Inosine shapes PD-1 blockade responses and synergizes with dual PD-1/CTLA-4 immunotherapy to enhance antitumor immunity

Abstract

Inosine, a bacterial metabolite and agonist of the adenosine A2A receptor, modulates antitumor immunity. However, its precise effects on immune checkpoint inhibitors remain unclear. This study aimed to evaluate the impact of inosine on the efficacy of anti-programmed cell death protein 1 (PD-1) therapy and explore strategies to counteract any potential inhibitory effects. In in vitro co-culture systems, inosine selectively suppressed cancer cell growth without impairing T-cell viability. In a murine subcutaneous tumor model, inosine treatment reduced tumor growth and was associated with elevated interferon-gamma levels in the tumor microenvironment, along with increased infiltration by tumor-infiltrating lymphocytes and enhanced splenic CD4⁺ and CD8⁺ T-cell frequencies. However, the combination of inosine with anti-PD-1 therapy attenuated the antitumor effect and increased cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) expression in splenic T cells compared to levels after anti-PD-1 monotherapy. To overcome this inhibitory effect, we tested whether adding an anti-CTLA-4 antibody could restore antitumor immunity. Notably, the combination of inosine with both anti-PD-1 and anti-CTLA-4 antibodies significantly enhanced antitumor efficacy. These findings suggest that inosine may synergize with dual ICI therapy and represent a promising adjunct to improve immunotherapeutic outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-025-04111-2.

Introduction

Immune checkpoint inhibitors (ICIs) attack cancer cells through the host immune system. In certain patients, ICIs remain effective even after treatment discontinuation. However, the efficacy of ICIs is currently approximately 10–20%, and enhancing their efficacy is a challenge [1]. The efficacy of ICIs depends on the tumor microenvironment (TME). Therefore, targeting the TME is necessary to enhance the efficacy of ICIs [2, 3].

Gut microbiota significantly influences host immune function and can regulate the TME [4]. The dynamic interaction between the gut microbiota and TME facilitates the activation of immune cells and creates a favorable environment for them to attack cancer cells, which is crucial for ICI efficacy [5]. The efficacy of ICI depends on the host gut bacteria [6–10], which modulates host immune functions through its metabolites. Intestinal bacteria degrade undigested products to produce metabolites, such as short-chain fatty acids, which influence the host immune function [11–13].

Inosine, a nucleoside produced by intestinal bacteria such as Bifidobacterium, is involved in host immune function and enhances ICI therapy [14–16]. It exerts antitumor effects by modulating the immune responses in the TME [14]. However, the immune-mediated antitumor effects of inosine and the efficiency of ICI treatment combined with inosine remain unclear.

In this study, the antitumor effects of inosine were assessed using both in vitro and in vivo models. Our results demonstrated the potential of a novel inosine combination therapy to maximize ICI efficacy.

Materials and methods

Cell culture

The human oral cancer cell line T3M-1 Clone2 (RRID:CVCL_8762) was provided by the RIKEN BioResource Research Center through the National BioResource Project of MEXT, Japan. The mouse colon cancer cell line CT-26 (RRID:CVCL_7254) was purchased from the American Tissue Culture Collection (Manassas, VA, USA). CT-26 and T3M-1 Clone2 were cultured in RPMI-1640 medium (Fujifilm, Tokyo, Japan) and Ham’s F-10 nutrient mix (Thermo Fisher Scientific, Waltham, MA, USA), respectively, supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific) and 1 × penicillin/streptomycin (Thermo Fisher Scientific).

T-cell activation and proliferation

Peripheral blood mononuclear cells (PBMCs) were isolated from the peripheral blood of healthy donors using Ficoll-Paque PLUS (Cytiva, Uppsala, Sweden) density gradient centrifugation following the manufacturer’s protocol. PBMCs were stimulated using plate-bound anti-CD3 (2 µg/mL, clone OKT3) and anti-CD28 (2 µg/mL, clone CD28.2) antibodies (BioLegend, San Diego, CA, USA) in the presence of 100 U/mL IL-2 (ProSpec, East Brunswick, NJ, USA). Cells were cultured in iMediam for T (GC Lymphotec, Tokyo, Japan) and diluted to 5 × 10⁵ cells/mL by adding fresh culture medium supplemented with 25 U/mL IL-2 on days 4, 7, and 10. On day 14, T cells were used for co-culture experiments. All protocols and experiments involving primary PBMCs were approved by the Institutional Review Board of Showa University. Informed consent was obtained from all the volunteers.

Mouse T cells were isolated from the splenocytes of female BALB/c mice (6–8 weeks old) through positive selection using CD90.2 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany), following the manufacturer’s protocol. Isolated cells were stimulated with plate-bound anti-CD3 ε (2 µg/mL, clone 145-2C11) and anti-CD28 (1 µg/mL, clone 37.51) antibodies (BioLegend) in the presence of 100 U/ml IL-2 (ProSpec). For mouse T-cell activation, cells were cultured in RPMI-1640 medium supplemented with 10% FBS and 1 × penicillin/streptomycin solution. The cells were diluted to 1 × 10⁶ cells/mL by adding fresh culture medium supplemented with 20 U/mL IL-2 on day 3. On day 4, T cells were used for the co-culture experiments.

Co-culture experiments

Cancer cells (T3M-1, 5 × 10⁴ cells and CT-26, 2.5 × 10⁴ cells) were seeded at least 2 h before T-cell addition, and activated T cells (human T, 5 × 10⁵ cells and mouse T, 2.5 × 10⁵ cells) were co-cultured in 6-well plates with the specified T-cell medium in a total volume of 2 mL. Inosine (Fujifilm, Osaka, Japan) was prepared in PBS at a 100-fold concentration, and 1/100th volume was added to the culture medium. The adenosine A2A receptor antagonist SCH58261 (TCI Chemicals, Tokyo, Japan) was added to the culture medium 30 min prior to the addition of T cells. The co-culture was maintained at 37 °C in 5% CO₂ for 72 h following the addition of inosine. The absolute numbers of T and cancer cells were determined through flow cytometry (FCM) using fluorescent counting beads (Flow-Count; Beckman Coulter, Brea, CA, USA). After collecting the floating cells, adherent cells were trypsinized and collected in the same tube containing a defined number of counting beads. Part of this suspension was stained with a fluorophore-conjugated anti-mouse CD3 antibody to determine the cancer-to-T-cell ratio. CD3 + cells were defined as T cells and all others as cancer cells. Additionally, co-cultured mouse T cells were assessed for surface activation markers using FCM.

Mouse tumor model

BALB/c mice were purchased from CLEA Japan (Tokyo, Japan) and maintained at a constant temperature (23 ± 1 °C) with a 12 h light/dark cycle under specific pathogen-free conditions. The experimental protocols were approved by the Institutional Animal Care and Use Committee of Showa University, and all experiments were conducted according to the relevant guidelines and regulations. Female mice (8–12 weeks old) were subcutaneously injected with CT-26 (1 × 10⁵ cells in 50 μL Hanks’ Balanced Salt Solution) in the right flank. Mice were intraperitoneally administered 1 g/kg inosine or an equivalent volume of saline solution every alternate day, starting from day 5 following tumor inoculation until the end of the experiment. Additionally, they received intraperitoneal injection of anti-PD-1 (clone RMP1-14, BioXcell, Lebanon, NH), anti-CTLA-4 (clone 9D9, BioXcell, Lebanon, NH), or control IgG (rat IgG for anti-PD-1 and mouse IgG for anti-CTLA-4, both from FUJIFILM). Unless otherwise stated, the dose of these antibodies was 150 µg per mouse, administered once a week from day 4. In the experiment investigating the effect of inosine on dual ICI, the dose of ICI was 100 µg per mouse and treatment started on day 11. For depletion of CD8 + T cells, 100 µg anti-CD8 (clone 2.43, BioXcell) antibody was administered on days 4, 11, and 18. Tumor length and width were measured twice a week using calipers, and tumor volumes were calculated as a length × width² × 0.5. On day 28 following tumor inoculation, the mice were euthanized for sample collection. Mice were categorized into responder and non-responder groups based on the final tumor volume. Mice with tumor volumes < 500 mm³ on the final day were classified as responders, while those with ≥ 500 mm³ were classified as non-responders.

Immunohistochemistry (IHC)

Tumor-infiltrating lymphocytes (TILs) were assessed using immunostaining with an anti-CD3 [13]. Sections of formalin-fixed, paraffin-embedded tumor tissues were mounted on silane-coated glass slides. Following deparaffinization, the antigens were retrieved by autoclaving in citrate buffer (pH 6; AS ONE, Osaka, Japan). Endogenous peroxidase was inhibited using 0.3% hydrogen peroxide, and non-specific staining was prevented using a non-specific blocking reagent (DAKO, Tokyo, Japan). Sections were incubated with a primary antibody (anti-mouse CD3 ε chain rabbit monoclonal antibody, clone SP7, Abcam, Cambridge, UK), followed by a peroxidase-linked secondary antibody (EnVision, DAKO). Color was developed using the DAB Liquid System (DAKO). TILs were assessed in a blinded manner by counting CD3 + cells in six high-power fields using 4–5 mice per group. Data were expressed as the number of CD3 + cells/mm².

Quantitative reverse transcription (RT)-PCR

Total RNA from tumor tissues was isolated using an RNAeasy mini kit with a QIAshredder (Qiagen, Hilden, Germany), followed by reverse transcription using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific), following the manufacturer’s protocol. Multiplex qPCR assays were performed on QuantStudio3 using TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific). These assays used a FAM-labeled probe for the target gene and a VIC-labeled mouse actin beta (Actb) endogenous control (Thermo Fisher Scientific). Predesigned PrimeTime qPCR assays for Foxp3 and Infg (assay ID Mm.PT.58.16779326 and Mm.PT.58.41769240, respectively) were purchased from Integral DNA Technologies (Coralville, IA, USA). Relative gene expression levels were calculated using the comparative Ct method and expressed as fold-change relative to the reference sample.

Flow cytometry (FCM)

Splenocytes and lymph node cells were isolated by grinding the spleen and draining lymph nodes (axillary and inguinal) in RPMI-1640 medium, respectively, followed by red blood cell lysis. TILs were isolated by cutting tumor tissues into < 1 mm³ pieces and digesting in 4 mL RPMI-1640 medium (without FBS) containing 0.02 mg/mL DNase I and 0.1 mg/mL Liberase (Roche, Mannheim, Germany) at 37 °C for 1 h with shaking. The reaction was neutralized by adding 5 mL of RPMI-1640 medium supplemented with 10% FBS and 2 mM ethylenediaminetetraacetic acid. The cell suspension was passed through a 100 μm cell strainer and resuspended in 2 mL of RPMI-1640 medium supplemented with 10% FBS. The tumor-cell suspension was layered on top of the 2 mL Ficoll-Paque PLUS and centrifuged at 400 × g for 20 min at 20 °C. The leukocyte layer was carefully transferred and resuspended in PBS with 2% FBS. Fc-mediated antibody binding was blocked using anti-CD16/32 (Tonbo Biosciences, San Diego, CA, USA) before reacting with antibodies against the target molecules. Antibodies used in this study for the analysis of mouse splenic T cells are listed in Tables S1–S3. For live cell gating, BD Horizon Fixable Viable Stain 780 (FVS780; BD Biosciences, Franklin Lakes, NJ, USA) was used. A transcription factor buffer set (BD Biosciences) was used for intracellular staining. For IFNγ staining, lymphocytes from tumor tissues were stimulated using 1 µg/mL ionomycin and 50 ng/mL phorbol 12-myristate 13-acetate with GolgiStop (BD Life Sciences) for 4 h at 37 °C. Cells were harvested, stained for cell surface antigen, and then fixed and permeabilized using a BD Cytofix/Cytoperm Plus Fixation/Permeabilization Kit with BD GolgiStop (BD Life Sciences) according to the manufacturer’s instructions. Cells were then stained for intracellular IFNγ. FCM was performed on a BD LSRFortessa or BD FACSLyric, and the data were analyzed using FlowJo v10.10.0 software (BD Life Sciences).

Statistical analysis

Data from three or more groups were analyzed using analysis of variance (ANOVA) with Dunnett’s multiple comparison test. Tumor growth curves from the mouse subcutaneous tumor model were compared using two-way ANOVA with Holm–Sidak’s multiple comparison test. Data from two groups were analyzed using an unpaired two-tailed Student’s t-test. For the analysis of mouse tumor models in which tumor sizes vary greatly within a group, areas under tumor growth curve were compared using Kruskal–Wallis test with Dunn’s multiple comparison test. Statistical analyses were performed using GraphPad Prism version 10 for MacOS (GraphPad Software, Boston, MA, USA; www.graphpad.com). Statistical significance was set at P < 0.05. To examine the additive effect of inosine on dual immune checkpoint blockade, tumor growth data were analyzed using a linear mixed-effects model (LMM) in R (Version 1.4.1106) with the lme4, lmerTest, and tidyverse packages. The model included fixed effects for time (Days), treatment group (inosine vs. control), and their interaction, with mouse identity specified as a random intercept: TumorVolume ~ Days * Group + (1 | Mouse). The significance of fixed effects was assessed using Satterthwaite’s approximation for degrees of freedom.

Study approval

All protocols and experiments involving primary PBMCs were approved by the Institutional Review Board of Showa University. Informed consent was obtained from all the volunteers. For animal studies, the experimental protocols were approved by the Institutional Animal Care and Use Committee of Showa University, and all experiments were conducted according to the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines.

Results

Initially, we assessed the effects of inosine, a gut bacterial metabolite, on the survival of cancer and T cells in a co-culture setting (Fig. 1a). Under co-culture conditions, inosine did not inhibit human T-cell survival. In contrast, the proliferation of the human squamous cell carcinoma cell line T3M-1 was inhibited in a concentration-dependent manner (≥ 1 mM). Additionally, the effects of inosine were assessed in a cell monoculture system. Inosine inhibited the proliferation of T3M-1 cells only at a concentration of 10 mM (Fig. S1a). Similar to that in co-culture conditions, inosine exhibited no inhibitory effect on the proliferation of activated T cells, even at 10 mM (Fig. S1b). These results indicated that inosine selectively inhibited the proliferation of cancer cells over T cells. Moreover, we assessed the effects of inosine on mouse T cells and CT-26 colon cancer cells under co-culture conditions (Fig. 1b). Inosine selectively suppressed the growth of CT-26 cells at a concentration of 10 mM, reproducing the results obtained in human cells. Furthermore, we assessed whether inosine exerted any modulatory effects on T cells under in vitro co-culture conditions (Fig. 1c). The expression levels of T-cell activation markers, such as inducible T-cell costimulator (ICOS) and cluster of differentiation 95 (CD95), in CD8 + T cells and CD95 in CD4 + T cells increased in an inosine concentration-dependent manner (Fig. 1c, S2). These results indicate that inosine exhibited stimulatory effects. The inosine-induced changes observed in the murine co-culture system were not reversed in the presence of the adenosine A2A receptor antagonist SCH58261 at 1 μM (Fig. S3). Assuming an EC₅₀ of 0.3 mM for inosine at the A2A receptor [17] and a Kd of 1 nM for SCH58261 [21], the predicted receptor activation at 10 mM inosine in the presence of 1 µM SCH58261 is approximately 3% based on a competitive binding model. Therefore, the inosine-induced effects observed in our system are unlikely to be mediated through A2A receptor activation.

Fig. 1.

Selective inhibition of cancer cells by inosine. a Effect inosine on the survival of T3M-1 Clone2 oral cancer cells (open circles) and human T cells (black circles) in co-culture. Data are represented as fold-changes relative to the control (treated with phosphate-buffered saline, PBS). b Effect of inosine on the survival of CT-26 colon cancer cells (open circles) and mouse T cells (black circles) in co-culture. Data are represented as fold-changes relative to the PBS control. c Expression of surface activation markers in CD8 + (upper panels) and CD4 + (lower panels) T cells in co-culture with CT-26 cells in various inosine concentrations. Data are presented as means with SEM from three independent experiments. *P < 0.05, **P < 0.01, and ****P < 0.0001 vs. PBS control

Subsequently, to assess whether the results obtained in vitro can be reproduced in a mouse tumor model, mouse colon cancer cells (CT-26) were subcutaneously implanted in the flanks of BALB/c mice and treated with inosine. Compared with that of the control group, the inosine-treated group demonstrated reduced growth of subcutaneous tumor tissue over time (Fig. 2a), and the tumor volume on day 28 was significantly smaller in the inosine-treated group than that in the control group (Fig. 2b). All mice in the control group exhibited tumor progression. Tumor growth was also observed in all mice in the inosine group, but of a lesser degree than that in the control group (Fig. 2c). None of the individuals exhibited tumor regression after inosine treatment (Fig. 2d). To determine whether the antitumor effect of inosine is mediated by the immune system, we depleted CD8 + T cells and assessed tumor growth. Combined treatment with anti-CD8 and inosine exacerbated the tumor growth compared to that of mice treated only with inosine (Fig. S4).

Fig. 2.

Antitumor effects of inosine in a mouse CT-26 subcutaneous model. a Tumor growth curve of mice treated with PBS + IgG (black) or inosine + IgG (orange). Data are expressed as means with SEM from 9 to 11 mice per group. b Tumor volume at the end of the experiment. Data are expressed as means with SEM. c Tumor growth curves of individual mice. d Pictures of resected tumors at the end of the experiment. *P < 0.05 by Student’s t-test

Subsequently, we aimed to elucidate the mechanism by which inosine inhibits tumor growth in vivo. TILs were quantified by immunostaining tumor tissues with anti-CD3. Compared with that of the control group, more T cells infiltrated the tumors in the inosine-treated group (Fig. 3a). Additionally, Ifng gene expression in tumor tissues was increased in the inosine-treated group compared to that in the control group (Fig. 3b). Moreover, the percentages of CD4 + and CD8 + T cells in the spleen increased in the inosine-treated group (Fig. 3c). However, the proportions of CD4 + and CD8 + T cells in tumor tissues were not altered by inosine treatment (Fig. S5a). Furthermore, flow cytometric analysis showed no change in IFNγ expression levels in tumor tissues (Fig. S5b). These results indicate that the antitumor effects of inosine are mediated through antitumor immunity.

Fig. 3.

Immunostimulatory effects of inosine. a Immunostaining of resected tumor tissue with anti-CD3. Left: representative micrographs. Scale bar = 200 µm. Right: number of CD3 + cells/mm². Data are expressed as means with SEM from four mice per group. b Relative gene expression of IFNG in tumor tissues from control and inosine-treated mice quantified through RT-qPCR analysis. c Percentage of CD4 + (left) and CD8 + (right) T cells in total splenocytes from tumor model mice at the end of the experiment. Data are expressed as means with SEM from 9 to 11 mice. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student’s t-test

TILs are predictive biomarkers for the efficacy of ICIs. Therefore, we hypothesized that inosine may potentiate the effects of ICIs by increasing the number of TILs. To test this, we treated a mouse subcutaneous tumor model with an anti-PD-1 alone or in combination with inosine, and compared the alterations in tumor growth over time (Fig. 4a, b). In the group treated with the anti-PD-1 alone, 9 out of 10 mice responded to the treatment (Fig. 4d, e). In contrast, only 7 out of 11 animals responded to the combined treatment (Fig. 4d, e). Notably, the tumor size of non-responsive individuals was larger in the inosine combination group than that in the anti-PD-1 monotherapy group (Fig. 4b). Furthermore, comparison of the area under the tumor growth curve showed that tumor burden tended to increase with inosine addition, although the difference was not statistically significant (Fig. 4c). These results indicate that contrary to expectations, inosine suppressed the antitumor effects of anti-PD-1.

Fig. 4.

Inosine suppresses the antitumor effects of anti-programmed cell death protein 1 (anti-PD-1) antibody in a mouse CT-26 subcutaneous model. a Tumor growth curve of mice treated with PBS + IgG (black), PBS + anti-PD-1 (red) and inosine + anti-PD-1 (blue). Data are expressed as means with SEM from 9 to 11 mice per group. b Tumor growth curves of individual mice. c Area under the tumor growth curve. Each dot represents an individual value. d Pictures of resected tumors at the end of the experiment. Blanks indicate the disappearance of the tumor. e Number of responders and non-responders

Subsequently, we assessed why inosine suppressed the antitumor effects of anti-PD-1. Anti-CD3 immunostaining of tumor tissues exhibited no significant reduction in TILs in the inosine combination group compared to that in the anti-PD-1 monotherapy group (Fig. 5a). Quantitative PCR of tumor tissues demonstrated that the expression of Foxp3, a master transcription factor for Tregs, was higher in the inosine combination group than that in the anti-PD-1 monotherapy group (Fig. 5b). FCM analysis of splenocytes demonstrated that the percentage of Tregs was higher in the inosine combination group than that in the anti-PD-1 monotherapy group (Fig. 5c, S7). These results indicate that the combination of an anti-PD-1 and inosine increased the number of Tregs. Additionally, we evaluated CTLA-4 expression in tumor tissues, draining lymph nodes, and spleen. CTLA-4 expression in splenic CD8 + T cells increased in the inosine combination group compared to that in the anti-PD-1 monotherapy group (Fig. 5d, S8). In tumor-draining lymph nodes and tumor tissues, CTLA-4 expression was unaffected by the addition of inosine (Fig. S8a, b). These findings imply that the addition of inosine to anti-PD-1 treatment may contribute to attenuated antitumor effects, potentially through increased Tregs and elevated CTLA-4 expression in splenic T cells. Furthermore, inosine stimulation increased CTLA-4 expression in in vitro cultured CD8⁺ T cells, mirroring the results observed in splenic T cells in vivo (Fig. S7).

Fig. 5.

Upregulation of cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) expression by the combination of inosine and anti-programmed cell death protein 1 (anti-PD-1) antibody. a Immunostaining of resected tumor tissue with anti-CD3. Left: representative micrographs. Scale bar = 200 µm. Right: number of CD3 + cells/mm². Data are expressed as means with SEM from four mice per group. ns, not significant. b Relative gene expression of FoxP3 in tumor tissues from mice treated with PBS + anti-PD-1 and inosine + anti-PD-1 quantified through RT-qPCR analysis. Data are expressed as means with SEM from 4 to 5 mice. Numbers indicate p values. c Percentage of CD4 + FoxP3 + T cells in spleen and CD3 + cells from tumor model mice at the end of the experiment. Data are expressed as means with SEM from 9 to 11 mice. Numbers indicate p values. d Expression of CTLA-4 in CD4 + (left) and CD8 + (right) splenic T cells. Data are expressed as means with SEM from 9 to 11 mice. MFI, mean fluorescence intensity. e Expression of CTLA-4 in CD4 + T (left), CD8 + T (middle), and Treg (right) cells in lymph nodes. Numbers indicate p values. *P < 0.05 by Student’s t-test

As the spleen is a major site of peripheral T-cell activation and regulation, we hypothesized that increased CTLA-4 expression in this compartment might contribute to systemic immune suppression and interfere with anti-PD-1 efficacy. Therefore, we tested whether CTLA-4 blockade could overcome this suppressive effect and restore antitumor immunity. Tumor growth was suppressed in the inosine and anti-PD-1 combination group compared to that in the untreated group (Fig. 6a). Moreover, 5 out of 10 individuals responded to the treatment (Fig. 6b, c). In contrast, in the group receiving anti-CTLA-4 in addition to the combination of inosine and anti-PD-1, tumor growth was further suppressed compared to group receiving combined inosine and anti-PD-1 (Fig. 6a). Tumor regression was observed in all individuals (Fig. 6b, c). These results indicate that the anti-CTLA-4 ameliorated the reduced antitumor effects caused by the combination of inosine and anti-PD-1. Finally, we investigated whether inosine enhances the effect of dual immune checkpoint blockade. To appropriately evaluate the potential additive effect of inosine, we reduced the dose of anti-PD-1 and anti-CTLA-4 antibodies from the standard 150 µg to 100 µg per mouse, since the full dose elicited strong antitumor effects on its own. Compared to those in the reduced-dose dual ICI group, mice receiving additional inosine tended to show less tumor growth (Fig. 6f, g). LMM revealed a significant interaction between treatment group and time (Days × Group: coefficient =  − 22.49, p = 0.012), indicating that inosine significantly attenuated tumor growth rate over time. These results suggest that inosine may enhance the antitumor effect of dual ICI therapy under the conditions of partial immune activation.

Fig. 6.

Anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) antibody overcomes the inhibitory effects of inosine in combination with anti-programmed cell death protein 1 (PD-1) antibody in a mouse CT-26 subcutaneous tumor model. a Tumor growth curve of mice treated with PBS (black), inosine + anti-PD-1 + IgG (blue), or inosine + anti-PD-1 + anti-CTLA-4 (magenta). The inosine-treated groups received intraperitoneal injections (150 µg per mouse) of anti-PD-1 and mouse IgG or anti-CTLA-4 on days 4, 11, 18, and 25. Data are expressed as means with SEM from 10 mice per group. b Area under the tumor growth curve. Each dot represents an individual value. Numbers indicate p values from Dunn’s multiple comparison test. c Tumor growth curves of individual mice. d Number of responders and non-responders. e Pictures of resected tumors at the end of the experiment. Blanks indicate the disappearance of the tumor. f Tumor growth curve of mice treated with anti-PD-1 + anti-CTLA-4 (green) or inosine + anti-PD-1 + anti-CTLA-4 (magenta). Data points represent mean tumor volumes ± SEM at each time point for each treatment group (n = 8–9). Solid lines indicate group-wise means of observed values, and dashed lines represent predicted tumor growth curves estimated by a linear mixed-effects model (LMM), incorporating fixed effects of treatment, time (Days), and their interaction. The LMM revealed a significant interaction between treatment and time (Days × Group: coefficient =  − 22.49, p = 0.012), indicating that inosine treatment significantly attenuated tumor growth over time. g Tumor growth curves of individual mice

Discussion

This study demonstrated the antitumor effects of inosine in both in vitro co-culture systems and in vivo subcutaneous tumor models. In monoculture systems, inosine directly inhibited tumor-cell proliferation, indicating a cytostatic or cytotoxic effect on cancer cells. Interestingly, a more pronounced inhibitory effect was observed in T-cell–cancer cell co-cultures, suggesting that inosine also enhances T-cell-mediated antitumor responses. In support of this, inosine increased the expression of multiple T-cell activation markers, such as ICOS and CD95, particularly in CD8 + T cells. Although inosine acts as an adenosine A2A receptor (A2AR) agonist, the T-cell-stimulatory effects observed in vitro were not reversed by co-treatment with the A2AR antagonist SCH58261, even at a concentration sufficient to block receptor signaling. This suggests that the immunostimulatory effects of inosine under these conditions are likely A2AR-independent. Instead, recent studies have proposed that inosine may influence T-cell function through receptor-independent mechanisms such as metabolic reprogramming [15, 16, 22]. Inosine serves as an alternative carbon source that fuels oxidative phosphorylation and supports effector T-cell survival and function under nutrient-limited or hypoxic conditions [15]. Thus, the enhanced T-cell activation observed in our co-culture model may, at least in part, result from inosine-mediated metabolic support rather than canonical A2AR signaling. Importantly, the 10 mM concentration of inosine used in our in vitro experiments is physiologically relevant, as Mager et al. reported similar circulating levels in mice [14]. Moreover, inosine affects T-cell metabolism at millimolar concentrations [15], supporting a receptor-independent metabolic mechanism in our system.

The suppression of tumor growth in the mouse subcutaneous tumor model appears to be immune-mediated, as it was accompanied by increased Ifng expression in the TME, enhanced infiltration by TILs, and elevated proportions of splenic CD4 + and CD8 + T cells. The concentration-dependent increase in T-cell activation markers observed in vitro further supports this interpretation. Moreover, depletion of CD8 + T cells abrogated the antitumor effect of inosine, demonstrating that these cells play a critical role in mediating its efficacy. The antitumor effect of inosine has been reported by Zhang et al., who observed tumor suppression in B16-F0 and B16-GM-CSF models, but not in the 4T1 breast cancer model [16]. Similarly, another study using B16-F10 cells showed no in vivo efficacy [15], suggesting that inosine’s effect depends on tumor type. Our study extends these findings by demonstrating that inosine alone suppresses tumor growth in a CT26 colon cancer model through a CD8 + T-cell-dependent mechanism. This highlights the broader potential of inosine as an immune-modulating agent and emphasizes the importance of tumor context in determining therapeutic response.

Another major finding of this study was the inhibitory effect of inosine on the antitumor effect of anti-PD-1 therapy. Because the antitumor effects of inosine were accompanied by T-cell activation and increased TILs, a synergistic effect of inosine and ICI was anticipated. Contrary to our expectations, the tumors were larger when inosine was combined with anti-PD-1 therapy than when anti-PD-1 was used alone. Several studies have indicated the synergistic effects of a combination of inosine and ICIs; however, these studies used various ICIs. No previous study has assessed the combined effect of anti-PD-1 therapy and inosine in a mouse model; the mechanistically closest setting was reported by Wang et al., who combined anti-PD-L1 treatment and inosine [15]. In their study, using a B16-F10 subcutaneous tumor model, inosine combined with anti-PD-L1 therapy suppressed tumor growth and prolonged individual survival, in contrast to our results. These discrepancies may stem from the use of different tumor cells and distinct mechanisms of PD-1 and PD-L1 inhibition. Additionally, a recent phase 2 clinical study by Zhao et al. reported that inosine enhanced the efficacy of PD-1/PD-L1 inhibitors in patients with advanced solid tumors [18]. The study demonstrated that the addition of inosine prolonged progression-free survival and improved objective response rates. However, the results varied across different tumor types and treatment regimens. These findings suggest that the effects of inosine in combination with ICIs may be context-dependent and influenced by the tumor type, immune microenvironment, and specific checkpoint inhibitors used.

Based on the increased splenic Tregs and FoxP3 expression in tumor tissue, as well as the elevated CTLA-4 expression in splenic CD8 + T cells, we hypothesized that CTLA-4-mediated immune regulation might contribute to the reduced efficacy of anti-PD-1 when combined with inosine. Indeed, the addition of anti-CTLA-4 reversed this attenuation and led to tumor regression in all treated animals, consistent with findings reported by Zhang et al. [16], although different tumor models were used. While these results suggest a role for CTLA-4-associated suppression, the underlying mechanism remains unclear. Notably, the increase in CTLA-4 expression was observed only in splenic T cells, with no changes in tumor or draining lymph node compartments, making it difficult to determine whether CTLA-4 directly mediates the suppressive effect of inosine. Previous studies have shown that CTLA-4 expression can be modulated by purinergic signaling, but whether this applies to inosine in our model is unclear. Together, our findings imply that inosine may activate compensatory regulatory pathways that limit PD-1-targeted therapy, but further investigation is required to clarify the molecular and cellular basis of this effect.

Our results suggest that inosine may enhance the efficacy of dual immune checkpoint blockade with anti-PD-1 and anti-CTLA-4 antibodies. Given that Bifidobacterium pseudolongum, a known inosine-producing species, is associated with improved responses to ICIs [14], both microbial composition and fecal inosine levels may serve as useful biomarkers to guide treatment strategies. Notably, microbial metabolites exert diverse effects on the immune system—short-chain fatty acids, for example, potentiate the efficacy of anti-PD-1 therapy while limiting that of anti-CTLA-4 [19, 20], in contrast to the actions of inosine observed in our study. These findings underscore that bacterial metabolites act through distinct and sometimes opposing mechanisms, and that their compatibility with specific ICIs may vary. Further studies are needed to elucidate the precise immunological roles of these metabolites and to determine how they might be leveraged—either alone or in combination with ICIs—to optimize cancer immunotherapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADORA2A

Adenosine 2A receptor

FBS

Fetal bovine serum

FCM

Flow cytometry

ICI

Immune checkpoint inhibitor

IFN-γ

Interferon gamma

PBMC

Peripheral blood mononuclear cell

RT-PCR

Reverse transcription polymerase chain reaction

TCGA

The cancer genome atlas

TME

Tumor microenvironment

Author contributions

AK, MH, and KY conceived and designed the study. YN, MM, EF, AS (Akiko Sasaki), YB, HT, JI, KT, RN, AS (Aya Sasaki), YM, YY, MS, and TT (Toshiaki Tsurui) performed the experiments. YN, AK, MH, YH, HA, TI, RS, RO, YK (Yutaro Kubota), and AH acquired and analyzed data. YN, AK, and KY wrote the manuscript, which was edited by TS, MT, SW, SK (Shinichi Kobayashi), TT (Takuya Tsunoda), SK (Sei Kobayashi), HK, TS, TO (Tatsunori Oguchi) and YK (Yujij Kiuchi). All authors read and approved the final manuscript.

Funding

This work was supported by Grant-in-Aid for Scientific Research (C) 22K09675.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interests

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

All protocols and experiments involving primary peripheral blood mononuclear cells were approved by the Institutional Review Board of Showa University. Informed consent was obtained from all blood donors. The animal experimental protocols were approved by the Institutional Animal Care and Use Committee of Showa University, and all experiments were conducted according to the relevant ARRIVE guidelines.