Tumor-derived miR-9-5p-loaded EVs regulate cholesterol homeostasis to promote breast cancer liver metastasis in mice

Abstract

Cancer cells secrete extracellular vesicles (EV) encapsulating bioactive cargoes to facilitate inter-organ communication in vivo and are emerging as critical mediators of tumor progression and metastasis, a condition which is often accompanied by a dysregulated cholesterol metabolism. Whether EVs are involved in the control of cholesterol homeostasis during tumor metastasis is still undefined and warrant further investigation. Here, we find that breast cancer-derived exosomal miR-9-5p induces the expression of HMGCR and CH25H, two enzymes involved in cholesterol synthesis and the conversion of 25-hydroxycholesterol from cholesterol by targeting INSIG1, INSIG2 and ATF3 genes in the liver. Notably, in vivo miR-9-5p antagomir treatment and genetic CH25H ablation prevents tumor metastasis in a mouse model of breast cancer. Thus, our findings reveal the regulatory mechanism of tumor-derived miR-9-5p in liver metastasis by linking oxysterol metabolism and Kupffer cell polarization, shedding light on future applications for cancer diagnosis and treatment.

Fig. 1. Breast cancer derived EV induced cholesterol biosynthesis in liver.

a Representative images for GFP signals in liver from mice xenografted MDA-MB-231/Lck-GFP cells (231/Ctrl) or MDA-MB-231/Rab27a KD/Lck-GFP cells (231/Rab27a KD) and tumor free mice. b Liver tissue collected from 231/Ctrl mice and 231/Rab27a KD mice for 50 days and tumor free mice were subjected to RNA-seq and GSEA, showing enrichment of genes related to indicated pathways (Permutation test, n = 3 mice per group). c Total cholesterol (TC) in liver and serum in indicated group (one-way ANOVA, n = 7 mice per group). d Liver sections were stained with oil red O. Scale bars, 50 μm. e Schematic of cholesterol synthesis pathway. f The qRT-PCR showing relative mRNA levels in the liver in indicated group (one-way ANOVA, n = 6 mice per group). g Western blots showing proteins related to cholesterol synthesis in the liver in indicated group. h IHC of indicated protein in liver sections. Representative images are shown. Scale bars, 100 μm. i GFP signals in liver from mice receiving tail vein injections of Lck-GFP-labeled EVs. j Liver tissue collected from mice that had received 5 weeks of i.v. injections of EVs or control mice were subjected to RNA-seq and GSEA, showing enrichment of genes set related to cholesterol biosynthesis pathway (Permutation test, n = 3 mice per group). k Total cholesterol (TC) in liver in indicated group (one-way ANOVA, n = 5 mice per group). l Left: Liver sections were stained with oil red O in NSG mice receiving tail vein injections of EVs. Scale bars, 50 μm. Right: Quantification of the lipid droplets of the liver (one-way ANOVA, n = 5 biological replicates). m Quantification of gene expression of several markers in cholesterol synthesis pathway using qRT-PCR in NSG mice receiving indicated EVs (one-way ANOVA, n = 6 mice per group). n Immunoblot analysis of liver sample collected from mice that had received MCF-10A or MDA-MB-231-derived EVs for 5 weeks. o Immunoblot analysis of proteins related to cholesterol synthesis in primary hepatocytes treated with indicated EVs, with PBS as control. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 2. Exosomal miR-9-5p induced cholesterol biosynthesis by targeting INSIG1/2.

a The enriched miRNAs from MDA-MB-231 EVs versus MCF-10A EVs overlapping with miRNA pools targeting both Insig1 and Insig2 from Targetscan database in cholesterol pathway. b Left: qRT-PCR-determined miR-9-5p levels in an equal amount of EVs derived from serum of normal (n = 6) and breast cancer patients (n = 12) (unpaired two-tailed t test). Right: The RNA quality of serum EVs was demonstrated. c RT-qPCR-determined miR-9-5p levels in liver (n = 5) and serum (n = 5) in indicated tumor-bearing or tumor-free mice (one-way ANOVA). d RT-qPCR-determined miR-9-5p levels in liver from mice receiving tail vein injections of indicated EVs (one-way ANOVA, n = 6 mice per group). e In situ hybridization for miR-9-5p in liver of mice from tumor free, 231/Ctrl and 231/Rab27a KD. Scale bar, 50 μm. f Predicted miR-9-5p binding sites in human and mouse Insigs genes (Insig1 and Inisg2, respectively). The corresponding sequences in WT and mutated reporters are shown. g Luciferase reporter assay showing responsiveness of the reporters to transfected human and mouse Insig1 or Insig2 in MCF-10A and MCF-10A/miR-9 cells (unpaired two-tailed t test, n = 3 independent batches). h Immunoblot analysis of liver sample collected from indicated mice received EVs injection. i, j Total cholesterol (TC) and triglycerides (TG) level in liver and serum in indicated mice received EVs (unpaired two-tailed t test, n = 6 mice per group). k Quantification of gene expression of HL7702 cell after EVs/PBS treatment for 48 h using qRT-PCR (one-way ANOVA, n = 3 independent batches). l Immunoblot analysis of HL7702 cell after EVs/PBS treatment for 48 h. m Immunoblot analysis of liver sample collected from indicated tumor bearing mice. n According to the gates in phasor plot, LD region was segmented in livers of indicated groups. Scale bars, 20 μm. o Total cholesterol (TC) and triglycerides (TG) level in liver (n = 6) and serum (n = 5) in indicated tumor bearing mice (unpaired two-tailed t test). p Liver sections were stained with oil red O in indicated groups. Scale bars, 50 μm. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 3. Circulating miR-9-5p promoted cholesterol to 25-HC conversion by targeting ATF3.

a Livers collected from mice xenografted with MDA-MB-231cells and 231/miR-9 KO cells were subjected to RNA-seq and GSEA, showing enrichment of genes related to oxysterol derived from cholesterol pathway (Permutation test, n = 3 mice per group). b Diagram of cholesterol conversion to oxysterol. c LC-MS/MS analysis of oxysterol derived from cholesterol in liver in indicated xenografted NSG mice (one-way ANOVA, n = 5 mice per group). d LC-MS/MS analysis of oxysterol derived from cholesterol in liver in indicated mice receiving tail vein injections of EVs (one-way ANOVA, n = 5 mice per group). e Quantification of gene expression in liver in indicated xenografted NSG mice by qRT-PCR (one-way ANOVA, n = 6 mice per group). f Quantification of gene expression in liver in indicated NSG mice receiving tail vein injections of EVs by qRT-PCR (one-way ANOVA, n = 6 mice per group). g Predicted miR-9-5p binding sites in human and mouse ATF3 genes. The corresponding sequences in WT and mutated reporters are shown. h Responsiveness of wild-type and mutant ATF3 3’UTR reporters to miR-9-5p in MCF-10A or MCF-10A/miR-9 cells. Luciferase reporter assay showing responsiveness of the reporters to transfected wild-type and mutant ATF3 3’UTR in MCF-10A and MCF-10A/miR-9 cell (unpaired two-tailed t test, n = 3 independent batches). i Quantification of gene expression in MCF-10A cell transfected with miR-9-5p mimic or control by qRT-PCR (one-way ANOVA, n = 3 independent batches). j Immunoblot analysis of MCF-10A cell transfected with miR-9-5p mimic or control. k Quantification of gene expression of primary hepatocytes after EVs/PBS treatment for 48 h by qRT-PCR (one-way ANOVA, n = 3 independent batches). l Immunoblot analysis of primary hepatocytes after EVs/PBS treatment for 48 h. m Immunoblot analysis of liver in indicated xenografted mice. n IHC of ATF3 and CH25H in liver sections in indicated xenografted mice. Scale bars, 100 μm. o Immunoblot analysis of liver in indicated mice received tail vein injections of EVs. p IHC of ATF3 and CH25H in liver sections in indicated mice received tail vein injections of EVs. Scale bars, 100 μm. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 4. Oncometabolite 25-HC promoted breast cancer liver and lung metastasis.

a Tumor volume followed in indicated xenografted mice (two-way ANOVA, n = 5 mice per group). b, c Representative bioluminescence images and quantification of mice inoculated with indicated group (one-way ANOVA, n = 5 mice per group). d, e Representative IHC images showing Ki67 staining and the overall percentage of Ki67⁺ tumor cells (one-way ANOVA, n = 5 mice per group). Scale bar, 50 μm. f Schematic diagram of 25-HC treated mouse model. 231/Ctrl or 231/miR-9 KO cells injected at day 0. On day 10, mice were treated with vehicle or 25-HC (10 mg/kg) every other day. At day 35, tissue was harvested, imaged and fluorescence quantified. Experimental timeline outlined above quantified data. g, h Representative bioluminescence images and quantification of mice inoculated with indicated groups (one-way ANOVA, n = 5 mice per group). i Quantification of liver nodules (one-way ANOVA, n = 5 mice per group). j 25-HC levels of livers in each group by ELISA kit. (one-way ANOVA, n = 6 mice per group). Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 5. Ch25h blockage inhibited BC liver and lung metastasis.

a LC-MS/MS analysis of 25-HC in serum in indicated CH25H liver conditional knock out (Ch25h^L–/–) and control (Ch25h^fl/fl) mice (unpaired two-tailed t test, n = 6 mice per group). b Total cholesterol (TC) and triglyceride (TG) levels were examined from CH25H liver conditional knock out (Ch25h^L–/–) and control (Ch25h^fl/fl) mice. (unpaired two-tailed t test, n = 5 mice per group). c Liver sections were stained with oil red O in Ch25h^–/– and Ch25h^fl/fl mice with E0771 xenograft. Scale bars, 100 μm. d Representative PET/CT imaging with ¹⁸F-FDG for accessing liver and lung uptake and quantification in indicated in CH25H liver specific KO and WT mice. e Representative bioluminescence images of metastasis site in liver and lung and quantification of mice inoculated with indicated groups (unpaired two-tailed t test, n = 5 mice per group). f Representative IHC images showing Ki67 staining and the overall percentage of Ki67⁺ tumor cells in indicated groups (unpaired two-tailed t test, n = 5 per group). Scale bar, 50 μm. g Immunoblot analysis of CH25H in liver in indicated groups (top), IHC and of CH25H in liver sections in indicated groups (bottom). Scale bar, 50 μm. h Representative bioluminescence images of metastasis site in liver and lung and quantification of mice inoculated with indicated groups (unpaired two-tailed t test, n = 5 per group). i Representative IHC images showing Ki67 staining and the overall percentage of Ki67⁺ tumor cells of liver and lung from indicated groups (unpaired two-tailed t test, n = 5 per group). Scale bar, 50 μm. j Representative PET/CT imaging with ¹⁸F-FDG for accessing liver and lung uptake and quantification in indicated mice. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 6. 25-HC induced Kupffer cells polarization into M2 type to promote liver metastasis.

a RAW264.7 cells treated with 25-HC and vehicle were subjected to RNA-seq and GSEA, showing enrichment of genes set related to macrophage pathway (Permutation test, n = 3 independent batches per group). b Quantification of purified Kupffer cells with CD11b⁺ and F4/80⁺ expression by flow cytometry in indicated groups. c IHC of F4/80 in liver and lung sections in indicated groups. Scale bar, 50 μm. d Representative bioluminescence images of liver and lung metastasis sites in indicated groups. e IHC of Ki67 in liver and lung sections in indicated groups. Scale bar, 50 μm. f Purified Kupffer cells were selected with TIM4⁺, I-A/I-E^int expression and stained and further analyzed for CD163 and CD11c expression. Representative flow cytometry results were shown. g Quantification of the M2/M1 ratio in flow cytometry results (one-way ANOVA, n = 3 independent batches). h RNA level in liver tissue of mice from indicated groups by qRT-PCR (unpaired two-tailed t test, n = 5 mice per group). i Heatmap showing the relative levels of selected genes related to tolerant macrophage pathway based on the RNA-seq data. j Relative mRNA level of Mafk in liver from indicated mice (unpaired two-tailed t test, n = 6 mice per group) (top); immunoblot analysis of liver in indicated mice (bottom). k The qRT-PCR of Mafk in the RAW264.7 cells after 25-HC treatment with a dose dependent for 48 h (unpaired two-tailed t test, n = 3 independent batches) (top); immunoblot analysis of Raw264.7 cells after 25-HC treatment with a dose dependent for 48 h (bottom). l Different expression and quantification of various chemokines in media derived from 25-HC treated or control-RAW264.7 cells was detected by a mouse chemokine array kit (n = 2 independent batches). Red boxes designate the chemokines whose expression was altered by 25-HC treatment. m TGF-β and IL-10 secreted from BMDM cells treated with vehicle and 25-HC were detected by ELISA kit (unpaired two-tailed t test, n = 6 independent batches). Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 7. Exogenous INSIGs and ATF3 blocked miR-9-5p-induced tumor metastasis.

a GFP signals in liver and lung from mice receiving tail vein injections of AAV-GFP, AAV-ATF3 and AAV-INSIG1 + INSIG2 virus. Scale bar, 100 μm. b Quantification of gene expression in liver from mice receiving tail vein injections of AAV-GFP, AAV-INSIG1 and AAV-INSIG2 virus. (unpaired two-tailed t test, n = 6 mice per group). c Western blots showing the expression of key proteins related to cholesterol synthesis in liver from mice receiving tail vein injections of AAV-GFP, AAV-INSIG1 and AAV-INSIG2 virus. d Total cholesterol (TC) and triglycerides (TG) in liver in indicated mice (unpaired two-tailed t test, n = 5 mice per group). e Western blots showing protein expression in liver from mice receiving tail vein injections of AAV-GFP and AAV-ATF3 virus. f Representative IHC images showing CH25H staining. Scale bar, 50 μm. g LC-MS/MS analysis of 25-HC level in liver in indicated groups (unpaired two-tailed t test, n = 5 mice for per group). h Body weight was monitored (two-way ANOVA, n = 6 mice per group). i, j Representative bioluminescence images and quantification of mice inoculated with indicated groups (one-way ANOVA, n = 5 mice per group). k Quantification of liver and lung metastatic nodules (unpaired two-tailed t test, n = 6 mice per group). l RNA level of Mki67 in liver and lung in indicated mice (unpaired two-tailed t test, n = 6 mice per group). m Representative IHC images showing Ki67 staining and the overall percentage of Ki67⁺ tumor cells (one-way ANOVA, n = 5 per group). Scale bar, 50 μm. n Western blots showing proteins in liver from mice receiving tail vein injections of AAV-GFP and AAV-ATF3 virus. o RNA level of M1 and M2 marker gene in liver and lung in indicated mice (unpaired two-tailed t test, n = 6 mice per group). p Purified Kupffer cells were selected with TIM4⁺, I-A/I-E^int expression and stained and further analyzed for CD163 and CD11c expression. Representative flow cytometry results were shown. q Quantification of the M2/M1 ratio in flow cytometry results (one-way ANOVA, n = 3 independent batches). Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 8. Silencing of miR-9-5p attenuated 25-HC-mediated tumor metastasis.

a Schematic diagram of antagomir treated mouse model. 231/Luc cells injected at day 0 in NSG mice. On day 14, tumor was detected. At day 28, mice were treated with antagomir (10 nmol) every 3 days untiled euthanized. 4T1/Luc cells injected at day 0 in BALB/c mice. On day 8, tumor was detected. At day 25, mice were treated with antagomir (10 nmol) every 3 days untiled euthanized, during which time the tumor was removed on day 28. b RT-qPCR-determined miR-9-5p levels in serum, tumor, liver and lung in 231/Luc cells xenografted mice treated with indicated antagomir and with 231/Ctrl, 231/Rab27a KD and 231/miR-9 KO mice (one-way ANOVA, n = 5 mice per group). c Representative PET/CT imaging with ¹⁸F-FDG for accessing liver and lung uptake and quantification in 4T1/Luc cells xenografted mice treated with indicated antagomir. d Representative bioluminescence images of metastasis site in liver and lung of xenografted mice treated with indicated antagomir and with 231/Ctrl, 231/Rab27a KD and 231/miR-9 KO mice. e Quantification of liver and lung metastasis of xenografted mice treated with indicated antagomir and with 231/Ctrl, 231/Rab27a KD and 231/miR-9 KO mice (one-way ANOVA, n = 5 mice per group). f Quantification of gene expression in liver from xenografted mice treated with indicated antagomir and with 231/Ctrl, 231/Rab27a KD and 231/miR-9 KO mice (one-way ANOVA, n = 5 mice per group). g Quantification of gene expression in liver from xenografted mice (one-way ANOVA, n = 5 mice per group). h Immunoblotting analysis of liver sample collected from xenografted mice treated with indicated antagomir and with 231/Ctrl, 231/Rab27a KD and 231/miR-9 KO mice. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.

Fig. 9. Circulating miR-9-5p was correlated with BC patients with metastasis.

a RT-qPCR-determined miR-9-5p levels in an equal amount of EVs derived from indicated cells (normalized to miR-16) (unpaired two-tailed t test, n = 3 independent batches). b RT-qPCR-determined miR-9-5p levels of EVs in serum from indicated xenografted mice. (one-way ANOVA, n = 3 independent batches). c Representative images of miRNA in situ hybridization (miRNA-ISH) in liver metastases from breast cancer patients. Scale bar, 50 μm. d RT-qPCR-determined miR-9-5p levels in serum from normal or BC patient with or without metastasis (unpaired two-tailed t test, n = 5). e Expression levels of miR-9-5p in BRCA tumor samples from the TCGA database. Statistical significance was assessed using two-tailed Mann–Whitney test (left, n = 1102 for Tumor and n = 104 for Normal) and paired two-tailed Student’s t test (right, n = 103). f Expression levels of miR‑9-5p in BRCA tumor samples from the GEO database. Box plots show minima, maxima, median bounds of box and quartiles (one-way ANOVA for left, n = 59 for Normal, n = 169 for Tumor, n = 54 for Metastasis; paired two-tailed t test for right, n = 59). g Expression levels of miR-9-5p in different stage of BRCA tumor samples. Box plots show minima, maxima, median bounds of box and quartiles (one-way ANOVA, n = 104 for normal, n = 182 for stage I, n = 609 for stage II, n = 244 for stage III, n = 20 for stage IV). h IHC analysis of HMGCR, CH25H, INSIG1, INSIG2 and ATF3 expression in liver metastases from breast cancer and non-BC individuals. Scale bar, 100 μm. i LC-MS/MS analysis of oxysterol derived from cholesterol in serum from normal and BC patients (unpaired two-tailed t test, n = 15 per group). j Correlation analysis of MRC1 and Ki67 expression IHC from BC patients metastatic liver. The two-tailed Pearson’s r was indicated (n = 7). k Schematic summary of this study. Data are presented as mean ± s.d. Source data and exact P value are provided as a Source data file.