Molecular Pathways: Targeting IDO1 and Other Tryptophan Dioxygenases for Cancer Immunotherapy

The immune system is composed of an immediate-acting innate arm comprised principally of granulocyte- and myeloid-lineage cells that quickly respond to cues of inflammation and/or injury, in addition to an adaptive arm, principally comprised of B and T cells that provide specificity and memory. Under normal circumstances, these immunologic arms are mutually dependent on one another for providing defense against infection, injury, and/or malignancy. T cells, which primarily mature following immunologic challenge(s), include CD4⁺ and CD8⁺ T lymphocytes that express a wide variety of cytokines based on the context of priming stimuli. Included in the CD4⁺ T-cell compartment are highly immunosuppressive regulatory T cells (Treg; CD4⁺CD25⁺FoxP3⁺CTLA-4⁺) that mature naturally in the thymus (nTreg) or are post-thymically induced from naïve CD4⁺Foxp3⁻ cells into Foxp3-expressing cells (iTreg; refs. 1–3). With respect to solid cancer(s), immunosuppressive mechanisms utilized to evade antitumor immunity include Treg accumulation (4, 5) effector T-cell expression of the PD-1 receptor (6), as well as high PD-L1 levels that localize to multiple types of cells in the tumor microenvironment (7, 8). Therefore, an active effort both clinically and preclinically is needed to develop strategies that reactivate a productive antitumor effector T-cell response, while simultaneously inhibiting immunosuppressive mechanisms.

Recent studies have demonstrated great promise at targeting immunosuppression in cancer, including clinical trials aimed at inhibiting PD-1, PD-L1, and/or CTLA-4 in patients diagnosed with late-stage melanoma, non–small cell lung cancer, and/or renal cell cancer (9–12). Follow-up studies have also shown that the benefit of combined PD-1/CTLA-4 inhibition is not restricted to those patients previously treated with systemic therapy (13). Preclinical work using multiple tumor models in immunocompetent mice further confirms that these immune checkpoint–targeted therapies require effector T cells for antitumor activity, with several studies reporting a coincident neutralization of tumor-infiltrating Treg (14–16). These clinical studies, combined with extensive preclinical validation of combinatorial approaches confirm that immunotherapy is a high-value strategy for treating patients with aggressive and immunosuppressive malignancies.

_L-Tryptophan (_L-Trp) is used in a variety of anabolic/catabolic processes and metabolized into serotonin, melatonin, protein, and Kyn. IDO1 and TDO are the primary enzymes that catalyze the rate-limiting cleavage of the Trp indole ring 2,3-double bond and incorporation of molecular oxygen. The product of this reaction is N-formylkynurenine, which is rapidly and spontaneously converted into _L-Kyn. The latter catabolite is further converted into downstream intermediates, including 3-hydroxy-_L-kynurenine (3-HK), 3-hydroxyanthranilate (3-HAA) and quinolinic acid (Quin), which also impact immune responses (17).

Although IDO1 and TDO both catabolize Trp, their quaternary structures (18, 19), expression in normal versus transformed tissue (20, 21) and regulation (22, 23) are quite distinct. While monomeric IDO1 acts on a broad range of substrates and is capable of cleaving both _D- and _L-Trp, homotetrameric TDO is enantiomer-specific and only catabolizes _L-Trp (24). IDO1 expression in adults is relatively limited to lymphoid tissues and placenta (20), whereas TDO is constitutively expressed in liver and brain (25, 26), likely reflecting their primarily immunomodulatory or energy regulating roles, respectively. Until 2007, IDO1 was the only known indoleamine dioxygenase acting at the 2, 3 double bond. Three independent groups then identified the novel paralog, IDO2 (27–29). While the IDO1 and IDO2 genes are 43% homologous and found directly adjacent to one another on chromosome 8, the K_m of human IDO1 and IDO2 for _L-Trp is 20.90 ± 3.95 μmol/L and 6,809 ± 917 μmol/L, respectively, indicating a substantial decrease in activity for the latter enzyme (30). This is particularly interesting given that the residues required for tryptophan catalytic activity are present in both gene products (27). Also notable is that mouse IDO2 has been shown to possess higher enzymatic activity than the human homolog, although the genetic depletion of mouse IDO2 has no impact on systemic Kyn levels (31), a dramatic contrast to the impact of IDO1 deficiency (32).

Because of IDO1 expression induced in response to infection, it was originally thought that it serves as an innate immune effector to restrict the amount of Trp required for microbial growth (33). This initial hypothesis was revised by Munn and colleagues, who demonstrated that the in vivo administration of an IDO1 inhibitor, 1-methyl tryptophan (1-MT), led to T-cell–dependent fetal allograft rejection (34). Subsequent work demonstrated that IDO1-expressing macrophages, dendritic cells (DC), and tumor cells mediate the inhibition of T-cell proliferation (35–38). IDO1 responses were found to be mediated by downstream stress–response pathways including general control non-depressible 2 (GCN2) and mTOR; both important regulators that sense amino acid sufficiency (Fig. 1). The GCN2 pathway is activated when amino acid deficiency increases overall uncharged tRNA levels, resulting in GCN2 kinase phosphorylation of the alpha subunit of translation initiation factor 2 α (eIF2α) and subsequent inhibition of translation. It was first discovered that GCN2-activated plasmacytoid DC could suppress T-cell proliferation in vivo by an IDO1-dependent mechanism (39). It was later discovered that the genetic deletion of IDO1, but not GCN2, prevented skin carcinogenesis in a mouse papilloma model, suggesting that additional critical pathways were downstream of IDO1 activity (40). In support of these findings, Metz and colleagues identified that IDO1-mediated Trp depletion suppressed mTOR, a critically important immunoregulatory kinase (40) that could be reactivated by treatment with _D-1-MT, a Trp mimetic, in vitro.

IDO1-mediated suppression of T-cell activity is hypothesized to rely on the depletion of free Trp. This premise requires cell-specific transport mechanisms that include both the transporter System L, which shuttles Trp and other large hydrophobic amino acids through a low-affinity (K_m = 20–30 μmol/L; ref. 41) interaction, as well as through an independent high-affinity (K_m = 200–300 nmol/L) interaction. Interestingly, the high-affinity transporter is upregulated in differentiated myeloid-derived macrophages (MDM) but not in T cells. In support of the requirement for transport, both Trp and the competitive inhibitor, _L-1-MT, inhibit Trp uptake into cells, collectively suggesting that competitive IDO1 inhibitors target the transporter and enzyme, simultaneously.

The literature is replete with redundant pathways that lead to IDO1 expression and activity. Proinflammatory signals including IFNγ, CpG DNA, and LPS are potent inducers of IDO1 expression (33, 42–44). Cytokines, including TNFα, IL6, and IL1β, synergize with each other to dramatically increase IDO1 expression. Other IDO1 modulators include soluble GITR, prostaglandin E2, the oncogene, c-Kit, as well as the tumor suppressor, Bin1 (45). Interesting new data suggests that Wnt5α also mediates IDO1 activity through β-catenin signaling in DC (46), while maintaining continuous expression through an AhR–IL6–STAT3 signaling loop in some cancer cell lines (47). Thus, based on the large number of pathways that modulate and/or sustain IDO1 expression/activity, the direct targeting of IDO1, rather than pathways that are up- or downstream, will likely be the most effective modality for controlling the overall impact mediated by this Trp dioxygenase.

Similar to IDO1, TDO mRNA expression has also been found in human tumors (21). Dominant factors that affect TDO expression and/or activity include sex steroid hormones (48) and glucocorticoids (22). New preclinical data also suggest that tumor-infiltrating T cells may regulate TDO expression based on findings from intracranially injected syngeneic murine brain tumors grown in Rag1^−/− mice (15). Notably, intraperitoneal injection of mastocytoma cells overexpressing TDO induces potent immunosuppression that can be reversed with a pharmacologic inhibitor of enzymatic activity, leading to immune-mediated tumor rejection (P < 0.001; ref. 21).

In contrast, the newest member of the tryptophan catabolic family, IDO2, has yet to be confirmed as a critical contributor to Kyn accumulation and tumor immunity. Notably, while mouse IDO2 possesses some capacity for Trp→Kyn conversion, the human ortholog is devoid of the same enzymatic capacity at physiologic Trp levels (30). Furthermore, transcriptome analysis of 129 human tumor samples and 25 human tumor cell lines has demonstrated limited IDO2 expression (49). As IDO2 was originally cloned from the liver (27), it is still unknown whether there are IDO2 splice variants specific to subtypes of differentiated and/or transformed tissues.

The interactions among inflammation, IDO1, and cancer (50, 51) are noteworthy and raise critical questions regarding how and when to optimally target tryptophan catabolism for therapeutic purposes. Furthermore, despite the presence of antigen-specific T cells within the microenvironment, tumors often escape, immunologically, without loss of antigen expression or presentation (MHC molecule) capacity. This effect is mediated, in-part, through the induction, upregulation, and/or enhanced participation of immunosuppressive T-cell–impairing ligands, CTLA-4 and PD-L1 (52). Similar to PD-L1, IDO1 expression also increases through a response to IFNγ released in the tumor microenvironment (53) as a potent compensatory mechanism contributing to the resistance of productive antitumor immunity (54). Interestingly, only a subset of patients have a T-cell–infiltrating presence within the tumor microenvironment, an observation reported for head and neck and bladder cancer, as well as melanoma, lung adenoma, and glioblastoma (55). A notable observation from those patients treated with the immune checkpoint inhibitor PD-1, correlates a high degree of clinical response to the pre-existence of tumor-infiltrating T cells (56). This observation, paired with the association of IDO1 induction by T-cell–derived IFNγ, leads to the hypothesis that IDO1 inhibitors will be most effective against T-cell–inflamed tumors, either de novo or caused by immunotherapeutic intervention. Preclinical studies support this hypothesis, establishing evidence that combinatorial immune checkpoint blockade and IDO1 pathway inhibition provide potent reactivation of tumor-infiltrating T cells and/or decreased tumor-resident immunosuppressive regulatory T cells (P < 0.01; refs. 15, 57).

No IDO1 inhibitor is currently approved by the FDA. However, results of recent phase I–II studies suggest that indoximod (_D-1-MT), INCB024360, and/or IDO1-targeting vaccines are well tolerated by cancer patients, with clinical anticancer effects in a subset of patients (58, 59). Notably, the number of clinical trials focused on IDO1 has recently grown in size, with many coupling multiple modalities to test the combinatorial benefit (Table 1). These recent reports, in addition to preclinical data suggest that combining tryptophan enzyme targeting with chemotherapy, radiotherapy, and/or immunotherapy may be an effective tool against a wide range of malignancies.

The seminal observation associating IDO1, immunosuppression, and cancer utilized a polyclonal antibody to identify the immunohistochemical frequency of expression among different human malignancies (60). Unexpectedly, recent analyses utilizing a novel monoclonal anti-human IDO1 antibody have demonstrated distinct differences compared with those original observations (20). While it was initially reported that 90% to 100% of human prostate and pancreatic tumors, as well as glioblastoma, were IDO1 positive, the latter study found only 42%, 38%, and 8% of those malignancies positive, respectively. As the antibodies were well vetted in both investigations, these conclusions present a cautionary tale that likely reflects more than simple differences in antibody specificity, but more broadly, the potential for alternative splice variants and/or posttranslational modifications resulting in antigenic variation. Thus, immunohistochemical studies associating IDO1 expression and survival should be interpreted carefully (61). Furthermore, these conflicting findings complicate strategies that would ideally use IDO1 IHC as a prognostic tool for selecting patients who would benefit most from IDO1 inhibition.

Recent work studying the Kyn/Trp ratio in patients with glioblastoma has suggested that analyzing a time point well after surgical tumor resection of 10+ weeks following the procedure, may be prognostically valuable to clinicians planning to enroll patients in immunotherapy trials (62). While this finding requires further validation in a larger patient cohort, it suggests the possibility that IDO1 activity increases well after glioblastoma patients are operated on, as well as highlighting the potential relevance of using a clinical inhibitor against IDO1 systemically. Similarly, the Kyn/Trp ratio was recently validated as a prognostic tool in cervical cancer patients whereby low Trp levels indicated a tumor size greater than 4 cm and metastatic spread to the lymph node (63). Accordingly, high Kyn/Trp ratios in patient sera were associated with lymph node metastasis, FIGO stage, tumor size, parametrial invasion, and poor disease-specific survival, further suggesting the relevance of IDO1 targeting based on a tryptophan catabolic signature. Similar work was recently shown in a clinical study that identified higher Kyn/Trp ratios in T-cell lymphotropic virus type-1 asymptomatic carriers when compared with healthy controls (64). Importantly, the serum Kyn/Trp ratio was a significantly independent detrimental prognostic factor in patients with adult T-cell leukemia/lymphoma. These collective analyses have begun to elucidate the relevance of determining an IDO1 enzymatic “signature” in patient sera, which preliminarily appears to be both prognostically valuable and clinically informative.

Given that the majority of clinical studies aimed at IDO1 inhibition that are currently ongoing have yet to report results, we can gain insight into preclinical analyses that have shown great potential in targeting this immunosuppressive mediator. However, these models possess limited usefulness when considering the potential effects that standard-of-care treatment have on IDO1 activity and/or expression, as well as the potential change of expression between primary and recurrent tumors. Given that inflammation is a primary driver of IDO1 expression, it may be relevant to prognostically stratify tumors that possess a wide range of T-cell–infiltrating heterogeneity, when compared with the primary, versus relapsed malignancy (65, 66).

Our substantial knowledge of the role and expression of IDO1 in cancer has continued to expand over the past two decades, yet critical questions regarding alternative functions regulated by posttranslational modifications, the role that IDO2/TDO plays in the absence or inhibition of IDO1, as well as the impact of tissue-specific alternative splicing, still remain. Most inhibitory strategies against IDO1 focus on disabling enzymatic activity. However, preclinical mouse tumor models suggest that this tactic alone will not lead to effective antitumor immunity, further suggesting that IDO1 inhibition is best suited for combinatorial therapeutic strategies. However, these findings also raise the intriguing, yet unproven possibility that IDO1 subsumes a new/alternative immunosuppressive role when Trp catabolism is abrogated in vivo. In support of this hypothesis, it is notable that indoximod (_D-1-MT), currently cast as an IDO1 pathway inhibitor, does not inhibit Trp to Kyn catabolism (refs. 67, 68; Supplementary Table S1). This combination of reported observations and untested hypotheses paints a blurry picture of a highly immunosuppressive player in tumor immunity. Unmistakably, IDO1 is a critical mediator that, given the normal limited expression throughout the body, makes it an ideal target for cancer immunotherapy. The central question going forward, thus becomes, how can we best inhibit the activity of this pleiotropic target?

No potential conflicts of interest were disclosed.

Conception and design: L. Zhai, G. Gritsina, F.J. Giles, D.A. Wainwright

Development of methodology: L. Zhai, D.A. Wainwright

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Zhai, D.A. Wainwright

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Zhai, F.J. Giles, D.A. Wainwright

Writing, review, and/or revision of the manuscript: L. Zhai, S. Spranger, D.C. Binder, G. Gritsina, K.L. Lauing, F.J. Giles, D.A. Wainwright

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.L. Lauing

Study supervision: D.A. Wainwright

The authors thank Michael Gallagher for his expertise in medical illustration and contribution toward creation of Fig. 1.

S. Spranger is supported by a Cancer Research Institute Postdoctoral Fellowship. D.A. Wainwright is supported by PHS grant number R00NS082381, awarded by the NINDS, U.S. Department of Health and Human Services; a Robert H. Lurie Comprehensive Cancer Center—Zell Scholar Program of the Zell Family Foundation Gift; and the Northwestern Brain Tumor Institute.