Soluble Tim-3 serves as a tumor prognostic marker and therapeutic target for CD8+ T cell exhaustion and anti-PD-1 resistance

Abstract

Resistance to PD-1 blockade in onco-immunotherapy greatly limits its clinical application. T cell immunoglobulin and mucin domain containing-3 (Tim-3), a promising immune checkpoint target, is cleaved by ADAM10/17 to produce its soluble form (sTim-3) in humans, potentially becoming involved in anti-PD-1 resistance. Herein, serum sTim-3 upregulation was observed in non-small cell lung cancer (NSCLC) and various digestive tumors. Notably, serum sTim-3 is further upregulated in non-responding patients undergoing anti-PD-1 therapy for NSCLC and anti-PD-1-resistant cholangiocarcinoma patients. Furthermore, sTim-3 overexpression facilitates tumor progression and confers anti-PD-1 resistance in multiple tumor mouse models. Mechanistically, sTim-3 induces terminal T cell exhaustion and attenuates CD8⁺ T cell response to PD-1 blockade through carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM-1). Moreover, the ADAM10 inhibitor GI254023X, which blocks sTim-3 production, reduces tumor progression in Tim-3 humanized mice and reverses anti-PD-1 resistance in human tumor-infiltrating lymphocytes (TILs). Overall, human sTim-3 holds great predictive and therapeutic potential in onco-immunotherapy.

Keywords: ADAM10; CEACAM-1; HCC; ICC; T cell exhaustion; anti-PD-1 therapy; hepatocellular carcinoma; intrahepatic cholangiocarcinoma; lung cancer; resistance to PD-1 blockade; sTim-3.

Graphical abstract

Figure 1

Serum sTim-3 is a potential biomarker of tumor progression and anti-PD-1 resistance (A) The serum sTim-3 concentrations in NSCLC patients (N = 65) and healthy people (N = 23) were determined by ELISA. (B) The serum sTim-3 concentrations in lung squamous cell carcinoma (LUSC) patients (N = 12), lung adenocarcinoma (LUAD) patients (N = 53), and healthy individuals (N = 23) were determined using ELISA. (C) The serum sTim-3 concentrations in different subtypes of lung adenocarcinoma, including adenocarcinoma in situ (AIS) (N = 10), minimally invasive adenocarcinoma (MIA) (N = 11), and invasive adenocarcinoma (IAC) (N = 32), as well as healthy individuals (N = 23) were determined using ELISA. (D) The serum concentrations of sTim-3 in LUAD patients at early (N = 10) and advanced stages (N = 43) according to the tumor/node/metastasis (TNM) classification system. (E) Correlation of serum sTim-3 concentrations with in situ tumor size in patients with LUAD. (F) Kaplan-Meier survival curve of NSCLC patients with high (N = 32) versus low (N = 32) sTim-3 expression. (G) sTim-3 concentrations in the serum of anti-PD-1-treated (N = 67) or non-treated (N = 54) NSCLC patients were evaluated using ELISA. (H) sTim-3 concentrations in the serum of NSCLC patients post anti-PD-1 therapy, including 16 responders and 26 non-responders, were evaluated using ELISA. (I) The serum sTim-3 concentrations in healthy individuals (N = 28), HCC patients (N = 20), gastric cancer patients (N = 16), colorectal cancer patients (N = 15), and cholangiocarcinoma (N = 9) patients. Data are presented as means ± SEM and were analyzed by the two-tailed unpaired Student’s t test (A, B, C, D, G, H, and I), linear regression (E) and log rank test (F). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 2

sTim-3 promotes tumor progression in multiple mouse models (A–H) Overexpression of sTim-3 in the AKT/c-Myc-induced intrahepatic HCC models. (A) Schematic diagram of human sTim-3-overexpressing or control HCC model. (B) Concentrations of serum sTim-3 in HCC model (n = 5). (C) Representative images and (D) quantification of bioluminescence in the liver at four weeks post tumor induction (n = 6). (E) Liver-to-body weight ratio (n = 5). (F) Gross appearances of the liver. (G) H&E staining and Ki67 immunohistochemistry staining of the intact liver tumors. (H) Survival curve (n = 6). (I–P) Overexpression of sTim-3 in the intrahepatic AKT-NICD1-induced ICC models. (I) Schematic diagram of sTim-3-overexpressing or control ICC model. (J) Concentrations of serum sTim-3 in ICC model (n = 5). (K) Representative images and (L) quantification of bioluminescence in the liver at four weeks post tumor induction (n = 7). (M) Liver-to-body weight ratio (n = 9). (N) Gross appearances of the liver. (O) H&E staining and Ki67 immunohistochemistry staining of the intact liver tumors. (P) Survival curve (Vector, n = 6; sTim-3, n = 7). (Q–T) Control (LV-Vector) or sTim-3-overexpressing (LV-sTim-3) B16-MO5 melanoma cells were subcutaneously injected into wild-type and Tim-3 knockout mice , separately. (Q) Schematic diagram of the B16-MO5 tumor model. (R) Concentrations of serum sTim-3. (S) Tumor growth curve (wild-type: LV-Vector, n = 5; LV-sTim-3, n = 4;Tim-3 knockout: LV-Vector, n = 4; LV-sTim-3, n = 4). (T) Survival curve (wild-type: LV-Vector, n = 7; LV-sTim-3, n = 8;Tim-3 knockout: LV-Vector, n = 4; LV-sTim-3, n = 4). All results are representative of at least three independent experiments. Data are presented as mean ± SEM and were analyzed by the unpaired Student’s t test (B, D, E, J, L, and M), two-way analysis of variance (ANOVA) (S), and log rank test (H, P, and T). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Figure 3

sTim-3 overexpression aggravates T cell exhaustion in tumor microenvironments (A–D) Tumor-infiltrating lymphocytes were isolated from control (n = 5) and sTim-3-overexpression (n = 5) HCC mice on day 30 post-HDI. (A) Schematic diagram of the immune phenotype analysis. (B) The number of CD8⁺ T cells in tumors. (C) Percentages of IFN-γ⁺, TNF-α⁺, and IFN-γ⁺TNF-α⁺ CD8⁺ TILs. (D) Percentage of PD-1⁺, Tim-3⁺, and PD-1⁺Tim-3⁺TCF1⁻ (T_EX) CD8⁺ T cells in tumor. (E–H) Tumor-infiltrating lymphocytes were isolated from control (n = 5) and sTim-3-overexpression (n = 5) ICC mice on day 30 post-HDI. (E) Schematic diagram of the immune phenotype analysis. (F) The number of CD8⁺ T cells in tumors. (G) Percentages of IFN-γ⁺, TNF-α⁺, and IFN-γ⁺TNF-α⁺ CD8⁺ TILs. (H) Percentage of PD-1⁺, Tim-3⁺, Tox⁺, and PD-1⁺Tim-3⁺Tox⁺TCF1⁻ (T_EX) tumor-infiltrating CD8⁺ T cells. (I–L) Tumor-infiltrating lymphocytes were isolated from control and sTim-3-overexpression B16-MO5 tumor control on day 26 post-tumor inoculation. (I) Schematic diagram of the immune phenotype analysis. (J) The number of CD8⁺ T cells per gram of tumor tissue from control (n = 5) and sTim-3-overexpressing (n = 8) B16-MO5 tumor-bearing mice. (K) Percentages of IFN-γ⁺, TNF-α⁺, and IFN-γ⁺TNF-α⁺ CD8⁺ TILs from control (n = 5) and sTim-3-overexpressing (n = 5) B16-MO5 tumors were measured. (L) Percentage of PD-1⁺, Tim-3⁺, LAG-3⁺, Tox⁺, and PD-1⁺Tim-3⁺Tox⁺TCF1⁻ (T_EX) tumor-infiltrating CD8⁺ T cells from control (n = 5) and sTim-3-overexpressing (n = 5) B16-MO5 tumor mice. All results are representative of at least three independent experiments. Data are presented as mean ± SEM and were analyzed by the unpaired Student’s t test. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Figure 4

sTim-3 abrogates the therapeutic effects of anti-PD-1 treatment (A) Schematic diagram of HCC mice or ICC mice treated with isotype control antibodies (IgG) or anti-PD-1 (α-PD-1). (B–D) (B) Representative images and (C) quantification of bioluminescence in the liver of HCC mice at 4 weeks post-HDI (n = 6). (D) Survival curve of HCC mice (Vector + IgG n = 7; Vector + anti-PD-1 n = 7; sTim-3 + IgG n = 9; sTim-3 + anti-PD-1 n = 8). (E–G) (E) Representative images and (F) quantification of bioluminescence in the liver of ICC mice at 3 weeks post-HDI (Vector + IgG n = 13; Vector + anti-PD-1 n = 10; sTim-3 + IgG n = 9; sTim-3 + anti-PD-1 n = 9). (G) Survival curve of ICC mice (Vector + IgG n = 8; Vector + anti-PD-1 n = 7; sTim-3 + IgG n = 10; sTim-3 + anti-PD-1 n = 9). All results are representative of at least two independent experiments. Data are presented as mean ± SEM and were analyzed by the Mann-Whitney U test (C and F) and log rank test (D and G). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; ns, no significance.

Figure 5

sTim-3 induces T cell dysfunction and reduces T cell anti-PD-1 response (A and B) WT (A, n = 7) or Tim-3 KO (B, n = 3) OT-Ⅰ CTLs were pretreated with sTim-3 protein (5 μg/mL) and stimulated with anti-CD3/CD28 antibodies for 6 h. IFN-γ, TNF-α, and IL-2 production was assessed using flow cytometry. (C) Cytotoxicity of control (RV-Vector) and sTim-3-overexpressing (RV-sTim-3) OT-Ⅰ CTLs against firefly luciferase-expressing B16F10 targets (n = 3). (D) Exhausted OT-Ⅰ CTLs were induced by chronic stimulation in the presence or absence of sTim-3. Percentages of PD-1⁺Tim-3⁺TOX⁺TCF1⁻ CD8⁺ T_EX cells were assessed using flow cytometry (n = 5). (E) OT-Ⅰ CTLs were pretreated with IgG or anti-PD-1 antibodies (5 μg/mL) in the presence of sTim-3 protein (5 μg/mL) or not and stimulated with anti-CD3/CD28 antibodies for 6 h. IFN-γ and TNF-α production were assessed using flow cytometry (n = 6). (F) TILs from HCC patients were pretreated with sTim-3 protein (5 μg/mL) and stimulated with anti-human CD3/CD28 beads. TNF-α and IFN-γ in supernatants were detected using ELISA 24 h later (n = 6). (G) TILs from HCC patients were pretreated with sTim-3 protein (5 μg/mL) and anti-PD-1 antibodies (5 μg/mL) and stimulated with anti-human CD3/CD28 beads. IFN-γ in supernatants was detected using ELISA 24 h later (n = 3). (H and I) Gene set enrichment analysis (GSEA) was performed to determine the specific enrichment in T cell activation signature gene set (H) and anti-PD-1 unresponsive gene set (I) in human sTim-3 protein-treated CTLs versus Mock CTLs. FDR, false discovery rate; NES, normalized enrichment score. (J) Representative immunoblots of sTim-3 protein-treated OT-Ⅰ CTLs or Mock OT-Ⅰ CTLs after stimulation with anti-CD3/CD28 antibodies for indicated time points. (K) Flow cytometry analysis of phospho-LCK, phospho-Zap-70 (Syk), and phospo-PLCγ-1 in mouse HCC tumor-infiltrating CD8⁺ T cells. All results are representative of at least three independent experiments. Data are presented as mean ± SEM and were analyzed by the unpaired Student’s t test (A, B, D, E, F, G, and K) and two-way ANOVA (C). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001; ns, no significance.

Figure 6

sTim-3 augments T cell dysfunction and anti-PD-1 resistance via CEACAM-1 (A–C) RV-shCEACAM-1 was used to knock down CEACAM-1 in OT-I CTLs cells. (A) shCEACAM-1 OT-Ⅰ CTLs were pretreated with sTim-3 protein (5 μg/mL), followed by stimulation with anti-CD3/CD28 antibodies for 6 h. IFN-γ and TNF-α production was assessed using flow cytometry (n = 3). (B) Exhausted shCEACAM-1 OT-Ⅰ CTLs were induced in the presence or absence of sTim-3. Percentages of PD-1⁺Tim-3⁺TOX⁺TCF1⁻ CD8⁺ T_EX cells were assessed using flow cytometry (n = 5). (C) shCEACAM-1 OT-Ⅰ CTLs were pretreated with IgG or anti-PD-1 antibodies (5 μg/mL) in the presence of sTim-3 protein (5 μg/mL) or not, followed by stimulation with anti-CD3/CD28 antibodies for 6 h. IFN-γ and TNF-α production was assessed using flow cytometry (n = 3). (D–F) RV-sgCEACAM-1 was used to knock out CEACAM-1 in OT-Ⅰ-Cas9^tdTomato CTLs cells. (D) IFN-γ and TNF-α production by control or sTim-3-treated sgCEACAM-1 OT-Ⅰ CTLs were assessed using flow cytometry (n = 3). (E) Exhausted sgCEACAM-1 OT-Ⅰ CTLs were induced in the presence or absence of sTim-3. Percentages of PD-1⁺Tim-3⁺TCF1⁻ CD8⁺ T_EX cells were assessed using flow cytometry (n = 5). (F) sgCEACAM-1 OT-Ⅰ CTLs were pretreated with IgG or anti-PD-1 antibodies (5 μg/mL) in the presence of sTim-3 protein (5 μg/mL) or not, followed by stimulation with anti-CD3/CD28 antibodies. IFN-γ and TNF-α production was assessed using flow cytometry (n = 3). (G and H) Jurkat cells (G) or CEACAM-1-overexpressing Jurkat cells (Jurkat-CEACAM-1) (H) were pretreated with sTim-3 protein (5 μg/mL) and then stimulated with anti-CD3/CD28 antibodies for 12 h. IL-2 expression at mRNA and protein level was assessed using qPCR and flow cytometry (n = 3). (I and J) Representative immunoblots of control (Mock) and sTim-3-treated Jurkat cells (I) or Jurkat-CEACAM-1 cells (J), which were stimulated with anti-CD3/CD28 antibodies for indicated time points. (K and L) PD-1-overexpressing Jurkat or Jurkat-CEACAM-1 cells were pretreated with anti-PD-1 antibody for 1 h and then stimulated with plate-coated anti-CD3/CD28 antibodies and PD-L1 protein for 6 h. RT-qPCR detected the IL-2 mRNA level of Jurkat-PD-1 cells (K) and Jurkat-CEACAM-1-PD-1 cells (L). All results are representative of at least three independent experiments. Data are presented as mean ± SEM and were analyzed by the unpaired Student’s t test (A, B, C, D, E, F, G, H, K, and L). ∗p < 0.05 and ∗∗p < 0.01; ns, no significance.

Figure 7

Blocking sTim-3 or ADAM10 suppresses tumor progression and reverses anti-PD-1 resistance (A and B) Tim-3 knockout mice were hydrodynamically injected with mTim-3/AKT/c-Myc plasmids to construct the sTim-3-overexpressing HCC model and then received anti-Tim-3 antibody (α-Tim-3) at the indicated times. (A) Schematic diagram of the sTim-3-overexpressing HCC in Tim-3 KO mice treated with anti-Tim-3 antibody. (B) Survival curve of Tim-3 KO mice bearing sTim-3-overexpressing HCC treated with IgG or anti-Tim-3 antibody (n = 7). (C–E) Tim-3 humanized mice were hydrodynamically injected with AKT/c-Myc plasmids to construct an HCC model, and mice were then treated with intraperitoneal DMSO or ADAM10 inhibitor GI254023X (GI, 20 mg/kg/d) at the indicated times. (C) Schematic diagram of the experiment. (D) Tumor growth was assessed using bioluminescence imaging (n = 7 per group), and (E) survival curves were monitored (DMSO: n = 7; GI: n = 8). (F and G) Human TILs from HCC patients were isolated and pretreated with ADAM10 inhibitor GI (5 μM) or DMSO together with anti-PD-1 antibodies (5 μg/mL) or IgG (5 μg/mL) for 3 days and then stimulated with anti-human CD3/CD28 beads for 24 h (n = 4). (F) Schematic diagram of the experiment. (G) The secretion of IFN-γ in the supernatants was detected using ELISA. All results are representative of at least two independent experiments. Data are presented as mean ± SEM and were analyzed by the unpaired Student’s t test (D and G) and log rank test (B and E). ∗p < 0.05 and ∗∗p < 0.01; ns, no significance.