Intercellular nanotube-mediated mitochondrial transfer enhances T cell metabolic fitness and antitumor efficacy

Abstract

Mitochondrial loss and dysfunction drive T cell exhaustion, representing major barriers to successful T cell-based immunotherapies. Here, we describe an innovative platform to supply exogenous mitochondria to T cells, overcoming these limitations. We found that bone marrow stromal cells establish nanotubular connections with T cells and leverage these intercellular highways to transplant stromal cell mitochondria into CD8⁺ T cells. Optimal mitochondrial transfer required Talin 2 on both donor and recipient cells. CD8⁺ T cells with donated mitochondria displayed enhanced mitochondrial respiration and spare respiratory capacity. When transferred into tumor-bearing hosts, these supercharged T cells expanded more robustly, infiltrated the tumor more efficiently, and exhibited fewer signs of exhaustion compared with T cells that did not take up mitochondria. As a result, mitochondria-boosted CD8⁺ T cells mediated superior antitumor responses, prolonging animal survival. These findings establish intercellular mitochondrial transfer as a prototype of organelle medicine, opening avenues to next-generation cell therapies.

Keywords: CAR T therapy; CD8(+) T cells; TCR-T therapy; TIL therapy; Talin 2; bone marrow stromal cells; cancer immunotherapy; immune metabolism; mitochondrial transfer; nanotubes.

Figure 1:. Intercellular nanotubes enable mitochondrial trafficking from BMSCs to CD8⁺ T cells.

(A-D) FESEM images showing nanotubes (yellow arrows) between BMSCs and CD8⁺ T cells in human (A, B) and mouse (C, D) co-cultures. The images show thin (A), thick (B), and branched (B1) nanotubes. (E-G) Bar graphs showing the number of nanotubes between BMSCs and CD8⁺ T cells (E), and the distribution of lengths (F) and widths (G) of nanotubes connecting the BMSC and CD8⁺ T cells, as calculated from the FESEM images. Data shown are mean ± s.e.m. (H, I) Confocal microscopy images of nanotube (yellow arrows) formation between Mito-DsRed BMSCs and CD8⁺ T cells in human (H) and mouse (I) co-cultures. Transfer of mitochondria has been observed inside the nanotube (red arrows). Co-cultures were fixed after 24 hrs and stained with phalloidin green and DAPI. Co-localization of DAPI and DsRed signals within the nanotube (H) indicates the trafficking of intact mitochondria from the BMSCs to CD8⁺ T cells.

Figure 2:. Mitochondrial transfer enhances CD8⁺ T cell metabolic fitness.

(A) Cartoon depicting the transwell co-culture system designed to promote mitochondrial transfer from Mito-DsRed BMSCs to CD8⁺ T cells. (B) Percentage of mouse CD8⁺DsRed⁺T cells 48 hrs after co-culture with Mito-DsRed BMSCs. Data shown are mean ± s.e.m., n = 19 independent co-culture experiments. (C) Flow cytometry plots of mouse CD8⁺ T cells 48 hrs after co-culture with Mito-DsRed BMSCs before (left) and after (right) sorting based on DsRed signal. Numbers indicate percentage after gating on live lymphocytes. (D) Representative confocal microscopy image showing FACS-sorted mouse CD8⁺ T cells that have received donor-labeled mitochondria from Mito-DsRed BMSCs. MitoTracker Deep Red FM was used to label total mitochondria after sorting. (E) Correlative confocal and transmission electron microscopy image of FACS-sorted mouse CD8⁺ T cells that have received donor-labeled mitochondria from MitoDsRed BMSCs. Overlay of nucleus (DAPI) and acquired mitochondria (DsRed) with the electron micrograph of the same section. E1,2: electron micrograph alone of transferred mitochondria. (F) Percent increase in mtDNA (as measured by mt-Co2 gene normalized to nuclear App gene) of mouse Mito⁺ cells relative Mito⁻ (n = 7). (G) Restriction enzyme analysis of Mito⁺ and Mito⁻ BALB/c CD8⁺ T cells after co-culture with C57BL/6-derived Mito-DsRed BMSCs. C57BL/6 cells have a single nucleotide polymorphism at A9348 in the mt-Co3 gene that creates an AspI restriction site. (H) Oxygen consumption rates (OCR) of FACS-sorted Mito⁺ and Mito⁻ mouse CD8⁺ T cells after co-culture with Mito-DsRed BMSCs that were left untreated or pre-treated with 200 ng/ml Ethidium bromide (EtBr) in DMEM complete medium supplemented with 50μg/ml uridine to render donor mitochondria dysfunctional. CD8⁺ T cells monocultured (CD8 mono) were included as additional control. Data were obtained under basal culture conditions and in response to the indicated molecules. (I) Basal respiration and (J) spare respiratory capacity (SRC) (n = 6–12, 2–4 technical replicates per 3-time points). *P < 0.05; ** P < 0.01; *** P < 0.001; *** P < 0.0001 (one-way ANOVA with Dunnett’s multiple-comparison test).

Figure 3:. Mitochondrial transfer from BMSCs to T cells is TLN2-dependent.

(A) GSEA of human and mouse Mito⁺ cells showing positive enrichment of genes upregulated in cancer cells acquiring mitochondria from neighboring cells (B) Heat map showing 26 of the enriched genes in the GSEA that are co-regulated in both human and mouse Mito⁺ cells. (C) Volcano plot showing changes in gene expression between Mito⁺ and Mito⁻ human CD8⁺ T cells. Gene expression was evaluated by RNA-seq on Mito⁺ and Mito⁻ cells FACS-sorted after 48 hrs co-culture with human Mito-DsRed BMSCs (n = 3 healthy donors), (dashed line, P adj=0.05)). (D) Heat map of top 22 selected DEG in Mito⁺ and Mito⁻ human CD8⁺ T cells (FDR corrected, Padj.hs <0.05) that are more robustly co-regulated in mouse CD8⁺ T cells (P.ms < 0.15). (E) Overrepresentation analysis of the 59 orthologous genes co-regulated in human and mouse Mito⁺ and Mito⁻ CD8⁺ T cells using GO ontology terms (baseMean.hs >100, Padj.hs <0.05). Bar graph illustrates the Top 5 gene ontology terms of genes enriched. (F) Percentage of DsRed⁺CD8⁺ T cells after BMSCs-CD8⁺ T cells co-cultures. Indicated doses of the farnesyltransferase and geranylgeranyltransferase inhibitor L-778123 were added to BMSCs and CD8 T cells before or during co-cultures. Data are shown as mean ± s.e.m. relative to untreated co-cocultures. (G) Double strand break repair, as indicated by MRE11 ChIP-seq signal at the TLN2 locus in CRISPR/Cas (TLN2-AA or TLN2-AB) and mock-treated T cells. Y-axis scale depicting the NGS read coverage at the left of each sample. Genomic coordinates on the x-axis. (H) Percentage of DsRed⁺CD8⁺ T cells after BMSCs-CD8⁺ T cell co-cultures in which TLN2 was deleted in the indicated cell type (magenta color). Data are shown as mean ± s.e.m. relative to control co-cultures in which CD2 was deleted. (I) qPCR of TLN2 mRNA of human Mito-DsRed BMSCs sorted following co-culture with either resting or activated human CD8⁺ T cells. Bars (mean ± s.e.m. of 3 healthy donors) relative to Actb. *P < 0.05 (one-way ANOVA, F, H; unpaired two-tailed Student’s t-test, I); **P < 0.01 (one-way ANOVA, F, H)

Figure 4:. Mitochondrial transfer enhances CD8⁺ T cell antitumor immunity against solid tumors.

Tumor size (A, mean ± s.e.m.) and survival curve (B) of sublethally irradiated B16_KVP tumor-bearing Ly5.2⁺ mice receiving 1.5 × 10⁵ Mito⁺ or Mito⁻ pmel-1 Ly5.1⁺CD8⁺ T cells generated as in Figure 2A in conjunction with recombinant human IL-2 (n = 5 mice/group). No Tx, no treatment (n = 4 mice). (C-D) Flow cytometry plot (C), frequency (D) of Mito⁺ and Mito⁻ pmel-1 Ly5.1⁺CD8⁺ T cells in the spleen 7 d after transfer. (E) Numbers of pmel-1 Ly5.1⁺CD8⁺ T cells per mg of tumor tissue, 7 d after treatment as in A,B. Data shown as mean ± s.e.m. (n = 4–5). (F, G) Confocal microscopy images of tumor sections stained with Ly5.1, and Hoechst 7 d after treatment with Mito⁺ pmel-1 Ly5.1⁺CD8⁺ T cells as in A, B. Arrows indicate tumor-infiltrating pmel-1 Ly5.1⁺ T cells retaining internalized Mito-Dsred signals. (H) Flow cytometry plot of Mito-DsRed⁺ cells in non-immune components (Ly5.1⁻Ly5.2⁻) of the tumor microenvironment. (I) Flow cytometry plot of Mito⁺ pmel-1 Ly5.1⁺CD8⁺ T cells in the tumor 7 d after transfer showing CTV signals versus Mito-DeRed. (J, K) Flow cytometry plot (J) and fold-change of Annexin V frequency (K) in Mito⁺ and Mito⁻ pmel-1 Ly5.1⁺CD8⁺ T cells in the tumor 7 d after transfer. Data shown as mean ± s.e.m. relative to Mito⁻ cells (n= 5–10, from two pooled independent experiments). *P < 0.05 (unpaired two-tailed Student’s t-test, A, D, E; log-rank [Mantel-Cox] test, B) **P < 0.01 (unpaired two-tailed Student’s t-test, K).

Figure 5:. Mitochondrial transfer counteracts CD8⁺ T cell exhaustion promoting effective effector responses.

(A, J) UMAP plot showing concatenated tumor-infiltrating (A) and splenic (J) pmel-1 Mito⁺ and Mito⁻ T cells 7d after tumor treatment as described in Figure 4A. (B, K) Density plot of pmel-1 Mito⁺ and Mito⁻ T cells as in A, J. (C, L) T cell subtype classification by ProjecTILs integrated with the Pauken proliferation signature dataset . (D, G, M) Violin plots showing the expression levels of exhaustion markers (D, M) and cytotoxic molecules (G) in each cluster. (E, F, H, I,N,O) Flow cytometry plots (E, H, N) and frequencies (F, I, O) of Mito⁺ and Mito⁻ pmel-1 Ly5.1⁺CD8⁺ T cells expressing the indicated combination of PD1 and LAG3 (F, O) or PD1 and Gzmb (I) 7d after treatment as in A, J. ns, not significant, *P < 0.05 (unpaired two-tailed Student’s t-test, F, I, O), *P <0.05; **P <0.01; ***P <0.001; ****P <0.0001 (Wilcoxon test followed by Benjamini Hochberg multiple-comparison test, D, G, M).

Figure 6:. Mitochondrial transfer enhances CD8⁺ T cell metabolism within the tumor microenvironment.

(A) Schematic representation of the SCENITH assay. Cells are incubated with DMSO (green), 2-DG (orange), oligomycin (purple), or harringtonine (grey) to inhibit the respective metabolic pathways before energy levels are read out by puromycin incorporation. 2-DG, 2-Deoxy-D-glucose; DMSO, Dimethyl sulfoxide; FA, fatty acids; AA, amino acids; FAAO, fatty acids and amino acids oxidation. (B) Histogram plots showing puromycin incorporation by splenic and intratumoral pmel-1 Mito⁺ and Mito⁻ T cells 7d after tumor treatment as described in Figure 4A (n= 5–9, from two pooled independent experiments) (C) Total metabolic capacity, (D) Glycolytic capacity and FAAO capacity (E) of splenic and intra-tumoral pmel-1 Mito⁺ and Mito⁻ T cells as assessed with SCENITH. (F, G) UMAP plots of splenic and intratumoral pmel-1 Mito⁺ and Mito⁻ T cells, color-coded by GSVA-based metabolic signatures . (H, I) Violin plots showing aerobic glycolysis (H) and OXPHOS (I) pathway scores. P = 0.0949; *P < 0.05, (unpaired one-tailed Student’s t-test, C).

Figure 7:. Mitochondrial transfer enhances human CD19-CAR CD8⁺ T cell antitumor immunity against systemic leukemia xenografts.

(A) Cytotoxicity assay using CD19-CAR Mito⁻ or Mito⁺ cells after co-culture with Mito-DsRed BMSCs that were left untreated or pretreated Ethidium bromide (EtBr) to render donor mitochondria dysfunctional. Data shows Green calibrated Unit (GCU) per μm²/image means ± s.e.m. after co-culture with NALM6-GL leukemia (E:T ratio 1:5) (n = 3 technical replicates per group). (B) Repetitive cytotoxicity assay using CD19-CAR Mito⁻ or Mito⁺ cells against NALM6-GL leukemia (E:T ratio 1:1). Data shows final GCU means ± s.e.m for each round of stimulation (n = 3 HD replicates per group) (C) Representative image of effector T cells:NALM6-GL leukemia co-cultures after six rounds of stimulation. Top row shows NALM6-GL cells alone. (D) Numbers of circulating NALM6-GL cells per 50 μl of blood 7 d after transfer of 1.25 × 10⁵ CD19-CAR Mito⁻ or Mito⁺ cells or CD19-CAR CD8 monocultured in conjunction with recombinant human IL-15 into sublethally irradiated NXG mice bearing NALM6-GL leukemia. (E) In vivo bioluminescent imaging and (F) survival of NALM6-GL-bearing NXG mice treated as in (C) (n = 4 or 6 mice/group). No Tx, no treatment (n = 6). (G) Confocal microscopy image of Mito-DsRed-BMSC pre-stained with MitoTracker DeepRed cocultured with MART-1TILs probed with anti-CD8-VioBlue antibody. (H) Flow cytometry plot of MART-1 TILs 48 hrs after co-culture with Mito-DsRed BMSCs. Numbers indicate the percentage of Mito-DsRed positive cells after gating on live lymphocytes. (I, J) Cytotoxicity assay using Mito⁻ or Mito⁺ MART-1 TILs against SK23-GFP melanoma (E:T ratio 1:5). Donor mitochondria were derived from immortalized (I) or primary (J) BMSCs. Data shows Target Green Object Area (TGOA) per μm²/image means ± s.e.m. (n = 3 technical replicates per group. Kruskal-Wallis test I; *P < 0.05 (Wilcoxon test, (A, I, J) 2-way ANOVA, Šídák’s multiple comparisons test (B) log-rank [Mantel-Cox] test, (F)); **P < 0.01 (2-way ANOVA, Šídák’s multiple comparisons test).