Transgelin 2 guards T cell lipid metabolism and antitumour function

Abstract

Mounting effective immunity against pathogens and tumours relies on the successful metabolic programming of T cells by extracellular fatty acids^1-3. Fatty-acid-binding protein 5 (FABP5) has a key role in this process by coordinating the efficient import and trafficking of lipids that fuel mitochondrial respiration to sustain the bioenergetic requirements of protective CD8⁺ T cells^4,5. However, the mechanisms that govern this immunometabolic axis remain unexplored. Here we report that the cytoskeletal organizer transgelin 2 (TAGLN2) is necessary for optimal fatty acid uptake, mitochondrial respiration and anticancer function in CD8⁺ T cells. TAGLN2 interacts with FABP5 to facilitate its cell surface localization and function in activated CD8⁺ T cells. Analyses of ovarian cancer specimens revealed that endoplasmic reticulum (ER) stress responses induced by the tumour microenvironment repress TAGLN2 in infiltrating CD8⁺ T cells, thereby enforcing their dysfunctional state. Restoring TAGLN2 expression in ER-stressed CD8⁺ T cells increased their lipid uptake, mitochondrial respiration and cytotoxic capacity. Accordingly, chimeric antigen receptor T cells overexpressing TAGLN2 bypassed the detrimental effects of tumour-induced ER stress and demonstrated therapeutic efficacy in mice with metastatic ovarian cancer. Our study establishes the role of cytoskeletal TAGLN2 in T cell lipid metabolism and highlights the potential to enhance cellular immunotherapy in solid malignancies by preserving the TAGLN2-FABP5 axis.

Extended Data Figure 1.. Analysis of FABP5 expression in T cells.

a, FACS histograms and quantitative analysis of lipid uptake (BODIPY 500/510) by CD8⁺ T cells in peripheral blood of cancer-free women (n = 6), or from peripheral blood (n = 4) and matched solid tumors (n = 4) of OvCa patients. b,c, Human naïve CD8⁺ T cells from peripheral blood of cancer-free women were activated via CD3/CD28 stimulation for 32 h, followed by 16 h in the absence or presence of 50% PBS. (b) Lipid uptake was assessed by FACS (n = 5). (c) FABP5, CD36, and FABP4 expression was determined via qRT-PCR and data were normalized to ACTB (n = 5). d, Schemes illustrating the strategy to detect cell surface-localized or total intracellular FABP5 in activated CD8⁺ T cells. FACS histograms depict FABP5 staining, with gray peaks representing isotype controls. e,f, Histograms and quantitative analysis of cell surface (e) and total intracellular (f) FABP5 expression determined by FACS in gated CD8⁺CD44⁺ T cells of the indicated genotypes (n = 4 per genotype). g-j, PBMCs from cancer-free women were cultured in the absence or presence of 50% HGSOC ascites supernatants for 16 h. FACS histograms and quantitative analysis of cell surface and total intracellular FABP5 expression in the indicated cell types (n = 6 per condition). k-m, WT female mice (n = 9) were intraperitoneally injected with ID8-Defb29/Vegfa OvCa cells and euthanized at the indicated times after tumor implantation (n = 3 per group). Experimental scheme and readouts (k). Representative images of peritoneal lavage (l) and omentum (m) from each group. Data are presented as mean ± s.e.m. a, One-way ANOVA with Tukey’s multiple comparisons test. b,g-j, Two-tailed paired Student’s t-test. c, Two-tailed unpaired Student’s t-test. e,f, Two-tailed unpaired Student’s t-test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 2.. Generation and characterization of mice selectively lacking TAGLN2 in T cells.

a, Description of the Talgln2 deletion strategy depicting floxed and deleted alleles. b,c, Deletion efficiency was analyzed in activated CD4⁺ or CD8⁺ T cells from Tagln2^fl/fl or Tagln2^fl/flCd4^Cre mice via qRT-PCR using a primer set that specifically detects the exon 3 region of Tagln2. Data were normalized to Actb (b). The intracellular levels of TAGLN2 were evaluated by FACS (c) (n = 3 per genotype). d, Histograms and quantitative analysis of TAGLN2 expression in the indicated cell types from the spleen and peripheral lymph nodes of Tagln2^f/f or Tagln2^f/f Cd4^cre mice (n = 3 per genotype). e, FACS plots and quantitative analysis of double negative (CD4⁻CD8⁻), double positive (CD4⁺CD8⁺), or single positive (CD4⁺ or CD8⁺) thymocytes frequencies and absolute number in the thymus (n = 5 per genotype). f, FACS plots and quantitative analysis of CD3⁺CD4⁺ or CD3⁺CD8⁺ T cell frequencies and absolute number in the spleen and peripheral lymph nodes (n = 5 per genotype). g,h, Expression of CD44 and CD62L on CD3⁺CD4⁺ (d) and CD3⁺CD8⁺ (e) T cells in the spleen and peripheral lymph nodes (n = 5 per group). FACS plots and quantitative analysis of the indicated cell populations and absolute number. Naïve (CD62L^highCD44^low), effector (CD62L^lowCD44^high) and Central memory (CD62L^highCD44^high). i,j, WT or TAGLN2-deficient naïve CD8⁺ T cells were activated via CD3/CD28 stimulation for 24 h. Histograms and quantitative analysis of CD44 (i) and Ki-67 (j) expression are shown (n = 3 per genotype). Data are presented as mean ± s.e.m. b-j, Two-tailed unpaired Student’s t-test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 3.. TAGLN2 is required for FABP5 surface expression and enables FABP5-dependent lipid uptake in activated CD8⁺ T cells.

a-c, WT or TAGLN2-deficient naïve CD8⁺ T cells were activated via CD3/CD28 stimulation for 24 h. Histograms and quantitative analysis of total FABP5 (a), surface FABP5 (b), and lipid uptake (c) by CD8⁺CD44⁺ T cells of the indicated genotypes are shown (n = 3 per genotype). d,e, WT naïve CD8⁺ T cells were activated via CD3/CD28 stimulation for 24 h and then Neon-electroporated with the indicated mRNAs, along with non-targeting (control) or Fabp5-specific siRNAs. Experimental scheme and readouts (d). Histograms and quantitative analysis of lipid uptake (BODIPY 500/510) by CD8⁺CD44⁺ T cells determined by FACS (n = 3 per condition) (e). f, WT, TAGLN2-deficient (Tagln2^KO) or FABP5-deficient (Fabp5^KO) naïve CD8⁺ T cells were activated via CD3/CD28 stimulation for 24 h and then Neon-electroporated with either control (Ctrl) or mouse Tagln2 mRNAs. After 48 h, lipidomic analyses were performed to evaluate the levels of diverse fatty acid species in CD8⁺ T cells. Quantification of the indicated fatty acids are shown (n = 3 per condition). Data are presented as mean ± s.e.m. a-c, Two-tailed unpaired Student’s t-test. e,f, One-way ANOVA with Tukey’s multiple comparisons test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 4.. Status of TAGLN2 in OvCa-infiltrating CD8⁺ T cells.

a, Volcano plot of differentially expressed genes in dysfunctional tumor-specific CD8⁺ T cells (TCR_TAG) compared to non-tumor-specific CD8⁺ T cells (TCR_OT1) from a mouse autochthonous liver tumor model (GSE126974). Selected differentially expressed genes with an adjusted P values <0.05 and Log₂ Fold Change > 1 or −1 are highlighted. (Two-sided Student’s t-test, P < 0.05, FC > 2). b, Relative expression dot plots of the indicated genes in different CD8⁺ T cell populations infiltrating human HGSOC tumors. The color of each dot represents the average normalized expression from high (red) to low (blue). The size of each dot represents the percentage (pct) of positive cells for each gene. c, FACS analysis of TAGLN2 expression by CD8⁺ T cells present in peripheral blood of cancer-free women (n = 14), malignant ascites (n = 15), or peripheral blood (n = 8) of HGSOC patients. d, TAGLN2 transcripts in the indicated CD8⁺ T cell populations from human peripheral blood mononuclear cells (PBMC). The resulting transcript expression values calculated as normalized transcript per million (nTPM), resulting from the internal normalized pipeline (The Human Protein Atlas; proteinatlas.org). e, TAGLN2 expression was quantified in the indicated CD8⁺ T cell populations from malignant ascites (n = 15). f, Correlation analysis for IFNG, TNFA or GZMB versus TAGLN2 mRNA in CD8⁺ T cells from ascites. Data were normalized to ACTB (n = 15). g, Correlation of IFN-γ concentration versus levels of TAGLN2 in the indicated CD8⁺ T cell subsets present in the ascites (n =14). h, Naïve peripheral CD8⁺ T cells from cancer-free women were activated via CD3/CD28 stimulation for 32 h and then incubated for 16 h with increasing amounts of HGSOC ascites supernatants (n = 5). Expression of TAGLN2, IFNG, and GZMB was assessed by qRT-PCR. Data were normalized to ACTB. i, FACS histograms and quantitative analysis of TAGLN2 expression in the indicated CD8⁺ T cells from peritoneal lavage of cancer-free female mice (n = 8; dotted lines) or mice bearing ID8-Defb29/Vegfa OvCa (n = 6; solid lines). Gray peaks denote isotype control staining. j,k, FACS analysis for CD44 (j) and Ki-67 (k) expression in the indicated T cells from the ascites of female mice bearing ID8-Defb29/Vegfa OvCa (n = 6 per group). l-n, Frequency of CD44⁺IFN-γ⁺ (l), CD44⁺TNF-α⁺ (m), and CD44⁺GZMB⁺ (n) cells within the indicated CD8⁺ T cells from the same mice described in (j) and (k). Data are presented as mean ± s.e.m. c,i-k, Two-tailed unpaired Student’s t-test. e,h, One-way ANOVA with Tukey’s multiple comparisons test. f,g, Spearman’s rank correlation test, Spearman coefficient (r) with exact P-value (two-tailed). l-n, Two-tailed paired Student’s t-test. Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 5.. Characterization of OvCa-associated T cells isolated from tumor-bearing mice devoid of TAGLN2 in T cells.

a-d, Female mice of the indicated genotypes bearing ID8-Defb29/Vegfa OvCa for 28 days. Peritoneal lavage, omentum, and spleen samples were harvested for immunophenotyping analyses. (a) Expression of TAGLN2 in CD45⁺CD3⁺CD8⁺ T cells. (b) Proportion of CD3⁺CD4⁺ and CD3⁺CD8⁺ T cells. (c) Proportion of naïve, effector, and memory CD3⁺CD8⁺ T cell populations based on CD44 and CD62L expression. (d) Proportion of PD-1⁺TOX⁺CD8⁺ T cells. FACS histograms and quantitative analyses are shown (n = 8 mice per group). Data are presented as mean ± s.e.m. a-d, Two-tailed unpaired Student’s t-test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 6.. Putative transcription factor binding sites in the Tagln2 promoter.

a, ECR browser analysis of the mouse and human Tagln2 locus is shown. The mouse genomic sequence was used as the base sequence on the x-axis. Schematic representation of the genomic positions of exons (E1–5) and putative binding sites of NF-kB and unfolded protein response (UPR) transcription factors in the Tagln2 promoter regions, mainly in the CNS1 region. Asterisks denote UPR transcription factors. UTR, untranslated region. b, Schematic representation of ER stress sensors and their corresponding downstream transcription factors. c-e, Naïve CD8⁺ T cells isolated from Eif2ak3^fl/fl or Eif2ak3^fl/flVav1^Cre mice (c), Atf6^fl/fl or Atf6^fl/flVav1^Cre mice (d), or Xbp1^fl/fl and Xbp1^fl/flCd4^Cre mice (e) were cultured under the indicated conditions. Expression of the Tagln2 transcript was determined by RT-qPCR, and data were normalized to endogenous levels of Actb in each sample (n = 3 per condition and genotype). f, Sequence of the mouse Tagln2 promoter from −646 to +138. XBP1s binding sites (Score > 10) are marked as BS1 and BS2. Location of ChIP-PCR primers (F/R) is indicated. g, Tagln2 promoter construct used for luciferase reporter assays. c-e, One-way ANOVA with Tukey’s multiple comparisons test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples.

Extended Data Figure 7.. Tagln2 overexpression in intratumoral CD4⁺ and CD8⁺ T cell subsets lacking XBP1s.

a, Scheme illustrating sample processing and experimental workflow. b, FACS sorting strategy for scRNA-sequencing. c, UMAP colored by genotype classifications of Xbp1^fl/fl (blue, n = 4,269 cells) and Xbp1^fl/flCd4^Cre (red, n = 3,325 cells). d, UMAP plot visualization of different T cell clusters colored by cell type. e, Heatmap showing the top 10 marker genes of the subclusters. f,g, Volcano plots showing differentially expressed genes (DEGs) with top 10 genes in CD4⁺ (f) or CD8⁺ (g) intratumoral T cell clusters from XBP1s-deficient compared to WT controls. Wilcoxon rank-sum test with Benjamini-Hochberg correction. (min_pct_expression > 10%, abs(avg_log2fc) > 0.2, −log10_pval_adj > 2). h, Enriched cellular pathways and functions in XBP1s-deficient CD8⁺ T cells at tumor sites. Z-scores greater than 2 indicate pathways and functions predicted to be significantly increased in XBP1s-deficient CD8⁺ T cells. i, Dot plot analysis showing the expression levels and distribution of seven major cytoskeletal genes (Tagln2, Wipf1, Wasf1, Wasf2, Hcls1, Was and Actr2) identified in each CD4⁺ and CD8⁺ T cell cluster in (d). j, Expression of previously reported RIDD target genes across the different CD3⁺ T cell clusters identified in (d).

Extended Data Figure 8.. Selective loss of XBP1s in T cells delays malignant progression and enhances TAGLN2, Ki-67, and CD44 expression in PPNM tumor-infiltrating CD8⁺ T cells.

a, Experimental scheme for mice of the indicated genotypes implanted with luciferase-expressing PPNM cancer cells. b,c, Assessment of peritoneal tumor burden over time in mice of the indicated genotypes (b) and quantification of bioluminescent signal for the same mice of the indicated genotype at different time points (n = 5 per genotype) (c). d, Overall survival curves for PPNM-bearing female mice of the indicated genotypes (n = 9 per genotype). e,f, Representative images of omentum (e) and solid tumors (f) from female mice of the indicated genotypes bearing PPNM-based HGSC for 40 days. Weight of omentum (e) and solid tumors (f) was determined in each group (Xbp1^fl/fl, n = 5; Xbp1^fl/flCd4 ^Cre, n = 8). g,h, Correlation of protein expression levels of TAGLN2 versus either Ki-67 or CD44 in the indicated intratumoral CD8⁺ T cell subsets in omentum (g) and solid tumor (h) from female mice of the indicated genotypes bearing PPNM-bearing HGSC for 40 days (Xbp1^fl/fl, n = 5; Xbp1^fl/flCd4 ^Cre, n = 8). Data are presented as mean ± s.e.m. c, One-way ANOVA with Tukey multiple comparisons test. d, Log-rank test for survival. e,f, Two-tailed unpaired Student’s t-test. g,h, Spearman’s rank correlation test, Spearman coefficient (r) with exact P-value (two-tailed). P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 9.. Selective suppression of FABP5 surface expression in ER-stressed CD8⁺ T cells.

a-e, WT naïve CD8⁺ T cells were stimulated via CD3/CD28 for 24 h and then electroporated with the indicated mRNAs. T cells were maintained under CD3/CD28 stimulation for an additional 32 h and then treated with the ER stressor TM for 16 h. (a) FACS histograms and quantitative analysis of TAGLN2, (b) lipid uptake, total and surface expression of (c) FABP5, (d) FABP4, and (e) CD36 in CD8⁺CD44⁺ T cells determined by FACS (n = 5 per condition). f-j, Pan B cells isolated from Tagln2^fl/fl or Tagln2^fl/flVav1^Cre mice were cultured under the indicated conditions (f). FACS histograms and quantitative analysis of TAGLN2 (g), surface FABP5 (h), total FABP5 (i) and lipid uptake (j) in CD19⁺ B cells. (n = 3 per condition). k-o, γδ T cells isolated from Tagln2^fl/fl or Tagln2^fl/flVav1^Cre mice were cultured under the indicated conditions (k). FACS histograms and quantitative analysis of TAGLN2 (l), surface FABP5 (m), total FABP5 (n) and lipid uptake (o) in CD3⁺γδ⁺ T cells. (n = 3 per condition). Data are presented as mean ± s.e.m. a-e,g-j,l-o, One-way ANOVA with Tukey multiple comparisons test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 10.. Induction of ER stress responses and progressive TAGLN2 loss in adoptively transferred CER T cells.

a, FSH-CER retroviral construct. b, Expression of the follicle-stimulating hormone receptor (FSHR) by PPNM cancer cells determined by immunoblot analysis. β-actin was used as loading control. c,d, CD45.1⁺CER T cells were isolated from the indicated tumor sites seven days after adoptive transfer into WT female mice developing PPNM HGSC. Experimental scheme (c). Xbp1s, Sec61a1, ERdj4, and Tagln2 expression in pre- or post-infusion CER T cells was determined via qRT-PCR. Data were normalized to Actb (n = 12) (d). e-h, FACS histograms and quantitative analysis of TAGLN2 and surface FABP5 expression in the indicated T cells isolated from peritoneal lavage (e) or omentum (f) at different days after adoptive transfer into WT female mice developing PPNM HGSC (D3: n = 6 per group, D7 and D14: n = 7 per group). Dark blue bar graph represents the expression of TAGLN2 and surface FABP5 in pre-infusion CER T cells (n = 1). FACS plots and quantitative analysis of CD44⁺IFN-γ⁺ (g) and CD44⁺GZMB⁺ (h) frequencies in the indicated CER T cells isolated from the same mice described in e and f (n = 6–7 per group). i,j, CD8⁺GFP⁺ sorted CER T cells were stimulated with recombinant chorionic gonadotropin alpha (CGα) in the absence or presence of TM (i). Xbp1s, Sec61a1, and Tagln2 expression was determined via qRT-PCR. Data were normalized to Actb (n = 3 per condition) (j). k, Experimental scheme for CER T cell killing assays. T cells were washed to remove TM and then cocultured with PPNM cancer cells at a 1:1 ratio. l, FACS plots and quantitative analysis of cancer cell death assessed by Annexin V and PI staining 18 h after incubation with mock-transduced CD8⁺ T cells (n = 3–4 per condition). Data are presented as mean ± s.e.m. d, Two-tailed paired Student’s t-test. e-h, One-way ANOVA with Tukey’s multiple comparisons test. j, Two-tailed unpaired Student’s t-test. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Extended Data Figure 11.. Therapeutic effects of CER-Tagln2 T cells in mice with OvCa.

a, FACS-based assessment of transduction efficiency using GFP expression as a marker. b, CER or CER-Tagln2 T cells were incubated in the presence or absence of TM for 16 h and TAGLN2 expression was then measured by FACS. c, Expression of the FSHR determined by immunoblot analysis in which GAPDH was used as loading control. d-g, Adoptive transfer of the indicated T cells into MP tumor-bearing female mice. (d) Experimental scheme. (e) Representative bioluminescence images of MP tumors over time and (f) quantification of peritoneal tumor burden in the indicated groups (n = 8 mice per group). (g) Overall survival curves for the same mice described in e and f. The initial number of mice allocated to each experimental group was n = 8, but only mice that developed OvCa were included in the survival analysis. h-j, WT female mice were challenged i.p. with PPNM cancer cells. After 8 days, mice were treated with either CER or CER-Tagln2 T cells alone or in combination with isotype control or anti–PD-1 antibodies as described in the Methods (n = 10 mice per group). (h) Experimental scheme and redouts. (i) Representative bioluminescence images of MP tumors over time. (j) Overall survival rates for the same mice described in i. The initial number of mice allocated to each experimental group was n=10, but only mice that developed OvCa were included in the survival analysis shown. k, Proposed model: ER stress responses induced by the TME disable a cytoskeletal-metabolic axis that maintains competent T cell anti-tumor function. TAGLN2 mediates optimal FABP5 surface localization to ensure robust T cell lipid uptake, fatty acid oxidation (FAO), mitochondrial respiration, and effector function. ER stress-driven IRE1α-XBP1s activation silences TAGLN2 in the OvCa TME. OvCa-reactive CER T cells overexpressing TAGLN2 bypass the detrimental effects of ER stress in the TME and show enhanced functionality, persistence, and anti-tumor activity. The figure was generated using BioRender. Data are presented as mean ± s.e.m. f, One-way ANOVA with Tukey’s multiple comparisons test. g,j, Log-rank test for survival. P<0.05 is considered statistically significant and exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Figure 1.. Defective lipid uptake and FABP5 surface localization in OvCa-infiltrating CD8⁺ T cells.

a, FACS histograms and quantitative analysis of lipid uptake (BODIPY 500/510) by CD8⁺ T cells in peripheral blood of cancer-free women (n = 14), malignant ascites (n = 15), or peripheral blood (n = 8) of OvCa patients. b,c, Human naïve CD8⁺ T cells from peripheral blood of cancer-free women were activated via CD3/CD28 stimulation for 32 h, followed by 16 h in the absence or presence of 50% HGSOC ascites supernatants. (b) Lipid uptake was assessed by FACS (n = 5). (c) FABP5, CD36, and FABP4 expression was determined via qRT-PCR and data were normalized to endogenous levels of ACTB (n = 5). d-f, Human naïve CD8⁺ T cells from peripheral blood of cancer-free individuals were activated via CD3/CD28 stimulation and then electroporated with the indicated mRNAs. Cells were expanded under the indicated treatments. FACS histograms and quantitative analysis of total FABP5 expression (d), lipid uptake (e), and cell surface FABP5 expression (f) in CD8⁺CD44⁺ T cells determined by FACS (n = 5 per group). g,h, FACS histograms and quantitative analysis of cell surface (g) and total (h) FABP5 expression in the indicated T cells isolated from the same specimens described in a. FACS-based kinetic analysis of surface FABP5 expression (i), total FABP5 expression (j), and lipid uptake (k) by the indicated T cells from peritoneal lavage or omentum at days 7, 14, or 28 after tumor implantation (n = 3 per group). Data are presented as mean ± s.e.m. a,d-h, One-way ANOVA with Tukey’s multiple comparisons test. b, Two-tailed paired Student’s t-test. c, Two-tailed unpaired Student’s t-test. Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Figure 2.. TAGLN2 is required for FABP5-mediated lipid uptake in activated CD8⁺ T cells.

a, Protein-protein interaction maps generated by the STRING database. b, Schematic of the immunoprecipitation coupled to mass spectrometry (IP-MS) approach. c, Relative intensity of FABP5-interacting proteins predicted in (a) after IP-MS analysis. N.D, not detected. d, Interaction between FABP5 and TAGLN2 in wild type (WT) activated CD8⁺ T cells. Representative image from three independent experiments is shown. e,f, Confocal images of activated CD8⁺ T cells of the indicated genotypes stained for TAGLN2 (green),FABP5 (red), and DAPI (blue). (e). The percentage of colocalization of the indicated markers is shown in (f) (Manders’ coefficient). g, WT naïve CD8⁺ T cells were activated via CD3/CD28 stimulation for 24 h and then Neon-electroporated with indicated mRNAs. Representative histograms and quantitative analysis of surface FABP5 expression in CD8⁺CD44⁺ T cells determined by FACS (n = 3 per condition) 48 h later. Gray peak represents isotype control staining. h, Experimental scheme and readouts. i, Histograms and quantitative analysis of lipid uptake (BODIPY 500/510) in CD8⁺CD44⁺ T cells determined by FACS (n = 3 per condition). j,k, Relative abundance (% of total fatty acid content) of the indicated fatty acids (j),and quantification of palmitoleic acid and oleic acid (k) in the indicated CD8⁺ T cells (n = 3 per condition). l, Relative fatty acid abundance in malignant ascites supernatants from mice bearing advanced ID8-Defb29/Vegfa OvCa. m,n, Relative abundance (m) and quantification (n) of oleic acid in CD8⁺ T cells of the indicated groups (n = 3 per condition). The same control data shown in (k) was used. Data are presented as mean ± s.e.m. f,g, Two-tailed unpaired Student’s t-test. i,k,n, One-way ANOVA with Tukey’s multiple comparisons test. Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Figure 3.. XBP1s restrains TAGLN2 expression in ER-stressed CD8⁺ T cells.

a, CD8⁺ T cells were treated as indicated. Xbp1s and Tagln2 expression was determined by qRT-PCR. Data were normalized to Actb (n = 3 per condition). b, TAGLN2 levels determined by FACS (n = 3 per group) in the same samples described in (a). c, Representative images for TAGLN2 expression in CD8⁺ T cells treated as indicated. d-g, CD8⁺ T cells of the indicated genotypes were exposed to various ER stressors and Tagln2 expression was determined by RT-qPCR. Data were normalized to Actb (d,f) (n = 3 per condition and genotype). FACS analysis of TAGLN2 expression in CD8⁺CD44⁺ T cells under the indicated conditions (e,g) (n = 3 per genotype). h, Xbp1s and Tagln2 levels determined by RT-qPCR. Data were normalized to Actb (n = 3 per condition). i, Conserved XBP1s-binding motifs. j, ChIP-PCR assays for XBP1s occupancy at the Tagln2 promoter. No XBP1s binding region (NR) was used as a negative control (n = 3 per condition). k, Luciferase reporter assays denoting repression of the Tagln2 promoter by XBP1s (n = 3). l,m, FACS analysis of TAGLN2 expression in T cells from the omentum (l) and solid tumors (m) from mice of the indicated genotypes bearing PPNM-based HGSC for 40 days (Xbp1^fl/fl, n = 5; Xbp1^fl/flCd4 ^Cre, n = 8). Gray peaks denote isotype control staining. n, UMAP plot of T cell subtypes from 11 human HGSOC specimens. o, GSEA enrichment plots for CD8⁺ tumor-infiltrating effector memory T cells (T_EM). NES, normalized enrichment score. FDR, false discovery rate. p,q, Correlation analysis for XBP1s versus TAGLN2 expression (n = 16) (p) and XBP1s versus TAGLN2 expression (q) in the indicated CD8⁺ T cells from the ascites of OvCa patients (n =14). Data are presented as mean ± s.e.m. a,b,d,f,h,k, One-way ANOVA with Tukey’s multiple comparisons test. e,g,j,l,m, Two-tailed unpaired Student’s t-test. p,q, Spearman’s rank correlation test, Spearman coefficient (r) with exact P-value (two-tailed). Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Figure 4.. TAGLN2 overexpression rescues the bioenergetic defects of ER-stressed CD8⁺ T cells.

a, Pre-activated CD8⁺ T cells were exposed to TM for 16 h and cell surface FABP5 expression was analyzed (n = 6 per condition). b, Pre-activated CD8⁺ T cells were treated with the indicated ER stressors and lipid uptake (BODIPY 500/510) by CD8⁺CD44⁺ T cells (n = 3 per condition) was analyzed. c,d, Pre-activated CD8⁺ T cells of the indicated genotypes were exposed to TM for 16 h and cell surface FABP5 expression (c) and lipid uptake (d) in CD8⁺CD44⁺ T cells (n = 3 per condition) were determined. e, OCR plots and quantification of basal and maximal respiration in CD8⁺ T cells exposed to TM. Data were normalized to total DNA content in each condition (n = 5). f-j, Pre-activated CD8⁺ T cells were electroporated with the indicated mRNAs and then exposed to TM. Lipid uptake in CD8⁺CD44⁺ T cells was determined by FACS (n = 3 per condition) (f). OCR plots and quantification of basal and maximal respiration normalized to total DNA content (n = 5 per condition) (g). Expression of CD44 and Ki-67 (n = 9 per group) (h). Proportion of CD44⁺IFN-γ⁺ (i) and CD44⁺GZMB⁺ (j) CD8⁺ T cells (n = 5 per group). k, OCR of TM-exposed Tagln2-overexpressing CD8⁺ T cells treated as indicated. Maximal respiration was normalized to total DNA content (n = 5 per condition). l, Levels of CD44 and Ki-67 in TM-treated Tagln2-overexpressing CD8⁺ T cells in the absence or presence of oleic acid (OA, 30 μM) (−OA: n = 6, +OA: n = 3). m, FACS analysis of mitochondrial membrane potential in CD8⁺ T cells treated as indicated (n = 4 per condition and genotype). Data are presented as mean ± s.e.m. a,f-l, Two-tailed paired Student’s t-test. c,d, Two-tailed unpaired Student’s t-test. b,m, One-way ANOVA with Tukey’s multiple comparisons test. Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.

Figure 5.. Preserving TAGLN2 enhances the therapeutic effects of CER T cells in HGSOC.

a, FACS analysis of cancer cell death assessed by Annexin V and PI staining (PPNM only: n = 4; n = 3 per condition). b, Schematics of retroviral CER expression constructs. FSHβ, follicle-stimulating hormone beta subunit. CGα, chorionic gonadotropin alpha subunit. IRES, internal ribosome entry site. GFP, green fluorescent protein. c, Experimental scheme for adoptive transfer of CER or CER-Tagln2 T cells into PPNM tumor-bearing mice. d-i, Expression of TAGLN2 (d), proportion of CD62L^hiCD44^hi cells (e), surface and total FABP5 expression (f), and proportions of CD44⁺IFN-γ⁺ (g), CD44⁺TNF-α⁺ (h), and CD44⁺GZMB⁺ (i) cells were assessed in the indicated CER T cell populations from peritoneal lavage or omentum at day 21 of tumor development (7 days after the second T cell infusion). (d-f, Peritoneal lavage; n = 8 per group, Omentum; n = 3 (CER T cells), n = 8 (CER-Tagln2 T cells), g,i, n = 8 mice per group). Gray peaks denote isotype control staining. j,k, Peritoneal carcinomatosis in female mice bearing PPNM-based HGSC and treated with the indicated CER T cells. Representative bioluminescence images of PPNM tumors over time (j) and quantification of peritoneal tumor burden (k) in the indicated groups (n = 16 mice per group). l,m, Representative images of omentum (l), and quantification of CD3⁺ T cell infiltration into omental samples (m) in female mice bearing PPNM tumors (n = 8 per group). n, Overall survival analysis for mice described in (j,k) (n = 16 per group). Data are presented as mean ± s.e.m. a,k, One-way ANOVA with Tukey’s multiple comparisons test. d-i, Two-tailed unpaired Student’s t-test. m, Fisher’s exact test was used to determine the association between the two categorical variables. n, Log-rank test for survival. Exact P-values are shown. The ‘n’ values represent biologically independent samples. gMFI, Geometric mean fluorescence intensity.