A selective IDO1 inhibitor, KHK2455, improves efficacy of PD-L1 blockade by modulating both innate and adaptive immunity in a mouse melanoma model

Abstract

Indoleamine 2,3-dioxygenase 1 (IDO1) facilitates tumor progression by catabolizing tryptophan into kynurenine (Kyn). While KHK2455, a selective IDO1 inhibitor, reduced Kyn levels in mouse tumors and plasma, it did not exert the expected antitumor activity in a mouse melanoma model. However, when combined with programmed death-ligand 1 (PD-L1) blockade, KHK2455 demonstrated enhanced antitumor effects compared with PD-L1 blockade alone. This study investigated the effects of IDO1 inhibition on the tumor microenvironment and mechanisms underlying the enhanced antitumor effects of combining IDO1 inhibition with PD-L1 blockade. PD-L1 blockade upregulated the pathways related to adaptive immunity including T-helper cells type 1 and 2 (Th1 and Th2) rather than innate immunity. On the other hand, IDO1 inhibition upregulated genes and pathways associated with innate immunity, such as natural killer cells, neutrophils, and macrophages. Furthermore, the combination of IDO1 inhibition and PD-L1 blockade upregulated both adaptive and innate responses more than PD-L1 blockade alone. These findings elucidate the differential effects of the two therapies on the immune system and provide valuable insights for future treatment strategies targeting IDO1.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-025-04232-8.

Introduction

Over the past decade, immune checkpoint inhibitors (ICIs), such as anti–programmed cell death protein 1 (PD-1) and anti–programmed death-ligand 1 (PD-L1) antibodies, have been used to treat various types of cancers and have significantly improved patient prognosis. However, ICI monotherapy benefits only a subset of patients, suggesting the need for combination strategies to enhance therapeutic efficacy. Various combination therapies are currently being investigated in clinical trials. One mechanism of resistance to ICIs is the induction of indoleamine 2,3-dioxygenase 1 (IDO1) expression by interferon γ (IFN-γ) that is secreted in response to PD-1/PD-L1 blockade. IDO1 catabolizes tryptophan (Trp) to kynurenine (Kyn), and activation of the IDO-Kyn pathway suppresses the T cell-mediated antitumor immunity by inducing and activating regulatory T cells (Tregs) [1]. Therefore, targeting this pathway has been regarded as a promising strategy in cancer immunotherapy. Several IDO1 inhibitors have been evaluated in combination with ICIs in clinical studies; however, whether IDO1 is an effective target for cancer immunotherapy remains controversial, as clinical trials have failed to demonstrate the anticipated efficacy [2]. For example, the most advanced IDO1 inhibitor, epacadstat, failed to show additional clinical benefits in combination with pembrolizumab in patients with unresectable or metastatic melanoma [3]. There have been new entrants into the IDO1 inhibitor field, including KHK2455 and linrodostat [4–6]. KHK2455 is a long-acting, potent, and selective IDO1 inhibitor that blocks the heme component of the IDO holoenzyme. KHK2455 and linrodostat differ from epacadostat by competing with heme for apoenzyme binding, and thereby prevent apo-IDO1 from forming an active complex and resulting in durable inhibition. Recently, a phase 1 study of KHK2455 in combination with mogamulizumab in patients with advanced solid tumors was reported; however, the report presented primarily clinical trial data and did not contain a description of the effects of KHK2455 on immune cells within tumors [5]. Since it remains unclear whether IDO1 inhibition is beneficial for patients with cancer, it is crucial to elucidate the effects of IDO1 inhibition on the tumor microenvironment (TME) and understand the mechanisms underlying the potential enhanced antitumor effects of combining IDO1 inhibitors with ICIs. This may enhance the success rate of clinical trials and facilitate the identification of patients who will benefit from IDO1 inhibition.

In this study, we conducted a comprehensive analysis of gene expression in mouse tumors treated with KHK2455, an anti–PD-L1 antibody, or their combination, to investigate the effects of IDO1 inhibition on the TME and explore the mechanisms underlying the enhanced antitumor effects of the combination therapy.

Materials and methods

Test articles

KHK2455 was synthesized by Kyowa Kirin Co., Ltd. and suspended in 0.5 w/v% methylcellulose 400 (MC400) (Fujifilm Wako Pure Chemical) to obtain a 1 mg/mL suspension. Anti–mouse PD-L1 antibody (clone 10F.9G2) and the isotype control antibody (clone RTK4530) were purchased from BioLegend, Inc. These antibodies were diluted to 1 mg/mL with phosphate buffered saline (PBS) (Thermo Fisher Scientific).

Cell lines

A mouse breast cancer cell line, 4T1-Luc2 cells (Caliper Life Sciences), a mouse pancreatic cancer cell line, PAN 02 cells (National Cancer Institute (NCI) Division of Cancer Treatment and Diagnosis (DCTD) Tumor Repository), and a mouse melanoma cell line, B16-F10-Luc cells (gifted from Takashi Murakami, Saitama Medical University), were cultured in RPMI 1640 medium (Life Technologies) supplemented with 10 vol% heat-inactivated fetal bovine serum (FBS) (Life Technologies). B16-F10 cells expressing luciferase and cancer testis antigen NY-ESO-1, B16.F10.Luc.NY-ESO-1 cells, were cultured in DMEM (Life Technologies) supplemented with 10 vol% heat-inactivated FBS, penicillin–streptomycin mixed solution (final 100 Units/mL penicillin, 100 μg/mL streptomycin, Nacalai Tesque), and sodium pyruvate (final 1.9 mmol/L, Life Technologies). All cells were maintained at 37 °C under 5% CO₂.

Establishment of B16-F10 cells with a stable expression of human NY-ESO-1

The details are available in supplemental Methods.

Animal experiments

Two independent pharmacological studies were conducted using the B16.F10.Luc.NY-ESO-1 mouse model. Study 1 was designed to evaluate the long-term antitumor efficacy of the test articles. Tumor growth was monitored for up to 43 days following treatment initiation (n = 10 per group). Study 2 was performed to evaluate immune-related gene expression changes in the tumor microenvironment induced by each treatment. To this end, tumors and blood samples were collected on day 10 after treatment initiation (n = 15 per group). B16.F10.Luc.NY-ESO-1 cells (1 × 10⁶ cells/0.05 mL/mouse) suspended in PBS were subcutaneously inoculated into C57BL/6 J mice. Seven days after the inoculation (designated as day 0), the mice were divided into four groups (control group, KHK2455 group, anti–PD-L1 group, and combination group). In Study 1, KHK2455 or MC400 was orally administered once daily for 21 days (day 0 to 20), and in Study 2, for 10 days (day 0 to 9), at a dose of 10 mg/kg. Additionally, anti–PD-L1 antibody or the isotype control antibody was intraperitoneally administered at a dose of 100 μg/mouse on days 0, 2, and 4 in both Study 1 and Study 2. The details, including mouse studies using Pan02 and 4T1-Luc2 cells, are available in supplemental Methods.

Preparation of the plasma and tumors for liquid chromatography-tandem mass spectrometry and nCounter analyses

The details are available in supplemental Methods.

Liquid chromatography-tandem mass spectrometry analysis

The details are available in supplemental Methods.

Immune-related gene expression analysis

Total RNA was extracted from mouse tumor specimens collected on day 10 using NucleoSpin RNA (MACHEREY–NAGEL). The nCounter Mouse PanCancer IO 360 Panel plate loaded with RNA samples was analyzed by the nCounter Analysis System (NanoString Technologies) according to the procedures for hybridization, detection, and scanning. The details are available in supplemental Methods.

Statistical analysis

Differences between the two groups were examined using the Wilcoxon rank-sum test. Survival curves were analyzed using the Kaplan–Meier method, and survival comparisons between groups were performed using the log-rank test with Bonferroni correction for multiple comparisons. Differences in V_AUC observed in the B16.F10.Luc.NY-ESO-1 model across treatment groups were evaluated using closed testing procedures with the hierarchical evaluation. For other multiple comparisons, the Holm method was used to correct for multiple tests. In this study, an adjusted P < 0.05 was considered significant. The details are available in supplemental Methods.

Results

The combination of IDO1 inhibition and PD-L1 blockade exerts significant antitumor effects compared with PD-L1 blockade alone in mice bearing B16.F10.Luc.NY-ESO-1 melanoma

The antitumor activities of KHK2455, both as a monotherapy and in combination with ICIs such as anti–CTLA-4 and anti–PD-L1 antibodies, were evaluated using several syngeneic mouse models bearing 3 different murine tumor-derived cells, including those expressing foreign antigens, luciferase (Luc), or human NY-ESO-1 (Pan02, B16-F10-Luc, 4T1-Luc2, and B16.F10.Luc.NY-ESO-1). KHK2455 alone did not exhibit significant antitumor effects in mice bearing Pan02 pancreatic cancer cells (Fig. S1A), 4T1-Luc2 breast cancer cells (Fig. S1B), or B16-F10-Luc melanoma cells [5]. However, when the 4T1-Luc2 and B16-F10-Luc models were treated with KHK2455 in combination with the anti–CTLA-4 antibody, significant antitumor effects were observed in both models. The antitumor effect was markedly stronger in the B16-F10-Luc model than in the 4T1-Luc2 model (Fig. S1B, and ref [5]). Notably, the antitumor effects of KHK2455, whether administered alone or in combination with ICIs, showed substantial inter-individual variability, even within the B16-F10-Luc model. In contrast, inter-individual variability in antitumor responses was significantly less in the B16.F10.Luc.NY-ESO-1 model than the parental B16-F10-Luc model. Moreover, significantly stronger antitumor effects were observed when KHK2455 was combined with PD-L1 blockade, compared with PD-L1 blockade alone (Fig. 1A). The combination therapy also resulted in significantly prolonged survival compared with monotherapy either KHK2455 or PD-L1 blockade (P < 0.001 and P < 0.05, respectively, Fig. 1B). At 43 days after the first administration, tumor disappearance rates were 10% in the anti–PD-L1 monotherapy group and 60% in the combination therapy group (n = 10 per group, Table 1). To investigate the effects of IDO1 inhibition on the TME and to explore the mechanisms underlying the enhanced antitumor effects of the combination therapy, mouse tumors and plasma samples were collected on day 10 by conducting another pharmacological study under the same conditions and were used for subsequent analyses (Fig. 1C). Even on day 10, the combination therapy exerted significantly stronger antitumor effects than PD-L1 blockade alone (area under the curve for tumor volume–time, V_AUC: P < 0.05, Fig. 1D). Relative to the control group, tumor growth inhibition (TGI) at Day 10 was 68.5% for PD‑L1 blockade and 83.6% for the combination therapy. Over the prespecified evaluation window, the AUC‑based TGI (TGI_AUC) was 47.2 and 61.2% for PD‑L1 blockade and the combination therapy, respectively.

Fig. 1.

Effect of KHK2455, anti–PD-L1 antibody, and their combination on the growth of B16.F10.Luc.NY-ESO-1 melanoma in C57BL/6 J mice. The control group was administered both MC400 and the isotype control antibody, the KHK2455 group was administered both KHK2455 and the isotype control antibody, the anti–PD-L1 group was administered both MC400 and anti–PD-L1 antibody, and the combination group was administered both KHK2455 and anti–PD-L1 antibody. Day 0 was defined as the first day of treatment. A Tumor growth curves in each group. Each plot represents the mean ± standard error of the tumor volume (mm³) (n = 10/group). B Survival curves were analyzed using the Kaplan–Meier method (n = 10/group). Statistical tests were performed using the log-rank test. C Tumor growth curves in each group. Each plot represents the mean ± standard error of the tumor volume (mm³) (n = 15/group). D Area under the curve of the tumor volume, V_AUC. The dots represent individual tumors, whereas the bars indicate mean values (n = 15/group). Differences in V_AUC between the groups were assessed using the Wilcoxon rank-sum test with closed testing procedures with hierarchical evaluation. MC400, methylcellulose 400; PD-L1, programmed death-ligand 1; V_AUC, area under the curve of the tumor volume

Table 1.

Number of mice with tumor disappearance on day 43 in each treatment group

	Number of mice (n = 10/group)
Control	0
KHK2455	0
Anti–PD-L1	1
Combination	6

IDO1 inhibition decreases the Kyn concentration and Kyn/Trp ratio in tumors as well as plasma

Kyn and Trp concentrations were analyzed in both the plasma and tumors to assess whether KHK2455 inhibits Kyn production in vivo (Fig. 2). None of the treatment groups showed any significant differences in the plasma Trp concentration compared with the baseline control. In contrast, the KHK2455 and combination groups showed significantly lower plasma Kyn concentrations and Kyn/Trp ratios compared with the baseline control. About 60% reductions in the plasma Kyn concentrations from baseline were observed on day 10. Similar trends were also found in the tumor samples. Significantly lower tumor Kyn concentrations and Kyn/Trp ratios were observed in the KHK2455 and combination groups, with a greater than 70% reduction in the tumor Kyn concentration compared to baseline. In contrast to the plasma concentration, the Trp concentration in the tumors of each treatment group was significantly higher than that in the baseline control group (P < 0.05). Moreover, PD-L1 blockade significantly increased the Kyn concentration in tumors (P < 0.05). To investigate the mechanism underlying this increase, the IDO1 inhibitory activity of KHK2455 was evaluated in C57BL/6 J mice treated with IFN-γ. Twenty-four hours after IFN-γ administration, plasma Kyn levels were significantly elevated, and this increase was inhibited by KHK2455 in a dose-dependent manner (Fig. S2A). Additionally, PD-L1 blockade significantly increased Ifng gene expression in tumors from the same B16.F10.Luc.NY-ESO-1 mice whose tumors were used for Trp and Kyn analyses (Fig. S2B). This suggests that the observed increase in Kyn following anti–PD-L1 treatment may be attributed to elevated IFN-γ levels, likely induced by T cell activation through PD-L1 blockade.

Fig. 2.

Tryptophan and kynurenine concentrations and their ratio in the plasma and tumor specimens. Trp and Kyn concentrations and their ratios were measured in A the plasma and B tumor specimens. The dots represent individual plasma or tumor specimens, and the bars indicate the mean values (n = 5/group). Differences in each value between the baseline control and the KHK2455, anti–PD-L1, or combination groups were assessed using the Wilcoxon rank-sum test, and P-values were adjusted for multiple comparisons by Holm’s method. Kyn, kynurenine; PD-L1, programmed death-ligand 1; Trp, tryptophan

The combination of IDO1 inhibition and PD-L1 blockade increases immune cell abundance in tumors more than PD-L1 blockade alone

To investigate the mechanisms by which the combination of IDO1 inhibition and PD-L1 blockade exerts strong antitumor effects, tumors were subjected to comprehensive immune-related gene expression analysis using the NanoString nCounter Mouse PanCancer IO 360 panel. First, the abundance of 14 types of immune cells in the tumors were estimated by the cell type profiling method. In this study, Tregs, B cells, and mast cells were excluded from the analysis because all the probes used for cell profiling of these cell types were discarded as their signals were below the threshold expression level.

PD-L1 blockade increased the cell type scores of most immune cells compared with the control group; however, KHK2455 did not change these scores much (Fig. 3A). In contrast, KHK2455 in combination with PD-L1 blockade further increased each score compared with PD-L1 blockade alone. Significantly higher scores were obtained for CD45 cells, neutrophils, macrophages, and exhausted CD8 T cells in the combination group compared with the group with PD-L1 blockade alone (P < 0.05, Fig. 3B).

Fig. 3.

Abundance of immune cells and tumor inflammation signature in tumors from mice treated with KHK2455, anti–PD-L1 antibody, or their combination. The heatmap displays the cell type scores for each immune cell (A) and the expression levels for each gene in the TIS (C). Tumor specimens with similar cell type or gene expression profiles were clustered using the Ward’s method. The color scale represents the level of the cell type scores or gene expression, with red indicating higher scores/expression levels. B, D The dots represent individual tumors, and the bars indicate the mean values (n = 7/group). Differences in each cell type score and TIS between the combination and anti–PD-L1 groups were assessed using the Wilcoxon rank-sum test. DC, dendritic cell; NK, natural killer; PD-L1, programmed death-ligand 1; Th1, T-helper cell type 1; TIS, tumor inflammation signature

Since the anti-tumor effects of both anti-PD-L1 and combination groups were accompanied by some inter-individual differences (Fig. S3A), we divided the mice with available gene expression data into responder and non-responder subgroups within each treatment group (Fig. S3B). When comparing cell type scores between anti–PD-L1 responders and combination therapy responders, the differences in scores for CD45 + cells, neutrophils, macrophages, and exhausted CD8 + T cells became more pronounced compared to our original analysis (Fig. S4).

The addition of IDO1 inhibition to PD-L1 blockade does not significantly enhance adaptive immunity compared with PD-L1 blockade alone

As the activation of adaptive immunity, including T cells, mainly contributes to the antitumor effects of PD-L1 blockade, we assessed whether KHK2455, in combination with PD-L1 blockade, activates adaptive immunity significantly more than PD-L1 blockade alone using the gene signature scoring method. The tumor inflammation signature (TIS), an indicator of a pre-existing but suppressed adaptive immune response within tumors, is often used to assess the immune activation status in the TME. However, TIS was originally developed as a predictive indicator based on human data. Therefore, in the present study, mouse orthologs of each human gene from the TIS were used to define the mouse TIS (mTIS) (Fig. 3C). When adaptive immunity is activated, the expression levels of genes in the TIS have been reported to increase, except for Cd276, whose expression level decreases. In this study, we excluded the downregulated gene from the mTIS score calculation to facilitate the interpretation and analysis. Significantly higher mTIS scores were obtained in both the anti–PD-L1 and combination groups compared with the control. However, no significant difference in mTIS scores was observed between the anti–PD-L1 and combination groups. Additionally, KHK2455 did not increase the mTIS score considerably compared with the control (Fig. 3D).

IDO1 inhibition induces innate immune responses in tumors

Next, differentially expressed genes (DEGs) between each group were identified and pathway enrichment analysis using the DEGs was performed to explore the effects of IDO1 inhibition on the TME and the mechanisms of enhanced antitumor effects of IDO1 inhibition combined with PD-L1 blockade. Comparing the KHK2455, anti–PD-L1, and combination groups with the control group, we identified 19, 163, and 233 DEGs in the KHK2455, anti–PD-L1, and combination groups, respectively (Table 2). The top 10 pathways with lower P-values identified using pathway enrichment analysis for each group are listed in TablesS1, S2, and S3, respectively. The pathways related to innate immunity were the most enriched, and some of them were suggested to be upregulated in the KHK2455 group, including pathways related to phagosome, pattern recognition receptors, natural killer (NK) cells, neutrophils, and macrophages (Table S1). In contrast, in the anti–PD-L1 group, pathways related to adaptive immunity, such as T-helper cells type 1 and 2 (Th1 and Th2), were enriched and suggested to be upregulated (Table S2). A similar trend was observed in the combination group (Table S3). P-values of pathways other than the “PD-1, PD-L1 cancer immunotherapy pathway” were lower in the combination group compared with the anti–PD-L1 group. Similarly, for the DEGs, most P-values in the combination group were lower than those in the anti–PD-L1 group (Fig. 4A).

Table 2.

The number of DEGs between each group

	Number of upregulated genes	Number of downregulated genes	Total number of DEGs
KHK2455 vs control	13	6	19
Anti–PD-L1 vs control	161	2	163
Combination vs control	225	8	233
Combination vs anti–PD-L1	92	1	93

DEG, differentially expressed gene; PD-L1, programmed death-ligand 1

Fig. 4.

Relationships between the DEGs identified among each pair of groups. A Relationship between the P-values of the DEGs identified between the combination vs control and anti–PD-L1 vs control groups. For each DEG, the -log10 (P-value) for the combination vs control is plotted on the x-axis, while the − log10 (P-value) for the anti–PD-L1 vs control is plotted on the y-axis. Red circles represent DEGs identified only between the anti–PD-L1 and control groups, blue dots represent DEGs identified only between the combination and control groups, and purple dots indicate genes that are common to both comparisons. Among these, the genes that are also DEGs between the combination and anti–PD-L1 groups are represented by asterisks. B Venn diagram of the relationships between DEGs among each pair of groups. The number of genes corresponding to each section of the diagram is indicated. C Volcano plot showing DEGs between the combination and anti–PD-L1 groups. The open blue dots represent the DEGs between the combination and anti–PD-L1 groups, while the solid blue dots represent the 27 DEGs identified in the combination and anti–PD-L1 groups and combination and control groups but not in the anti–PD-L1 and control groups. D The top 10 enriched pathways from the 90 DEGs identified in both the combination vs control and combination vs anti–PD-L1 groups. Orange, blue, and gray bars represent upregulated, downregulated, and undetermined pathways, respectively. DEG, differentially expressed gene; PD-L1, programmed death-ligand 1

The combination of IDO1 inhibition and PD-L1 blockade activates not only adaptive immunity but also innate immunity more than PD-L1 blockade alone

Lastly, differences in gene expression profiles in the TME between the combination and anti–PD-L1 groups were analyzed. Ninety-three DEGs were identified between the groups (Table 2), and the relationships between the DEGs in each group (combination vs control, anti–PD-L1 vs control, and combination vs anti–PD-L1) are shown in Fig. 4B. Of the 93 DEGs identified between the combination and anti–PD-L1 groups, 90 were also DEGs between the combination and control groups, and 63 of these were also DEGs between the anti–PD-L1 and control groups. The remaining 27 genes exhibited no significant differences in expression between the anti–PD-L1 and control groups (Fig. 4B and C). To explore the mechanisms underlying the enhanced antitumor effects of combining IDO1 inhibition and PD-L1 blockade, pathway enrichment analysis was performed using the 90 DEGs identified in both the combination vs control groups and the combination vs anti–PD-L1 groups. The pathways related to innate immunity were the most enriched, and most of them were suggested to be upregulated, including those related to neutrophils and macrophages (Fig. 4D, Table S4).

Discussion

This study demonstrated that the selective IDO1 inhibitor, KHK2455, decreased Kyn levels in mouse tumors and plasma and enhanced antitumor efficacy when combined with PD-L1 blockade. Comprehensive gene expression analysis suggested that this combination therapy activates not only adaptive immunity but also innate immunity, providing new insight into the multifaceted immunomodulatory effects of IDO1 inhibition.

One of the mechanisms of action of KHK2455 is the inhibition of IDO1-mediated immunosuppression induced by IFN-γ from T cells or NK cells. Therefore, we hypothesized that a certain level of immunostimulation is required for KHK2455 to exert its antitumor effects. Yuan et al. reported that NY-ESO-1-specific antibody and CD8+ T cell responses were correlated with clinical benefit in melanoma patients treated with the anti–CTLA-4 antibody ipilimumab [7]. Based on this, a B16-F10-Luc cell line expressing human NY-ESO-1 (B16.F10.Luc.NY-ESO-1) was established and used in this study.

KHK2455 effectively inhibited Kyn production in both the mouse plasma and tumors. However, a significantly stronger antitumor effect was not observed with KHK2455 compared with the control, suggesting that for exerting strong antitumor effects, Kyn suppression alone may not be sufficient or the degree of Kyn suppression may be inadequate. As tryptophan 2,3-dioxygenase (TDO) also helps in the metabolism of Trp to Kyn [8], TDO inhibition might also be important to inhibit Kyn production sufficiently. In contrast, PD-L1 blockade increased the intratumoral Kyn levels, consistent with the previous findings that PD-1–PD-L1 axis blockade upregulates IDO expression via IFN-γ, which in turn induces Kyn production [9]. Indeed, IFN-γ treatment increased plasma Kyn concentrations, and this increase was inhibited by KHK2455 in a dose-dependent manner, supporting the notion that IFN-γ induced by PD-L1 blockade enhances Kyn production via IDO1 activation.

Cell type profiling revealed that PD-L1 blockade increased the infiltration of multiple immune cell types, whereas KHK2455 alone had little effect. Notably, the combination therapy further increased the scores of neutrophils, macrophages, and exhausted CD8 T cells, compared with PD-L1 blockade alone. The present study relied on gene expression data to estimate changes in tumor-infiltrating immune cells. Several studies have demonstrated that results from this method correlate well with those obtained from flow cytometry or immunohistochemistry [10, 11]. However, it has also been shown that, while flow cytometry reveals changes in specific cell populations, the NanoString nCounter analysis identifies activation of multiple immune pathways from comprehensive gene expression data, thereby providing complementary information [12]. Thus, further investigations using flow cytometry or immunohistochemistry are warranted.

The TIS score is widely recognized as a measure of the immune activation state of the TME and extent of immune cell infiltration into tumors, particularly T cell infiltration [13, 14]. When applying gene signatures developed based on human data to our mouse gene expression data, two issues must be considered. First, species differences in the immune system exist, which raises the question whether human gene signatures are applicable to the mouse model. Second, the scoring formula of some gene signatures, including TIS, is not yet publicly available. Therefore, we used a signature scoring method “singscore” for calculating mTIS, to address the above-mentioned limitations [15]. As expected, the mTIS was not increased by KHK2455, suggesting that IDO1 selective inhibition and suppression of Kyn production are not sufficient to make the TME both sensitive to T-cell–based immunotherapy and immunologically hot. High positive correlations were found between the TIS scores obtained from the singscore, gene set variation analysis (GSVA) [16], and single-sample gene set enrichment analysis (ssGSEA) methods [17] (data not shown), and similar trends were found between the pathway analysis results and our signature analysis results. Moreover, high positive correlations were confirmed between the mTIS score and cell type scores such as CD8 T cells and cytotoxic cells. High negative correlations were also confirmed between the mTIS and antitumor effects (V_AUC and tumor volume at day 10, data not shown). These results support the validity of our analysis method. Since the original TIS algorithm is applicable only to human data, our approach enables its extension to mouse models, thereby providing a useful tool for preclinical research and offering broad relevance to the field.

The cell type profiling and mTIS analysis suggested that the antitumor effects of combining IDO1 inhibition with PD-L1 blockade may be more influenced by the innate immune system, including macrophages and neutrophils, rather than by the adaptive immune system centered on T cells. The analysis of DEGs between the KHK2455 and control groups revealed the upregulation of innate immunity-related pathways, including those associated with phagocytotic activity and the responses of NK cells, neutrophils, and macrophages. This finding is consistent with previous reports that the IDO1-Kyn pathway suppresses NK cells [18] and decreases the macrophage phagocytic activity [19]. However, in general, adaptive immunity, including Th1 and CD8+ T cells, is considered to be important for antitumor effects [20]. This may partly explain why the antitumor effect of KHK2455 monotherapy was not as strong as we had expected in our mouse model.

The analysis of DEGs between the combination and anti–PD-L1 groups also revealed the upregulation of innate immunity-related pathways, including those involving neutrophils and macrophages. Tumor-associated neutrophils (TANs) and macrophages (TAMs) are often correlated with poor prognosis and resistance of immunotherapy. However, they are heterogeneous and considered to be composed of tumor-promoting (N2-TAN, M2-TAM) and antitumor (N1-TAN, M1-TAM) populations [21]. In this study, we were unable to determine whether the immune cell populations that increased following the combination therapy acted as antitumor or tumor-promoting cells. Moreover, the activation states of each immune cell type and its associated pathways were inferred solely from transcriptomic data. Future studies are needed to clarify these points through more direct approaches, such as flow cytometry and single-cell analyses.

The anti–mouse PD-L1 antibody used in this study (10F.9G2) is IgG2b, which is the main antibody isotype mediating effector functions such as antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis in mice. Furthermore, Dahan et al. demonstrated that the anti–mouse PD-L1 antibody, 10F.9G2, displayed significantly enhanced antitumor activity when activating fragment crystallizable gamma receptor (FcγR) engagement was optimized [22]. Our data suggest that IDO1 inhibition by KHK2455 enhanced the activities of FCγR+ effector cells, such as NK cells and macrophages. Additionally, KHK2455 activates NK cells and increases ADCC activity [23]. The FCγR-mediated antitumor effects may contribute to the enhanced antitumor effects of the combination therapy of KHK2455 and PD-L1 blockade. KHK2455 was evaluated in a phase 1 study in combination with avelumab, an anti–PD-L1 antibody with ADCC activity, in patients with locally advanced or metastatic urothelial carcinoma (NCT03915405, 2455–002). Thus, further investigations including analysis for clinical specimens are warranted to verify this hypothesis.

The clinical failures of IDO inhibition provide critical lessons for guiding future drug development [24, 25]. A key challenge that must be addressed is how to overcome the gap between preclinical results and clinical outcomes, particularly those arising from interspecies differences, including variations in the immune system and differences in tumor dependence on the IDO1–Kyn pathway. Furthermore, Kyn has been shown to increase PD-1 expression on T cells in an AhR-dependent manner [26]. It is also possible that inhibition of PD-1 pathway by PD-1 blockade is more complete in humans than in mice, suggesting that mouse models may have a window of antitumor effect enhancement through IDO1 inhibition that could be absent in humans [25]. These indicate that alternative combination strategies with IDO1 inhibition, other than PD-1/PD-L1 blockade should be considered as well. In this study, an additive antitumor effect of IDO1 inhibition in combination with PD-L1 blockade was observed in mice; however, further clinical validation is required to determine whether such an effect also occurs in humans. For such validation, it may be necessary to stratify patients according to IDO1 expression levels or the degree of tumor dependence on the IDO1-Kyn pathway, and to consider additional concomitant therapies beyond PD-1/PD-L1 blockade.

Overall, these findings suggest that KHK2455 in combination with PD-L1 blockade has the potential to not only enhance adaptive immunity but also activate innate immunity, resulting in stronger antitumor effects compared with PD-L1 blockade alone (Fig. 5). Although multiple mouse tumor models were used to evaluate antitumor efficacy in this study, comprehensive gene expression profiling of the TME was performed only in one model, in which 770 immune-related genes were analyzed. Therefore, further investigations regarding changes in the TME with IDO1 inhibition using more in-depth analyses, such as whole transcriptome analysis and single-cell analysis, are warranted.

Fig. 5.

Summary of tumor microenvironment modulation and antitumor responses across different therapeutic regimens

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M.S., T.I., R.O., Y.K., and T.Y. designed the research. M.S., K.K., T.I., and S.K. performed experiments. M.S., K.K., T.I., S.K., R.O., Y.K., and T.Y. analyzed and interpreted data. All authors read and approved the final manuscript.

Funding

This work was financially supported by Kyowa Kirin Co., Ltd. through collaborative research.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available as there are no public repositories for this type of dataset. The data are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

M.S., K.K., S.K., T.I., R.O., and Y.K. are employees of Kyowa Kirin Co., Ltd. T.Y. is a founder of Metagen Inc., Metagen Therapeutics Inc., and digzyme Inc. Metagen Inc. focuses on the design and control of the gut environment for human health. Metagen Therapeutics Inc. focuses on drug discovery and development which utilizes microbiome science. All of the companies had no control over the interpretation, writing, or publication of this work. The terms of these arrangements are being managed by the Institute of Science Tokyo in accordance with its conflict of interest policies.

Ethics approval

All animal studies were performed in accordance with the Standards for Proper Conduct of Animal Experiments at Kyowa Kirin under the approval of the company’s Institutional Animal Care and Use Committee. Kyowa Kirin is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care, International.