MIRO2-mediated mitochondrial transfer from cancer cells induces cancer-associated fibroblast differentiation

Abstract

Cancer-associated fibroblasts (CAFs) are key components of the tumor microenvironment that commonly support cancer development and progression. Here we show that different cancer cells transfer mitochondria to fibroblasts in cocultures and xenograft tumors, thereby inducing protumorigenic CAF features. Transplantation of functional mitochondria from cancer cells induces metabolic alterations in fibroblasts, expression of CAF markers and release of a protumorigenic secretome and matrisome. These features promote tumor formation in preclinical mouse models. Mechanistically, the mitochondrial transfer requires the mitochondrial trafficking protein MIRO2. Its depletion in cancer cells suppresses mitochondrial transfer and inhibits CAF differentiation and tumor growth. The clinical relevance of these findings is reflected by the overexpression of MIRO2 in tumor cells at the leading edge of epithelial skin cancers. These results identify mitochondrial transfer from cancer cells to fibroblasts as a driver of tumorigenesis and provide a rationale for targeting MIRO2 and mitochondrial transfer in different malignancies.

Fig. 1. Cancer cells transfer mitochondria to fibroblasts through TNTs.

a, Coculture setup with LifeAct A431 cells (red) stained with MitoTracker green and unstained HPFs. This image was created with BioRender.com. b, Immunofluorescence images of the cocultures, counterstained with Hoechst. c, Representative photomicrographs of cocultures of A431 cells prestained with MitoTracker green and HPFs immunostained for COLI (white) and counterstained with phalloidin (red) and Hoechst (blue). TNT-like structures are indicated by white rectangles together with their length (n = 3 A431–HPF cocultures). d, Percentage of MitoTracker-high HPFs in direct or transwell coculture with A431 cells (n = 3 cocultures per setup). e, Percentage of MitoTracker-high HPFs after coculture with A431 cells in the presence of carbenoxolone (CBX) or vehicle (n = 3 cocultures per treatment group). f, RT–qPCR for GJB2 (encoding connexin 26) relative to RPL27 using RNA from A431 cells transfected with control (scrambled) or connexin 26 (Cx26) siRNA, and percentage of MitoTracker-high HPFs after coculture of siCtrl or siCx26 A431 cells (n = 3 cultures per group). g, Holotomographic imaging showing mitochondrial transfer (white arrows) from A431 LifeAct–MitoTracker green cells to HPFs (unstained) (Supplementary Video 1). h, Representative image of a coculture of A431 cells stained with MitoTracker green and HPFs, immunostained for COLI and counterstained with phalloidin and Hoechst (n = 3 A431–HPF cocultures). White arrows point to TNT-like structures. i, Percentage of MitoTracker-high HPFs after coculture with A431 cells in the presence of nocodazole (Noc), dihydrocytochalasin B (Cyto B) or vehicle (n = 3 cocultures per treatment group). j, Western blot analysis for SEC3 and SEC5 using lysates from A431 cells transfected with siCtrl, siSEC3 or siSEC5 (n = 2 cultures per group). Graph shows the percentage of MitoTracker-high HPFs after coculture with MitoTracker-stained control or SEC3–SEC5-knockdown A431 cells (n = 3 cocultures per group). k, Percentage of MitoTracker-high HPFs after coculture with HaCaT or A431 cells (n = 3 cocultures per cell line). l, Representative immunofluorescence images depicting cocultures of MDA-MB-231 and PANC1 cells prestained with MitoTracker green and HPFs immunostained for COLI (white) and counterstained with Hoechst (blue) (n = 3 cocultures per cell line). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (d–f,i,k) or two-sided one-way ANOVA with Bonferroni post hoc multiple comparison test (j) was used to determine statistical significance. Scale bars, 50 μm (b,g), 20 μm (c) or 25 μm (h,l). Source data

Fig. 2. Cancer cells transfer mitochondria to fibroblasts in vitro and in vivo.

a, qPCR for the human mtDNA encoding tRNA-Leu(UUR) relative to mouse nuclear DNA encoding beta2-microglobulin (B2m) using DNA from MitoTracker-positive and MitoTracker-negative mouse fibroblasts sorted from n = 3 cocultures. Total mtDNA content was calculated on the basis of C_t values. b, RT–qPCR for human and mouse FN1 and Fn1 relative to RPL27 or Rps29, respectively, using RNA from MitoTracker-high and MitoTracker-low mouse fibroblasts sorted from n = 3 cocultures. c, Transfer efficiency in human–human and human–mouse cocultures (n = 3 cocultures per group). d, Comparison of SNPs within the 16S rRNA gene region of A431 cell mitochondria with those of control and recipient mouse fibroblasts. SNPs from A431 cells in recipient fibroblasts are indicated with rectangles (n > 300,000 cells pooled from three independent cocultures). e, Experimental setup and fluorescence images of HPF Su9–RFP and A431 Su9–GFP cocultures. This image was created with BioRender.com. Bottom, (1) colocalization of A431 and HPF mitochondria (orange), (2) HPF periphery with own mitochondrial network (red) and (3) nonrecipient HPFs (red). f, Section of a xenograft tumor formed by Su9–RFP A431 cells, showing Su9–RFP fluorescence in mitochondria of keratin 14 (K14)-negative cells and cultured fibroblasts from these tumors showing Su9–RFP fluorescence (red) and vimentin expression. g, qPCR for the mtDNA-encoded human tRNA-Leu gene relative to the mouse B2m gene using total DNA from cultured mouse fibroblasts isolated from noninjected ear skin (NS) or A431 xenograft tumors (n = 3 normal skin samples and n = 3 tumor samples from different mice). h, Representative immunofluorescence staining of A431 xenograft tumors for COLI (green) and human mitochondria (red). Costaining (yellow) of stromal cells adjacent to tumors was confirmed by colocalization analysis (site indicated with an asterisk). The white arrow indicates the line along which the intensity values of the different fluorescence signals were measured, starting from the initial position at the base of the arrow and ending at the arrowhead. Separate channels of zoomed-in regions are displayed (n = 3 sections from different tumors). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (a,c,g) or two-sided one-way ANOVA with Bonferroni post hoc multiple comparison test (b) was used to determine statistical significance. One control value was set to 1. Scale bars, 25 μm (e,f,h). Source data

Fig. 3. Transferred cancer cell mitochondria induce a CAF phenotype.

a–c, Volcano plots displaying differentially expressed genes in MitoTracker-low versus control (a), MitoTracker-high versus control (b) and MitoTracker-high versus MitoTracker-low HPFs (c) sorted from n = 3 cocultures with A431 cells. d,e. RNA-seq data from sorted HPFs depicting expression of INHBA, IL6, ACTA2 and COL1A1 (d) or PDGFRA, PDGFRB, S100A4, FAP and CD74 (e) in MitoTracker-high, MitoTracker-low and control groups sorted from n = 3 cocultures. f–h, Comparative analysis of gene signatures in MitoTracker-high versus MitoTracker-low HPFs with published CAF datasets^,,, showing similarities of MitoTracker-high HPFs with myCAFs and iCAFs. i, Volcano plot displaying differentially abundant proteins in MitoTracker-high versus MitoTracker-low HPFs sorted from n = 4 cocultures with A431 cells. j, Correlation analysis of gene and protein expression in MitoTracker-high versus MitoTracker-low HPFs. Significantly regulated pathways (q < 0.1) are highlighted (blue, Hallmarks of Cancer pathways; purple, Wikipathways). k, Percentage of Ki67-positive HPFs and relative levels of intracellular ATP and MitoSOX in sorted HPFs (n = 3 cocultures for Ki67 and 6 cocultures for ATP and MitoSOX). l, Percentage of Ki67-positive A431 cells in spheroids cultured with CM from control HPFs and sorted MitoTracker-high and MitoTracker-low HPFs (n = 3 spheroids per treatment group). m, Transwell migration of A431 cells in CM from sorted MitoTracker-low and MitoTracker-high and control HPFs (n = 3 transwell cultures per treatment group). n. Relative colony size of A431 cells plated on dECM from sorted HPFs (n = 3 cultures). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (k (right and middle),n) or two-sided one-way ANOVA with Bonferroni post hoc multiple comparison test (d,e,k (left),l,m) was used to determine statistical significance. One control value was set to 1. Source data

Fig. 4. CAF reprogramming through transplantation of cancer cell mitochondria.

a, Representative fluorescence images of HPFs MitoCepted with MitoTracker green-stained A431 mitochondria (MitoCepted HPFs) or mock treatment, counterstained with Hoechst (blue) (n = 3 cultures per group). b, qPCR for the mtDNA encoding tRNA-Leu(UUR) relative to the nucDNA encoding B2M using DNA from MitoCepted (MC) or mock-treated (Ctrl) HPFs. Relative mtDNA content (based on C_t values) is indicated (n = 3 cultures per group). c, Representative confocal image in the xyz plane showing HPFs prestained with MitoTracker red and MitoCepted with MitoTracker green-labeled A431 mitochondria (left) and TOM20–GFP-expressing HPFs MitoCepted with mitochondria from A431 Su9–RFP cells (right). Yellow staining indicates mitochondrial fusion. d, Percentage of Ki67⁺ MitoCepted or control HPFs among all cells (n = 3 cultures per group). e, RT–qPCR for INHBA, IL6, ACTA2 and COL1A1 relative to RPL27 using RNA from MitoCepted or mock-treated HPFs (n = 3 cultures per group). f, FluidFM experimental setup, adapted from a previous study. g, Image of FluidFM-mediated injection of mitochondria into HPFs. h, Percentage of Ki67⁺ fibroblasts in HPFs injected with A431-derived mitochondria using FluidFM or mock treatment (n = 3 cultures per group). i, RT–qPCR for INHBA using RNA from HPFs subjected to MitoCeption with mitochondria from HaCaT, HaCaT-Ras or A431 cell lines (n = 3 cultures per cell line). j, RT–qPCR for INHBA using RNA from (1) mock-treated HPFs or HPFs subjected to MitoCeption with mitochondria (2) from keratinocytes of a healthy individual, (3) from normal keratinocytes of a person with SCC or (4) malignant cancer cells of the same person with SCC. Right, representative FN1–COLI immunofluorescence stainings with quantification of staining intensity in the dECM produced by HPFs after MitoCeption with mitochondria from the different primary donor cells (n = 3 MitoCeptions per cell type). k, Percentage of Ki67⁺ HPFs subjected to MitoCeption with different amounts of mitochondria isolated from A431 cells. Numbers on the x axis show the ratio of donor A431 cells (used for mitochondrial isolation) and recipient HPFs (n = 3 cultures per group). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (b,d,e,h) or two-sided one-way ANOVA with Bonferroni post hoc multiple comparison test (i–k) was used to determine statistical significance. Scale bars, 100 μm (a,j), 10 μm (c) and 25 μm (g). Source data

Fig. 5. Functional cancer cell mitochondria are required for CAF reprogramming.

a, Basal respiration of HPFs subjected to MitoCeption with A431 or HaCaT mitochondria or mock treatment in comparison to A431 cells (n = 5 independent MitoCeptions per cell type). b, Proton leak in the same cultures as in a. c, Percentage of Ki67⁺ A431 cells and transwell migration of A431 cells cultured in CM of MitoCepted or mock-treated HPFs (n = 3 cultures per group). d, RT–qPCR for INHBA and IL6 using RNA from MitoCepted (mitochondria from MDA-MB-231 breast cancer cells) or mock-treated HPFs (n = 3 cultures per group). e, Clonogenicity of MDA-MB-231 cells cultured in CM from HPFs subjected to MitoCeption with MitoTracker green-stained mitochondria from MDA-MB-231 cells or mock treatment (n = 3 cultures per treatment group). f, Representative image of 3-week-old ear xenograft tumors (arrowheads) following intradermal coinjection of A431 cells and MitoCepted (with A431 mitochondria) or mock-treated HPFs and tumor volume at various time points (n = 5 tumors per group from different mice). g, Representative immunofluorescence stainings of tumors formed by A431 cells and MitoCepted or mock-treated HPFs for E-cadherin and FN1 (green) and MECA32 (red), counterstained with Hoechst (blue) (n = 5 tumors per group from different mice). h, Percentage of Ki67⁺ HPFs subjected to MitoCeption with A431 lmt mitochondria or mock treatment and RT–qPCR for INHBA using RNA from HPFs subjected to MitoCeption with A431 lmt mitochondria or mock treatment (n = 9 Ki67 or n = 3 RT–qPCR cultures per treatment group). i, Percentage of Ki67⁺ A431 cells (left) or transwell migration of A431 cells (right) cultured in CM from HPFs subjected to MitoCeption with A431 lmt mitochondria or mock treatment (n = 9 Ki67 or n = 3 transwell migration cultures per treatment group). j, Left, tumor volume at various time points during tumor development by A431 cancer cells coinjected with MitoCepted HPFs, which received mitochondria from control or lmt A431 cells (n = 4 tumors per group from different mice). Right, Histological stainings of a tumor from each group. Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (c–e,h,i) or two-sided one-way (a,b) or two-way (f,j) ANOVA with Bonferroni post hoc multiple comparison test was used to determine statistical significance. Scale bars, 200 μm (g) and 1 mm (j). Source data

Fig. 6. MIRO2 is overexpressed at the leading edge of SCCs.

a, Dot plot showing expression of mitochondrial trafficking genes; violin plot showing expression of MIRO2 in different cell types in SCCs (n = 5,799 myeloid cells, 4,644 tumor cells, 1,495 epithelial cells, 584 fibroblasts, 413 lymphoid cells, 169 endothelial cells and 129 melanocytes). b, Violin plot showing expression of MIRO2 in tumor cell subpopulations in SCCs based on scRNA-seq data (n = 296 tumor-specific keratinocytes (TSKs), 1,385 basal tumor keratinocytes (KC), 725 cycling tumor keratinocytes and 2238 differentiating tumor keratinocytes). c, Feature plots showing spatial distribution of MIRO2 transcripts in human skin SCC; violin blots showing MIRO2 transcripts at the tumor leading edge versus the total tumor and its microenvironment (TME) (n = 2 tumors from different patients; P2 and P6). d, Western blot of lysates from HPFs, HaCaT, HaCaT-Ras and A431 cells for MIRO2 and GAPDH. e, MIRO2 and K14 immunofluorescence stainings of sections from 3D organotypic skin cultures with HPFs and HaCaT or A431 cells and quantification of the MIRO2-positive area (n = 3 3D cultures per epithelial cell line). Scale bar, 100 μm. f, Forest plot showing the 5-year disease-specific survival (DSS) associated with MIRO2 expression across solid cancers based on TCGA. Hazard ratios (HRs) and 95% confidence intervals (CIs) based on Cox proportional hazard model are shown. The last point represents the estimate from the random-effects meta-analysis (n = 8,941 patients). g, Pearson correlation coefficient (ρ) and 95% CIs between the enrichment score of the leading edge (LE) signature and MIRO2 expression across the different solid cancers in TCGA. The last point represents the estimate from the random-effects meta-analysis (n = 10,238 patients). h, Dependency of different cancers on MIRO2 expression as documented in the DepMap Portal. Gene effect scores are derived from DEMETER2 or CERES. A lower score denotes a greater dependency on expression. Violin plots in a–c show the median (center line), 25th and 75th percentiles (box bounds) and whiskers extending to the most extreme data points within 1.5 times the interquartile range from the box. Points outside this range are plotted as outliers. The graph in e shows the mean ± s.e.m. A Mann–Whitney U-test for comparison between two groups (a,b) or unpaired two-sided Student’s t-test (e) was used to determine statistical significance. Source data

Fig. 7. MIRO2 is required for mitochondrial transfer.

a, RT–qPCR for MIRO2 using RNA from siCtrl or siMIRO2 A431 cells; Western blot of total and mitochondrial lysates from these cells for MIRO2, vinculin or HSP60 (loading controls) (n = 3 cultures per group). b, Fluorescence images of LifeAct A431 cells (red) stained with MitoTracker green and transfected with siCtrl or siMIRO2 in coculture with HPFs, counterstained with Hoechst. White arrowheads indicate A431 cells. c, Percentage of MitoTracker-high HPFs in cocultures with siMIRO2 or siCtrl A431 cells, normalized to the number of cancer cells (n = 3 cocultures per group). d, qPCR for mtDNA encoding tRNA-Leu(UUR) relative to nucDNA encoding B2M using DNA from A431 cells transfected with siCtrl or siMIRO2. Total mtDNA content was calculated on the basis of C_t values (n = 3 cultures per group). e, Mitochondrial mass in siCtrl and siMIRO2 A431 cells based on MitoTracker green mean fluorescence intensity (MFI) (n = 3 per group). f, Confocal images of siCtrl or siMIRO2 A431 cells incubated with MitoTracker green. The dashed line marks the outer edge of the cell (n = 3 cultures per group). g, RT–qPCR for INHBA and IL6 using RNA from HPFs incubated with CM of siCtrl or siMIRO2 A431 cells (n = 3 cultures per treatment group). h, RT–qPCR for INHBA and IL6 using RNA from sorted HPFs cocultured with siCtrl or siMIRO2 A431 cells (n = 3 cocultures per group). DC, direct culture. i, OCR of siCtrl or siMIRO2 A431 cells. The time of drug injection is indicated (n = 3 cultures per group). j, RT–qPCR for MIRO2 using RNA from A431 cells transfected with control or MIRO2 overexpression vectors (OE-MIRO2) (n = 3 cultures per group). Western blot of lysates from control or MIRO2-overexpressing A431 cells for MIRO2 or GAPDH (n = 2 cultures per group). k, Percentage of MitoTracker-high HPFs after coculture with MitoTracker-stained control or MIRO2-overexpressing A431 cells (n = 3 cocultures per group). l, Percentage of Ki67⁺ HPFs after coculture with control or MIRO2-overexpressing A431 cells (n = 9 cocultures per group). m, Percentage of MitoTracker-high HPFs after coculture with MitoTracker-stained control or MIRO2-overexpressing A431 cells, with or without treatment with dihydrocytochalasin B (n = 3 cocultures per group). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (a,c–e,g,h,l) or two-sided one-way ANOVA with Bonferroni post hoc multiple comparison test (m) was used to determine statistical significance. Scale bars, 100 μm (b) and 25 μm (f). Source data

Fig. 8. MIRO2 depletion in cancer cells reduces mitochondrial transfer and tumor growth.

a, Percentage of Ki67⁺ A431 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). b, Relative viability of A431 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). c, Transwell migration of A431 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). d, Relative spheroid area of a single hanging drop formed by siCtrl or siMIRO2 A431 cells and representative images of the spheroids (n = 9 spheroids per group). e, Percentage of Ki67⁺ SCC13 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). f, Relative viability of SCC13 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). g, Transwell migration of SCC13 cells 24 h after transfection with siCtrl or siMIRO2 (n = 3 cultures per group). h, Relative spheroid area of a single hanging drop formed by siCtrl or siMIRO2 SCC13 cells and representative images of the spheroids (n = 15 spheroids per group). i, Photo of 5-week-old ear xenograft tumors (indicated by arrowheads) formed after injection of 200,000 A431 cells transfected with siMIRO2 or siCtrl and tumor volume at different time points (n = 3 tumors per group from different mice). j, Representative photomicrographs of Herovici-stained tumors (left) formed by siMIRO2 or siCtrl A431 cells and immunofluorescence staining of sections from these tumors for K14 (green) and MIRO2 (red), counterstained with Hoechst (blue). Inset, a tumor cell island with persistent MIRO2 knockdown (n = 3 sections from independent tumors per group). k, Normalized cell count of A431 LifeAct–RFP cells transfected with siCtrl or siMIRO2, cocultured with or without HPFs in spheroids and analyzed by FACS after 5 days (n = 3 spheroids per group). l, Tumor volume during tumor development following coinjection of HPFs (with or without A431-derived mitochondria, introduced using MitoCeption) and A431 cells transfected with either siMIRO2 or siCtrl (n = 5 tumors per group). m, Representative H&E stainings of tumors from each group (n = 5 mice). Graphs show the mean ± s.e.m. An unpaired two-sided Student’s t-test (a–h) or two-sided one-way (k) or two-way (i,l) ANOVA with Bonferroni post hoc multiple comparison test (i,k,l) was used to determine statistical significance. Scale bars, 100 μm (d), 200 μm (j) and 1 mm (m). Source data

Extended Data Fig. 1. Different cancer cells transfer mitochondria to fibroblasts in vivo.

a. Representative fluorescence images of FACS-sorted HPFs. Cells were sorted based on the intensity of MitoTracker Green staining, re-seeded for fluorescence analysis, and counterstained with Hoechst (blue). b. Representative fluorescence stainings of HPFs incubated with A431 CM (±MitoTracker Green staining) and counterstained with Hoechst (blue). c. Representative immunofluorescence staining of A431 skin xenograft sections for PDGFR-α (green) and an antibody specific for human mitochondria (red), co-stained with Hoechst (blue). Co-staining of stromal cells adjacent to tumors was confirmed by co-localization analysis (site indicated with an asterisk). White arrows in (c) and (d) indicate the line along which the intensity values of the different fluorescence signals were measured, starting from the initial position at the base of the arrow and ending at the arrowhead. Separate channels of zoomed-in regions are displayed. n = 3 sections from different tumors. d. Representative sections of breast and pancreatic cancer xenograft tumors from LM2 and PANC1 cells, respectively, immunostained for COLI (green) and co-stained with an antibody that specifically detects human mitochondria (red). Co-staining (yellow) of stromal cells adjacent to tumors was confirmed by co-localization analysis (site indicated with an asterisk). Separate channels of zoomed-in regions are displayed. n = 3 sections from different tumors. Scale bars: 100 μm (a–d). Source data

Extended Data Fig. 2. RNA-Seq data of MitoTracker-high and -low HPFs in co-cultures.

a. FACS gating strategy for isolation of HPFs from A431-HPF co-cultures, demonstrating the arbitrary division into two HPF populations. FSC: Forward scatter; SSC; side scatter. b. Principal Component Analysis (PCA) of RNA-seq data from MitoTracker-high and -low HPFs and control HPFs that underwent the sorting procedure but were not maintained in co-cultures. n = 3 co-cultures per group. c, d. Volcano plots showing differentially expressed genes in HPFs in co-culture (MitoTracker-high (c) or -low (d)) vs. control HPFs in monoculture. Selected CAF marker genes are in red. n = 3 co-cultures and sorting experiments. e. RNA-seq data showing expression of ISGs in MitoTracker -high and -low HPFs after co-culture with A431 cells vs. control HPFs. n = 3 co-cultures per group. f. qRT-PCR for INHBA and IL6 relative to RPL27 using RNA from MitoTracker-high and -low HPFs after co-culture with A431 cells vs. control HPFs (repetition experiment). n = 3 co-cultures per group. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test (f) or two-sided one-way ANOVA with Bonferroni post-host multiple comparison test (e) were used to determine statistical significance. Source data

Extended Data Fig. 3. Proteomics analysis of HPFs in co-cultures and CAF features of recipient fibroblasts.

a. PCA of proteomics data from MitoTracker-high and -low and control HPFs. n = 4 co-cultures and sorting experiments. b. Volcano plot showing differentially abundant proteins in sorted MitoTracker-low vs. control HPFs. n = 4 co-cultures per group. Selected CAF marker proteins are in red. c. Volcano plot showing differentially abundant proteins in sorted MitoTracker-high vs. HPFs. n = 4 co-cultures per group. Selected CAF marker proteins are in red. d. Unbiased pathway analysis of MitoTracker-high vs. - low HPFs referenced to Hallmarks of Cancer (left) and WikiPathways (right). e. Gating strategy (top panel) and representative histogram of flow cytometry for Ki67⁺ cells among RFP-high and -low and control HPFs (bottom panel). Gating was set based on the negative control. f. Percentage of Ki67⁺ cells among RFP-high- and -low HPFs after co-culture with A431 cells expressing Su9-RFP. n = 3 co-cultures per group. g. qRT-PCR for INHBA, IL6, ACTA2 and COL1A1 relative to RPL27 using RNA from RFP–high and -low HPFs after co-culture with A431 cells expressing Su9-RFP. n = 3 co-cultures per group. h, j. Percentage of Ki67⁺ cells and transwell migration of primary cutaneous SCC cells (h), SCC13 cells (i) and HaCaT-Ras cells (j) cultured with CM from control or MitoTracker-low or -high HPFs sorted after co-culture with the respective cancer cell lines. n = 3 co-cultures per group. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test (f, g) or two-sided one-way ANOVA with Bonferroni post-host multiple comparison test (h–j) were used to determine statistical significance. Source data

Extended Data Fig. 4. HPFs acquire CAF properties after MitoCeption with cancer cell mitochondria.

a. qRT-PCR for INHBA, IL6, COL1A1 and ACTA2 relative to RPL13A using RNA from MitoCepted (A431 mitochondria) or mock-treated HPFs. n = 3 cultures per group. b. qRT-PCR for ISG15, ISG20, and OASL relative to RPL27 using RNA from MitoCepted or mock-treated HPFs. n = 3 cultures per group. c. BCA protein quantification in mitochondrial lysates from 10⁷ HaCaT or A431 cells. n = 3 cultures per cell line. d. Seahorse analysis of HPFs subjected to MitoCeption with A431 or HaCaT mitochondria or mock treatment, and of A431 cells. n = 5 cultures per group. e. Basal respiration and proton leak of HPFs, subjected to MitoCeption with HaCaT or HaCaT-Ras mitochondria or mock treatment, determined by Seahorse XF stress test. n = 5 cultures per group. f. qRT-PCR for INHBA, IL6, COL1A1 and ACTA2 relative to RPL27 using RNA from HPFs subjected to MitoCeption with MitoTracker Green-stained mitochondria from A431 cancer cells, pre-treated with oligomycin or mock treatment. n = 3 cultures per group. g. Percentage of Ki67-positive HPFs subjected to MitoCeption with MitoTracker Green-stained mitochondria from A431 cancer cells, pre-treated with oligomycin or mock treatment. n = 3 cultures per treatment group. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test (a–c) or two-sided one-way ANOVA with Bonferroni post-host multiple comparison test (e–g) were used to determine statistical significance. Source data

Extended Data Fig. 5. HPFs undergo senescence five days post-MitoCeption.

a. qRT-PCR for INHBA and IL6 using RNA from MitoCepted (mitochondria from PANC1 pancreatic cancer cells) or mock-treated HPFs. n = 3 cultures per group. b. Clonogenicity of PANC1 cells cultured in CM from HPFs subjected to MitoCeption with MitoTracker Green-stained mitochondria from PANC1 pancreatic cancer cells or mock treatment. n = 3 cultures per group. c. Percentage of Ki67-positive HPFs cultured for 5 days after MitoCeption with A431 mitochondria or mock treatment. n = 3 cultures per group. d. Representative photomicrographs of HPFs cultured for 5 days after MitoCeption with A431 mitochondria or mock treatment and stained for SA-β-Gal, and quantification of the percentage of SA-β-Gal positive cells. n = 3 cultures per group. e. qRT-PCR for CDKN1A and CDKN2B using RNA from HPFs subjected to MitoCeption with mitochondria from A431 cancer cells or mock treatment; 5 days after MitoCeption. n = 3 cultures per treatment group. f. qRT-PCR for INHBA, IL6, COL1A1 and ACTA2 using RNA from HPFs subjected to MitoCeption with MitoTracker Green-stained mitochondria from A431 cancer cells or mock treatment; 5 days after MitoCeption. n = 3 cultures per group. g. Representative xenograft tumor sections (two weeks after injection) showing MitoCepted Su9-RFP-positive fibroblasts that had received mitochondria from A431-Su9-RFP A431 cells. Co-localization analysis reveals the presence of Su9-RFP-positive mitochondria in collagen I-positive fibroblasts (indicated with an asterisk). The white arrow indicates the line along which the intensity values of the different fluorescence signals were measured, starting from the initial position at the base of the arrow and ending at the arrowhead. n = 3 sections from different tumors. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test (a–f) was used to determine statistical significance. Scale bars: 20 μm (d), 100 μm (g). Source data

Extended Data Fig. 6. Functional cancer cell mitochondria are required to induce CAF properties.

a. Percentage of Ki67-positive HPFs subjected to MitoCeption with mitochondria isolated from A431 cells pre-treated with CCCP or vehicle. n = 3 cultures per treatment group. b. Seahorse analysis of HPFs subjected to MitoCeption with mitochondria isolated from A431 cells pre-treated with CCCP (n = 12 cultures) or DMSO (n = 5 cultures). c. qPCR for the mtDNA encoding tRNA-Leu relative to nucDNA encoding B2 microglobulin using total DNA from A431 control or low mtDNA (lmt) cells. The relative mtDNA content was calculated based on the Ct values. n = 3 cultures per group. d. BCA protein quantification in mitochondrial lysates from 10⁷ A431 Ctrl and A431 lmt cells. n = 3 cultures per group. e. Oxygen consumption rate (OCR) of A431 Ctrl and A431 lmt cells, determined by Seahorse XF stress test. The time of drug injection is indicated. n = 8 cultures per group. f. Basal and maximal respiration and spare respiratory capacity of A431 Ctrl and A431 lmt cells. n = 8 cultures per group. g. Viability and cell count of A431 Ctrl and A431 lmt cells. n = 3 cultures per group. h. qRT-PCR for IL6, COL1A1 and ACTA2 relative to RPL27 using RNA from HPFs subjected to MitoCeption with mitochondria from A431 Ctrl or A431 lmt cells. n = 3 cultures per group. i. Tumor volume at various time points during development of tumors formed by control or lmt (low mtDNA) A431 cells. n = 5 tumors per group from different mice. j. Representative images of H&E stainings of tumors from each group. n = 5 tumors per group from different mice. Scale bar: 1 mm. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test (a, c, d, f–h) or two-sided two-way ANOVA with Bonferroni post-host multiple comparison test (i) were used to determine statistical significance. Scale bar: 1 mm (j). Source data

Extended Data Fig. 7. MIRO2 is predominantly expressed in tumor keratinocytes of SCCs.

a. Ridge plot showing abundance of mRNAs encoding mitochondrial trafficking proteins in different cell types based on published scRNA-seq data from human cutaneous SCCs. n = 10 tumors and patient- and site-matched normal skin collected from 10 individuals. b. Ridge plot showing abundance of mRNAs encoding mitochondrial trafficking proteins in different cell types based on published scRNA-seq data from human cutaneous SCCs. n = 5 invasive SCCs and n = 6 normal skin samples. c. Odds ratio analysis of MIRO2 transcript co-localization with CAF subtypes in patient tumor sections. The table correlates MIRO2 expression with expression of PDGFRA and specific markers for CAF subtypes—MMP11 for CAFinfla, ACTA2 for CAFmyo, and CFD for CAFadi—quantified by odds ratios (OR) across patient samples. n = 6 tumor sections. Accompanying the table, patient tissue sections are illustrated, with transcript-positive spots for MIRO2 and CAF markers denoted by colored dots, showcasing the spatial interplay within the tumor microenvironment. OR values denote co-localization strength, and p-values indicate statistical significance. P2_rep2: n = 646, P4_rep1: n = 744, P5_rep1: n = 590, P9_rep2: n = 1,071, P6_rep1: n = 3,650, P10_rep1: n = 608. Significant values are in red. d. Comparison of CAF score distribution between MIRO2+ spots and their neighbors (True) and remaining spots (False). n = 2 tumors from different patients (referred to as P2 and P6 from Ji et al.). Violin plots in (d) show the median (center line), 25^th and 75^th percentiles (box bounds), and the whiskers extend to the most extreme data points within 1.5 times the interquartile range from the box. Points outside this range are plotted as outliers. Fisher’s exact test was used to determine statistical significance (c). Source data

Extended Data Fig. 8. Global MIRO2 expression does not correlate with survival across different human tumors.

Kaplan-Meier plots showing the disease specific survival (DSS) rate by cancer type in TCGA patients with high or low expression of MIRO2. T1 (Tercile 1; lowest): low expression of MIRO2; T3 (Tercile 3; highest): high expression of MIRO2. Source data

Extended Data Fig. 9. MIRO2 promotes mitochondrial transfer from cancer cells to HPFs via TNTs.

a. Representative photomicrographs of co-cultures of LifeAct-A431 cells, pre-labeled with MitoTracker DeepRed (red) and transfected with siCtrl or siMIRO2 pre-labeled with Label IT® siRNA Tracker™ fluorescein (green), with HPFs. Cultures were counterstained with Hoechst (blue). Note the exclusive detection of the labeled siRNA in A431 cells. Scale bar: 25 μm. n = 3 co-cultures. b. qRT-PCR for MIRO2 relative to RPL27, using RNA from HPFs sorted following co-culture with siCtrl or siMIRO2 A431 cells. n = 3 co-cultures per group. c. Western blot analysis of total lysates from A431 cells transfected with siCtrl, siMIRO1, siTRAK1, or siTRAK2 for MIRO1, TRAK1, TRAK2, and GAPDH (loading control). n = 2 cultures per group. d. Mitochondrial transfer efficiency between A431 cells transfected with siMIRO1, siTRAK1, siTRAK2, or siCtrl and HPFs in co-culture, based on the proportion of MitoTracker-high fibroblasts. n = 3 co-cultures per group. e. qRT-PCR for INHBA and IL6 using RNA from HPFs subjected to MitoCeption with mitochondria from A431 cells, which had been transfected with siCtrl or siMIRO2. n = 3 transfected cultures per group. f. Western blot analysis for MIRO2 using lysates from HaCaT or SCC13 cells transfected with control or MIRO2 overexpression vectors (OE-MIRO2). n = 3 cultures per group. Bar graphs show percentage of MitoTracker-high HPFs after direct co-culture (DC) with control or MIRO2-overexpressing cells. n = 3 co-cultures per group. Graphs show mean ± SEM. Unpaired two-sided Student’s t-test was used to determine statistical significance (b, d–f). Source data