The cytokine Meteorin-like inhibits anti-tumor CD8+ T cell responses by disrupting mitochondrial function

Abstract

Tumor-infiltrating lymphocyte (TIL) hypofunction contributes to the progression of advanced cancers and is a frequent target of immunotherapy. Emerging evidence indicates that metabolic insufficiency drives T cell hypofunction during tonic stimulation, but the signals that initiate metabolic reprogramming in this context are largely unknown. Here, we found that Meteorin-like (METRNL), a metabolically active cytokine secreted by immune cells in the tumor microenvironment (TME), induced bioenergetic failure of CD8⁺ T cells. METRNL was secreted by CD8⁺ T cells during repeated stimulation and acted via both autocrine and paracrine signaling. Mechanistically, METRNL increased E2F-peroxisome proliferator-activated receptor delta (PPARδ) activity, causing mitochondrial depolarization and decreased oxidative phosphorylation, which triggered a compensatory bioenergetic shift to glycolysis. Metrnl ablation or downregulation improved the metabolic fitness of CD8⁺ T cells and enhanced tumor control in several tumor models, demonstrating the translational potential of targeting the METRNL-E2F-PPARδ pathway to support bioenergetic fitness of CD8⁺ TILs.

Keywords: CD8(+) T cell hypofunction; Meteorin-like protein; T cell exhaustion; anti-tumor immunity; mitochondrial dysfunction.

Figure 1:. METRNL is differentially upregulated in TILs.

Differential expression analysis of experienced CD8⁺TIL samples and activated CD8 PBMC samples for GBM, PRAD, RCC and BLCA cohorts. (A) After FDR adjustments, genes that meet an FDR <= 0.05 are highlighted. Genes that meet an FDR <= 0.00001 are labeled. Red indicates higher expression while blue indicates lower expression in TIL compared with PBMC. (B) GSEA pathways analysis for differentially expressed genes based on the Biological Hallmarks dataset. Gene sets were selected if FDR < 0.01 in one of the tumor types. The color scale is log10(FDR), with orange indicating higher expression in CD8 TILs vs. activated PBMCs and blue indicating higher expression in activated PBMCs vs. CD8 TILs. (C) Estimation of the underlying high expression and low expression distributions via expectation maximization. (D) Venn diagrams displaying significantly differentially expressed (FDR ≤ 0.01) genes across tissues for triple positive experienced CD8 TIL contrasted against all other samples (left) and significantly differentially expressed (FDR ≤ 0.0001) genes across tissues for triple positive experienced CD8⁺ TIL contrasted against triple positive activated PBMC CD8⁺ samples (right). (E) Expression of METRNL in CD8 T cell subsets from murine scSeq data by Andreatta et al. and (F) from human prostate cancer scSeq data by Chen et al. See also Figure S1.

Figure 2:. METRNL is an immunosuppressive cytokine secreted during tonic signaling.

(A) ELISA measurement of METRNL and (B) IFN-γ secretion by immune checkpoint-expressing CD8⁺ TILs sorted from orthotopic GL261 gliomas (from pooled and sorted TILs, n = 7 biological replicates were plated of checkpoint-negative TILs, n = 4 for PD-1⁺LAG-3⁻TIM-3⁻ TILs and n = 4 for PD-1⁺LAG-3⁻TIM-3⁺ TILs. n = 3 biological replicates of DCs only w/o TILs). (C) ELISA measurement of METRNL during in vitro activation and re-activation of CD8 T cells (n = 5-6 biological replicates for each treatment condition). (D) METRNL expression from bulk RNA-seq of in vitro day 10 constitutive and regulatable (SNIP) CAR-T cells targeting B7H3, GD2 or HER2 (n = 1 biological replicate). (E) METRNL expression on day 7 and 11 from CAR-T cells without tonic signaling (Always OFF), with tonic signaling (Always ON), with tonic signaling until indicated day followed by rest (OFF from day 7) (n = 3 biological replicates for each condition). Graphs show the mean +/− SEM. Statistics were calculated by one-way ANOVA with Tukey’s multiple-comparisons post hoc test (A, B, C, E right panel), two-tailed Student’s t test (E left panel). *p ≤ 0.05; **p ≤ 0.01; ****p ≤ 0.0001; ns = not significant. See also Figure S2.

Figure 3:. Metrnl ablation enhances CD8 TIL effector function and viability and delays tumor growth.

(A) Schematic for paracrine suppression assay using WT and Metrnl^−/− T cells (B) Flow cytometry assessment of in vitro proliferation of CFSE-labeled naive CD8 T cells co-cultured with pre-activated WT or pre-activated Metrnl^−/− T cells in the paracrine suppression assay (n = 5 biological replicates for each condition). (C) Survival of Metrnl^−/− and WT mice with orthotopic GL261 glioma (WT, n = 7; KO, n = 6), (D) B6CaP prostate cancer flank tumors (WT, n = 10; KO, n = 9), (E) MC38 colorectal cancer flank tumors (WT, n = 9; KO, n = 6). (F) Tumor growth in WT and Metrnl^−/− mice with MC38 flank tumors in response to CD4 and CD8 depletion (WT + isotype, n = 6; WT + anti-CD8, n = 6; WT + anti-CD4, n = 7; KO + isotype, n = 7; KO + anti-CD4, n = 5; KO + anti-CD8, n = 5). (G) Flow cytometry analysis of TILs from MC38 flank tumors harvested from WT and Metrnl^−/− mice (n = 6 mice per group). (H) Survival and growth curves of MC38 flank tumors following IV injection of siRNA against Metrnl (scramble siRNA treated mice, n = 10; Metrnl siRNA treated, n = 9). (I) MC38-OVA flank tumor growth curves with adoptive transfer of WT OT-1 and Metrnl^−/− OT-1 cells (OT-1 transfer, n = 9; Metrnl^−/− OT-1 transfer, n = 7). (J) Schematic of adoptive transfer of CFSE-labeled OT-1 and Metrnl^−/− OT-1 T cells to MC3 8-OVA tumor bearing WT and Metrnl^−/− recipients. (K) Flow cytometry assessment of in vivo proliferation of CFSE-labeled OT-1 or Metrnl^−/− OT-1 T cells in tumors of WT or Metrnl^−/− mice (WT receiving OT-1 T cells, n = 5; WT receiving Metrnl^−/− OT-1 T cells, n = 5, Metrnl^−/− receiving OT-1 T cells, n = 4; Metrnl^−/− receiving Metrnl^−/− OT-1 T cells, n = 5). Differences in survival were calculated by the Mantel-Cox log-rank test. Graphs show the mean +/− SEM. Statistics for tumor volumes were calculated with 2-way ANOVA with Sidak’s multiple-comparisons test. Statistics were calculated using two-tailed Student’s t test (B, G) and and one-way ANOVA with Tukey’s multiple-comparisons post hoc test (K). *p ≤ 0.05; **p ≤ 0.01; ****p ≤ 0.0001; ns, not significant. See also Figure S3.

Figure 4:. Exogenous rMETRNL depolarizes mitochondria in CD8 T cells.

Flow cytometry analysis of (A) mitochondrial potential-dependent dye CMXRos (n = 4 biological replicates) and (B) mitochondrial potential-independent dye Mitotracker green (n = 5 biological replicates) in naïve CD8 T cells treated with exogenous rMETRNL during activation with PMA/ionomycin for 5-6 hours. (C) Transmission electron microscopy (TEM) imaging of CD8 T cells activated with anti-CD3/CD28 beads with and without rMETRNL (top panels show representative cells from each group and bottom panels show magnified image of the mitochondria from each condition). (D) Quantification of mitochondria from TEM imaging of T cells activated with anti-CD3/CD28 beads in the presence of rMETRNL for 2 days and rested for 1 or 2 days without rMETRNL. (E) Flow cytometry analysis of mitochondrial potential-dependent dye Mitotracker DeepRed and MTG (n = 6 biological replicates) in WT and Metrnl^−/− CD8 T cells treated with FCCP, followed by washing and stimulation with PMA/ionomycin, and quantification of mitochondrial parameters. (F) Flow cytometry analysis of ROS MFI and positivity in CD8 T cells treated with escalating doses of rMETRNL during activation with PMA/ionomycin (n = 4 biological replicates for each dose). (G) 53BP1 staining of naïve CD8⁺ T cells activated with anti-CD3/anti-CD28 beads in the presence of rMETRNL. For each treatment, 3 to 4 images were quantified, and the experiment was repeated four times and statistical analysis incorporating all repeats were determined. CD8 T cells with 53BP1 loci greater than 4 per cell were quantified. (H) 53BP1 foci quantification with rMETRNL exposure and following recovery. (I) Representative images of Apopxin and Hoescht staining of activated CD8⁺ T cells with and without rMETRNL treated at indicated timepoints. For each treatment three to four images were quantified. Graphs show the mean +/− SEM. Statistics were calculated by one-way ANOVA with Tukey’s multiple-comparisons post hoc test (A,B,F) and Sidak’s correction (D, H), two-tailed Student’s t test (G) and 2-way ANOVA with Holm-Sidak correction (E,I).*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; ns = not significant. See also Figure S4.

Figure 5.. METRNL alters CD8 T cell respiration and induces compensatory glycolysis.

(A) Real time oxygen consumption rates (OCR) were measured under basal conditions and following addition of oligomycin, FCCP and rotenone/antimycin in untreated and rMETRNL treated Jurkat T cells (n = 7 biological replicates). (B) Real OCR were measured under basal conditions and following addition of oligomycin, FCCP and rotenone/antimycin in WT and Metrnl^−/− CD8⁺ T cells in reactivation condition (n = 7 biological replicates for WT and Metrnl^−/− CD8 T cells). (C) Heatmap visualization of the top 25 metabolite changes between in vitro reactivated CD8 T cells with and without rMETRNL, measured by LC-MS. (D) Relative quantities of metabolites related to glycolytic flux in CD8 T cells with and without rMETRNL (n = 5 biological replicates for CD8 T cells and n = 5 biological replicates for CD8 T cells with rMETRNL). (E) Volcano plot of metabolites plotting log2 fold change versus −log 10 (FRD-corrected p value), with colored dots representing significant metabolite changes. (F) Heatmap visualization of the top 40 metabolite changes between in vitro reactivated WT and Metrnl^−/− CD8⁺ T cells, measured by LC-MS. (G) Relative quantities of metabolites related to glycolytic flux in WT and Metrnl^−/− CD8 T cells (n = 3 biological replicates for WT and n = 4 biological replicates for Metrnl^−/− CD8 T cells). (H) Volcano plot of metabolites plotting log2 fold change versus −log 10 (FRD-corrected p value), with colored dots representing significant metabolite changes. (I) Extracellular acidification rates (ECAR) were measured under basal conditions and following addition of glucose, oligomycin and 2-DG in untreated and rMETRNL CD8 T cells (n = 7 biological replicates). (J) Real time OCR were measured under basal conditions and following addition of oligomycin, FCCP and rotenone/antimycin in WT and Metrnl^−/− CD8 T cells in glucose-deficient condition (n = 9 biological replicates for WT and Metrnl^−/− CD8 T cells). (K) Real time OCR were measured under basal conditions and following addition of rMETRNL, oligomycin, FCCP and rotenone/antimycin in CD8 T cells (mitochondrial stress test, n = 10 biological replicates). (L) Real time ECAR were measured under basal conditions and following addition of rMETRNL, glucose, oligomycin and 2-DG in CD8 T cells (glycolytic stress test, n = 10 biological replicates). Graphs show mean +/− SEM. Statistics were calculated by Student’s t test for metabolomics assay (C,E,F,H) *p ≤ 0.1; **p ≤ 0.01; ***p ≤ 0.001; two-tailed Student’s t test (A,B,I,J), and one-way ANOVA with Tukey’s multiple-comparisons post hoc test (K,L) *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; ns = not significant. See also Figure S5.

Figure 6.. METRNL signals via the E2F-PPARδ pathway.

(A) Volcano plot of all differentially expressed genes between reactivated WT andMetrnl^−/− CD8⁺ T cells (n = 3 biological replicates). (B) Results of GSEA Hallmark analysis showing top 25 enriched gene sets at FDR < 25% and a nominal p value = 0. A positive Normalized Enrichment Score (NES) value indicates enrichment in Metrnl^−/− CD8⁺ T cells, a negative NES indicates enrichment in WT CD8⁺ T cells. (C) GSEA showing enrichment of the Reactome gene sets of MAPK signaling pathway, Metabolism, HIF-1α hydroxylation and DAP12 signaling. (D) Z-values of E2F motifs from ISMARA analysis of bulk RNA sequences (E) Luciferase-based luminescence from CD8 T cell containing a plasmid with luciferase construct downstream of PPARD promoter and treated with rMETRNL (F) Assessment of DNA-binding activity of PPARδ using ELISA based detection of PPARδ bound to immobilized peroxisome proliferator response element, in a dose dependence and (G) time course manner. (H) Luciferase-based luminescence of cells containing a luciferase construct downstream of PPRE, and treated with rMETRNL in a dose dependence and (I) time course manner. (J) Real time OCR were measured under basal conditions and following addition of oligomycin, FCCP and rotenone/antimycin in WT and Metrnl^−/− CD8⁺ T cells with PPARδ agonist GW0742 and maximum respiration and spare respiratory capacity were compared among the different groups. (K) Real time OCR of WT and Metrnl^−/− CD8⁺ T cells with increasing doses of PPARδ antagonist GSK3787 were measured, and maximum and spare respiratory capacity were compared. Multiple hypothesis test corrections with BH were used to obtain FDR values. Statistics were calculated with one-way ANOVA with Fisher’s LSD test (J, K). *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; ns = not significant. See also Figure S6.