattern and role of miR-130b-3p in MCF7 xenografts of immune-deficient mice.

A Mammary fat pad injection of MCF7 cells or PBS (negative control). The graph shows tumor growth in the MCF7-injected group. B Mouse plasma was collected at weeks 2, 3, and 4 (each group, n = 4) after tumor cell injection. miR-130b-3p in plasma was quantified through RT-qPCR assay and normalized by the average Ct levels of miR-23a-3p, miR-191-5p, miR-425-5p, and miR-451a. C Tumors from the MCF7 group and normal fat pads from the PBS group were isolated at weeks 2, 3, and 4. Tissues were homogenized in QIAzol lysis reagent and the miR-130b-3p level was examined. miRNAs from tissues were normalized by the expression of RNU6. D, Co-staining of miR-130b-3p (green) and SPIN90 (red) and the fibroblast markers α-SMA for CAFs and Vimentin for normal fibroblasts (magenta) in tumor and adjacent normal tissues (n = 4 per group) from the MCF7 group. White dotted lines and arrows, stromal fibroblasts. The merged images show samples co-stained for miR-130b-3p and SPIN90, and with DAPI. Scale bar, 25 μm. E Expression pattern of miR-130b-3p and SPIN90 in tumor and normal tissues. The fluorescent intensity in the stroma was analyzed by the ImageJ software. F Representative IHC images of xenografted tumors derived from MCF7 NC-i (MCF7 cells stably expressing negative control for the inhibitor) and MCF7 130b-i (MCF7 cells stably expressing the oligonucleotide inhibitor of miR-130b-3p). Relative expression levels of α-SMA and SPIN90 of randomly selected ten areas were calculated using the ImageScope (Leica) software. Scale bar, 100 μm. G MCF7 cancer cells and HBF cells stably expressing miR-130b-3p inhibitor or negative control oligonucleotide were co-injected into the mammary fat pads of mice (n = 5). Tumor tissues were harvested 6 weeks later and weighed. All data are presented as mean ± standard deviation. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (Student’s t-test).

Fig. 6. Expression of miR-130b-3p in CAFs derived from human breast cancer tissue is inversely correlated with that of SPIN90.

A Expression of miR-130b in human luminal A breast cancer (n = 29) and normal tissues (n = 15) from the GSE45666 dataset. ***p ≤ 0.001 (Student’s t-test). B KM-plot survival analysis for ER-positive patients with respect to miR-130b expression. The data were retrieved from the TCGA (n = 451) and METABRIC (n = 996) datasets. HR hazard ratio. C Representative images of miR-130b-3p (red) and SPIN90 (green) staining in serial tissue sections from a tumor microarray (BR804b). Tumor and paired normal tissues from luminal A breast cancer patients (n = 16) were analyzed. The enlarged images at the upper right were not merged with DAPI. Cell boundaries are marked with white dotted lines. White arrows indicate stromal fibroblasts. Scale bar, 50 μm. D Correlation between the expression levels of miR-130b-3p and SPIN90 in the stroma. Each value was calculated as the log ratio of [intensity in tumor stroma] relative to [intensity in normal stroma] using ImageJ software.