IDO1 (indoleamine 2,3-dioxygenase 1) is a member of a unique class of mammalian haem dioxygenases that catalyse the oxidative catabolism of the least-abundant essential amino acid, L-Trp (L-tryptophan), along the kynurenine pathway. Significant increases in knowledge have been recently gained with respect to understanding the fundamental biochemistry of IDO1 including its catalytic reaction mechanism, the scope of enzyme reactions it catalyses, the biochemical mechanisms controlling IDO1 expression and enzyme activity, and the discovery of enzyme inhibitors. Major advances in understanding the roles of IDO1 in physiology and disease have also been realised. IDO1 is recognised as a prominent immune regulatory enzyme capable of modulating immune cell activation status and phenotype via several molecular mechanisms including enzyme-dependent deprivation of L-Trp and its conversion into the aryl hydrocarbon receptor ligand kynurenine and other bioactive kynurenine pathway metabolites, or non-enzymatic cell signalling actions involving tyrosine phosphorylation of IDO1. Through these different modes of biochemical signalling, IDO1 regulates certain physiological functions (e.g. pregnancy) and modulates the pathogenesis and severity of diverse conditions including chronic inflammation, infectious disease, allergic and autoimmune disorders, transplantation, neuropathology and cancer. In the present review, we detail the current understanding of IDO1’s catalytic actions and the biochemical mechanisms regulating IDO1 expression and activity. We also discuss the biological functions of IDO1 with a focus on the enzyme's immune-modulatory function, its medical implications in diverse pathological settings and its utility as a therapeutic target.

INTRODUCTION

First identified and isolated from the rabbit intestine in 1967 [1,2], IDO1 (indoleamine 2,3-dioxygenase 1) [UniProt ID P14902 (human); EC 1.13.11.52] is a non-secreted intracellular haem enzyme that catalyses the conversion of the least-abundant essential amino acid, L-Trp (L-tryptophan) into NFK (N-formylkynurenine), which represents the initial and rate-limiting enzyme of the Kyn (kynurenine) pathway of L-Trp metabolism.

Primarily induced at sites of inflammation and infection, IDO1 depletes L-Trp from the local microenvironment, resulting in the concomitant production of downstream bioactive Kyn pathway metabolites that together contribute to the physiological and pathological actions of IDO1, including suppression of the immune response during pregnancy, cancer, inflammation and autoimmunity, as well as representing a host response against micro-organisms. Although IDO1-mediated immune suppression is protective under many circumstances, it is increasingly recognised that IDO1 is employed by certain types of cancers or microbial agents to avoid immune clearance, thus positioning IDO1 as an attractive therapeutic drug target. Notably, IDO1-targeted inhibitory drugs are in clinical trials as a novel class of immunotherapeutic anti-cancer adjuvant drugs. Moreover, it is becoming increasingly clear that the ability of IDO1 to influence immune responses and disease pathogenesis or severity is complex, involving various different immune cell types, several modes of biochemical or metabolic signalling, and, in some cases, the ability to mediate immune stimulatory actions. IDO1 and the Kyn pathway are also capable of mediating several non-immune biological actions, including roles in neuropathology, cataractogenesis, control of vascular tone and blood pressure and bone remodelling.

The present review provides a comprehensive examination of the current understanding of the dioxygenase catalytic mechanism and other reactions catalysed by IDO1, describes the molecular mechanisms controlling IDO1 expression and enzyme activity, and details the role of IDO1 under both physiological and pathophysiological settings, with a particular focus on the enzyme's role as a prominent immune regulatory mechanism and key host response to microbial infection.

IDO1 and the Kyn pathway of L-Trp metabolism

Mammals are incapable of synthesising L-Trp. As such, this amino acid is obtained solely via dietary intake. Approximately 95% of the free L-Trp in the body is metabolised via the Kyn pathway, with 1–5% shuttled down the methoxyindole pathway for conversion into serotonin and melatonin, and the remainder employed for protein synthesis [3]. Of the essential amino acids, L-Trp is the least abundant, with plasma concentrations of ∼40–80 μM in humans [4] and ∼60–100 μM in mice [5,6].

IDO1 is an initial and rate-limiting enzyme of the Kyn pathway, where it catalyses the oxidative cleavage of the indole ring of L-Trp, producing NFK, which is further converted into formic acid and the stable end-product Kyn, either spontaneously under acidic conditions or enzymatically via formidase [7] (Figure 1). Depending on the cell type and the different expression levels of other Kyn pathway enzymes, a cascade of reactions occur, yielding various biologically active Kyn pathway metabolites including Kyn, kynurenic acid, anthranilic acid, 3-HK (3-hydroxykynurenine), 3-HAA (3-hydroxyanthranilic acid), picolinic acid and QA (quinolinic acid) (Figure 1). NAD+ can be subsequently generated from QA. As NAD+ is an essential cofactor for many important cellular reactions and processes, ranging from ATP synthesis to DNA repair, the maintenance of adequate intracellular levels of NAD+ is critical for cellular homoeostasis, function and survival [8]. Accordingly, IDO1 activation and resultant L-Trp metabolism along the Kyn pathway can protect against oxidative stress via promoting de novo NAD+ synthesis in a human astroglioma cell line exposed to hydrogen peroxide (H2O2) [9].

Kyn pathway of L-Trp metabolism

Figure 1
Kyn pathway of L-Trp metabolism

Approximately 95% of the free L-Trp in the body is metabolised down the Kyn pathway, generating several biologically active metabolites, including kynurenine (Kyn), kynurenic acid, 3-hydroxykynurenine (3-HK), 3-hydroxyanthranilic acid (3-HAA), picolinic acid and quinolinic acid (QA), as well as NAD, an essential cofactor for many cellular reactions, including the synthesis of ATP. The relative proportion of each of these metabolites produced is dependent on the expression of the other Kyn pathway enzymes, which is cell-type- and stimuli-dependent. The remaining L-Trp is either shuttled through the methoxyindole pathway to generate serotonin and melatonin or used for protein synthesis. Incorporation of L-Trp into protein requires tryptophanyl-tRNA synthetase (TTS), an enzyme that catalyses the association of L-Trp with its tRNA.

Figure 1
Kyn pathway of L-Trp metabolism

Approximately 95% of the free L-Trp in the body is metabolised down the Kyn pathway, generating several biologically active metabolites, including kynurenine (Kyn), kynurenic acid, 3-hydroxykynurenine (3-HK), 3-hydroxyanthranilic acid (3-HAA), picolinic acid and quinolinic acid (QA), as well as NAD, an essential cofactor for many cellular reactions, including the synthesis of ATP. The relative proportion of each of these metabolites produced is dependent on the expression of the other Kyn pathway enzymes, which is cell-type- and stimuli-dependent. The remaining L-Trp is either shuttled through the methoxyindole pathway to generate serotonin and melatonin or used for protein synthesis. Incorporation of L-Trp into protein requires tryptophanyl-tRNA synthetase (TTS), an enzyme that catalyses the association of L-Trp with its tRNA.

Cytosolic expression of active IDO1 can act as an intracellular sink for L-Trp, resulting in a reduction in local extracellular tissue levels of the amino acid. This is because L-Trp is transported into IDO1-expressing cells from the extracellular environment via LATs (system L amino acid transporters), which have modest affinity for L-Trp and other large neutral amino acids, or a more recently identified tryptophan-selective transporter with 100-fold higher affinity for L-Trp (Km<300 nM) that is expressed by macrophages, but not in resting T-cells [1012]. Studies in human colonic epithelial cell and murine DC (dendritic cell) lines indicate that the cellular uptake of L-Trp can also be enhanced by stimulation with IFNγ (interferon-γ) and AhR (aryl hydrocarbon receptor) agonists, including Kyn [13]. In essence, this means that under in vitro conditions L-Trp can be completely consumed from the medium by intracellular active IDO1 and that in vivo, the local L-Trp concentration in tissue microenvironments can become significantly reduced [14,15].

The Kyn and other downstream metabolites generated following IDO1 activation can be released by IDO1-expressing cells into the extracellular milieu for uptake by neighbouring cells via LAT1 and binding to the AhR in the cytosol, to represent a paracrine mode of cell signalling (in the case of Kyn), or human organic anion transporters hOAT1 and hOAT3 (for kynurenic acid) [1618]. Also, QA is an NMDA (N-methyl-D-aspartate) receptor agonist, whereas kynurenic acid is an antagonist of this receptor [19], as well as a ligand of the α7 nicotinic acetylcholine receptor [20]. In this manner, IDO1-expressing cells can influence the function and phenotype of surrounding cells that do not necessarily express IDO1 themselves (e.g. T-cells; see the ‘Lymphocytes’ section below).

IDO1 and other L-Trp-catabolising haem dioxygenase enzymes

IDO1 (also referred to as IDO throughout the literature) belongs to a unique class of mammalian haem dioxygenase enzymes that includes IDO1, TDO (tryptophan 2,3-dioxygenase) [UniProt ID P48775 (human); EC 1.13.11.11] and a more recently discovered novel IDO isoform, IDO2 [UniProt ID Q6ZQW0 (human)]. Although these enzymes both catalyse the first reaction in the Kyn pathway, there are significant differences between them.

Unlike IDO1, which is highly induced by various inflammatory stimuli in many cell types and tissues in mammals, TDO is primarily localised to the liver and brain and appears largely unresponsive to immunological stimuli [21]. In the liver, TDO is thought to principally function as a homoeostatic enzyme necessary for the maintenance of physiological concentrations of L-Trp under normal conditions [21]. Recently, however, increased TDO has been detected in the brains of Alzheimer's disease patients [22] and in various types of cancer cells including breast cancer, ovarian carcinoma and gliomas [23,24]. Hence, similar to IDO1 (see the ‘Cancer’ section below), TDO has now been identified as playing a role in cancer progression and immune suppression [23,24]. Enhanced TDO activity, in concert with IDO1, has also recently been shown to play a key role in mediating endotoxin tolerance [25] (see the ‘Bacteria’ section below).

Although IDO1 and TDO catalyse the same dioxygenase reaction and contain similar haem active sites, there is only 10% sequence identity between the two enzymes [26], with IDO1 being a 42–45 kDa monomer and TDO a ∼190 kDa (eukaryotic) and ∼120 kDa (prokaryotic) homotetramer [27]. Moreover, TDO and IDO1 also differ significantly in several of their biochemical properties including substrate specificity and binding, and enzyme reactivity. For example, in contrast with TDO, which primarily binds L-Trp, 5-fluorotryptophan and 6-fluorotryptophan as substrates, IDO1 has a much higher affinity for D- and L-Trp and a broader substrate specificity that allows it to also accept a variety of indoleamines such as tryptamine, 5-hydroxytryptophan, 5-methyltryptophan, serotonin (5-hydroxytryptamine) and melatonin into its active site [2628]. Recent studies show that IDO1 and TDO also significantly differ with respect to the implications of their reaction with the ROS (reactive oxygen species) H2O2 [29,30] (see the ‘Hydrogen peroxide’ section below for details).

Identified in 2007, IDO2 is thought to have arisen from gene duplication ∼360 million years ago and shares 43% sequence identity with IDO1 [31,32]. Whereas IDO2 catalyses the oxidation of the indole ring of indoleamines, it appears to be less efficient when compared with IDO1 in recombinant protein assays and in vitro cellular models, although the optimal in vitro assay conditions and cofactors for IDO2 may not yet have been identified [3134].

Although there is a relative paucity of research concerning IDO2, current data suggest that the induction, expression and distribution of IDO2 are, in general, dissimilar to those of IDO1. For example, in human DC subtypes [35], IDO2 is primarily found in the cerebral cortex, kidney, epididymis and testis, and in the liver of mice, and, in these tissues, IDO2 is often present in cells that are distinct from those expressing IDO1 [31,33,36]. These data suggest differential and non-redundant functions of IDO2 in vivo that are only now beginning to be revealed. For instance, recent data support the suggestion that IDO2, but not IDO1, is linked to the pathogenesis of arthritis in mice by promoting the production of autoantibodies [37], whereas, unlike IDO1, IDO2 is not involved in the progression of inflammatory skin cancer in a mouse model [38]. Despite this discrepancy in the roles of IDO1 and IDO2, IDO2 can act similarly to IDO1 in promoting the generation of immunosuppressive Tregs (regulatory T-cells) and Ido1 and Ido2 gene-deficient mice both exhibit reduced skin-contact hypersensitivity responses [35,38]. In fact, a recent study found that IDO1-dependent generation of immunosuppressive Tregs is impaired in Ido2 gene-deficient mice [38], suggestive of a co-operative interaction of the two enzymes. IDO1−/− mice are characterised by the up-regulated expression of IDO2 within the epididymis, a tissue constitutively expressing IDO1 [36]. Importantly, similar to IDO1, IDO2 can also be inhibited by the commonly employed IDO inhibitor 1-MT (1-methyltryptophan) [32,39]. Moreover, macrophages from IDO1−/− mice reportedly exhibit impaired IDO2 function involving altered message splicing [38]. Therefore the numerous studies covered in the present review investigating the role of IDO1 employing 1-MT and IDO1−/− mice should be viewed in the light of these recent findings on IDO2 where a potential involvement of the more recently discovered IDO isoform cannot be excluded and needs to be considered. New insights clarifying the role of IDO2 including distinctions and potential interrelationships with IDO1 will no doubt be forthcoming with further studies employing IDO2−/− mice [38].

The present review focuses on IDO1 for which considerable advances in knowledge of the biochemistry of the enzyme and its roles in health and disease have been achieved, particularly over the last 20 years.

Tissue and cellular expression of IDO1

IDO1 is expressed in a wide range of tissues and cell types either constitutively or upon stimulation by relevant inflammatory and immune stimuli. Constitutive IDO1 protein expression under normal physiological conditions occurs primarily in mucosal tissues and subsets of immune cells. In rodents, this includes the caput of the epididymis, the prostate, adipose tissue, the placenta during pregnancy, in the uterus during the secretory phase, in lymphoid organs such as the thymus, the marginal zone of the spleen and certain lymph nodes (e.g. mesenteric), in the gastrointestinal tract (particularly the small intestine), the CNS (central nervous system), kidney, pancreas, lung and in various regions of the eye [4043]. Constitutive IDO1 expression in the adult mouse large intestine, small intestine and mesenteric lymph nodes is dependent on commensal micro-organisms and IFNγ, and is thought to play a homoeostatic role in regulating mucosal immunity in response to the local intestinal microbiota [44,45]. Human data indicate constitutive IDO1 expression in the placenta, eye and pancreas [4648], certain DC subsets [49] and eosinophils [50].

In many cell types, however, IDO1 is not expressed under normal physiological conditions. Indeed, IDO1 expression is robustly induced in response to inflammation and infection, with IFNs representing principal inducers (see the ‘Control of IDO1 gene expression’ section below). At the cellular level, elevated IDO1 is commonly expressed in myeloid cells including professional APCs (antigen-presenting cells) such as monocyte-derived macrophages and DC subsets [49,51], neutrophils [52,53] and eosinophils [50], as well as in non-haematopoietic cell types including endothelial cells [54], epithelial cells [55], fibroblasts [56] and SMCs (smooth muscle cells) [57]. Within IDO1-expressing cells, the enzyme is principally detected within the cytosolic and perinuclear regions of the cell. Recent data indicate that functional IDO1 protein also resides within microvesicles derived from human amniotic fluid stem cells [58]. IDO1 tissue and cellular expression in different disease states in humans and experimental animals is detailed in the ‘Biological roles of IDO1’ section below.

IDO1 BIOCHEMISTRY

In the light of the emerging clinical relevance and importance of IDO1 as a fundamental immune regulatory mechanism and response to infection (detailed in the ‘Biological roles of IDO1’ section below), it is essential to have a thorough understanding of the fundamental molecular and biochemical mechanisms controlling IDO1 expression and activity. The following sections review the current understanding of the biochemical aspects of IDO1.

Control of IDO1 gene expression

The IDO1 gene (also referred to as Indo or Ido1 in mice) is located on chromosome 8 (8p12-p11 in humans, 8 A2 in mice) and contains ten exons, which span approximately 15 kb [5961]. It encodes a monomeric IDO1 protein of 403 amino acids with a molecular mass of approximately 42–45 kDa. A number of genetic variants of IDO1 have been identified in humans, including SNPs (single nucleotide polymorphisms) and base pair deletions, some of which correlate with increased or reduced IDO1 protein expression and activity [6264]. The extent to which the expression of different genetic variants in individuals relates to changes in disease severity and outcome is worthy of investigation, with data thus far indicating an association of a specific IDO1 SNP with autoimmune systemic sclerosis [63] or an IDO1 genetic variant correlating with reduced risk of recurrent vulvovaginal candidiasis [64] in human patients. IDO1 transcription is also under epigenetic control involving histone deacetylation by HDAC (histone deacetylase), which removes acetyl groups from histones, allowing tighter binding to DNA, thereby inhibiting gene transcription. Accordingly, inhibition of histone deacetylation by treating murine bone-marrow-derived DCs with HDAC inhibitors enhances IDO1 gene expression [65].

The IDO1 promoter region contains two ISREs (IFN-stimulated response elements) and three GASs (IFNγ-activated sites), with which IRF1 (IFN regulatory factor 1) and STAT1 (signal transducer and activator of transcription 1) respectively interact [66] (Figure 2). The most potent stimulus for IDO1 gene expression in vitro and in vivo is IFNγ, which induces robust activation of the JAK (Janus kinase)/STAT and PKCδ (protein kinase Cδ) signalling pathways [67,68]. Accordingly, JAK, STAT1 and IRF1 are required for optimal IFNγ-mediated IDO1 gene transcription [66,6871]. STAT1 signalling leading to IDO1 expression can be influenced by PI3K (phosphoinositide 3-kinase), as PI3Kα-mediated STAT1 phosphorylation signals for maximal activation of STAT1 in IFNγ-treated human cord-blood-derived mesenchymal stem cells leading to robust IDO1 expression [72]. Expression of functional IDO1 in murine splenic CD8α+ DCs in response to IFNγ also requires the expression of the IRF8 transcription factor [73].

Multi-layered control of IDO1 expression and activity

Figure 2
Multi-layered control of IDO1 expression and activity

Cellular IDO1 expression and enzymatic activity are tightly regulated at various levels involving a multitude of molecular signalling pathways and events. Several different types of stimuli, including cytokines, TLR ligands, pathogens and reverse signalling via receptor–ligand interactions, induce IDO1 gene transcription in various cell-types. IFNγ [66], LPS or GpC-ODNs [92,107], viral infection [109] and reverse signalling though CD80/CD86 or GITRL [95,99] are typical examples of these classes of stimuli. (A) IFNγ principally signals through the JAK/STAT and IRF pathways and increasing evidence shows that IFNγ can also induce IDO1 expression via COX2-derived PGE2 [97,98]. (B) Furthermore, LPS, HIV and reverse signalling can each activate the non-canonical NF-κB pathway to induce IDO1 expression. (C) Negative regulators of IDO1 transcription include IL-4 [116], the transcriptional repressor BLIMP-1 (B-lymphocyte-induced maturation protein 1) that is activated by L-Trp deficiency [119], the BAR adaptor encoded by Bin1, which limits STAT1- and NF-κB-mediated IDO1 gene transcription [120], DAP12 (which must be down-regulated for IDO1’s activity) [121], as well as histone deacetylation [65]. (D) Once transcribed and translated into the IDO1 apoprotein, the haem prosthetic group synthesised de novo from the haem precursor δ-aminolaevulinic acid (E) must be incorporated to form the active haem-containing IDO1 holoenzyme [160]. Alteration of cellular haem availability by δ-ALAS and HO-1, enzymes that control haem synthesis and degradation respectively, can potentially influence IDO1 activity by influencing cellular levels of catalytically inactive IDO1 apoprotein compared with active IDO1 holoenzyme [160]. (F) Activation of IDO1 requires the one-electron reduction of ferric (Fe3+)-IDO1 to the active ferrous (Fe2+) state, facilitating L-Trp and O2 binding [136]. Potential intracellular reducing cofactors include superoxide anion radical (O2•−) [139141] and cytochrome b5 reductase (cyto-b5) [142]. (G) Active IDO1 catalyses the oxidative conversion of L-Trp into NFK, necessary for the formation of Kyn and other bioactive downstream metabolites. (H) IDO1-derived Kyn is itself capable of further enhancing IDO1 transcription by binding to and activating the AhR [25]. (I) IDO1 dioxygenase activity can be controlled at the post-translational level by endogenous molecules, such as NO [180,188] or exogenous small-molecule inhibitors, such as 1-L-MT [411]. (J) IDO1 can be post-translationally modified further by Src family kinases that phosphorylate IDO1 at Tyr115 and Tyr253 in response to TGFβ-mediated stimulation of the PI3K pathway or Kyn-mediated AhR activation (not shown) [25,113]. (K) Tyrosine-phosphorylated IDO1 recruits the protein tyrosine phosphatases SHP-1 and SHP-2, leading to the activation of the non-canonical NF-κB pathway, and the sustained production of TGFβ and expression of type I IFNs and other non-canonical NF-κB-dependent genes such as IDO1 [113]. In this manner, IDO1 acts as an intracellular signalling scaffold, independent of its enzymatic activity, to promote downstream signalling events and self-amplify its own gene expression. (L) IDO1 tyrosine phosphorylation can also signal for the binding of SOCS3 to the phosphorylated tyrosine residues and recruitment of ECS, resulting in the ubiquitination of IDO1 and resultant proteasomal degradation [213].

Figure 2
Multi-layered control of IDO1 expression and activity

Cellular IDO1 expression and enzymatic activity are tightly regulated at various levels involving a multitude of molecular signalling pathways and events. Several different types of stimuli, including cytokines, TLR ligands, pathogens and reverse signalling via receptor–ligand interactions, induce IDO1 gene transcription in various cell-types. IFNγ [66], LPS or GpC-ODNs [92,107], viral infection [109] and reverse signalling though CD80/CD86 or GITRL [95,99] are typical examples of these classes of stimuli. (A) IFNγ principally signals through the JAK/STAT and IRF pathways and increasing evidence shows that IFNγ can also induce IDO1 expression via COX2-derived PGE2 [97,98]. (B) Furthermore, LPS, HIV and reverse signalling can each activate the non-canonical NF-κB pathway to induce IDO1 expression. (C) Negative regulators of IDO1 transcription include IL-4 [116], the transcriptional repressor BLIMP-1 (B-lymphocyte-induced maturation protein 1) that is activated by L-Trp deficiency [119], the BAR adaptor encoded by Bin1, which limits STAT1- and NF-κB-mediated IDO1 gene transcription [120], DAP12 (which must be down-regulated for IDO1’s activity) [121], as well as histone deacetylation [65]. (D) Once transcribed and translated into the IDO1 apoprotein, the haem prosthetic group synthesised de novo from the haem precursor δ-aminolaevulinic acid (E) must be incorporated to form the active haem-containing IDO1 holoenzyme [160]. Alteration of cellular haem availability by δ-ALAS and HO-1, enzymes that control haem synthesis and degradation respectively, can potentially influence IDO1 activity by influencing cellular levels of catalytically inactive IDO1 apoprotein compared with active IDO1 holoenzyme [160]. (F) Activation of IDO1 requires the one-electron reduction of ferric (Fe3+)-IDO1 to the active ferrous (Fe2+) state, facilitating L-Trp and O2 binding [136]. Potential intracellular reducing cofactors include superoxide anion radical (O2•−) [139141] and cytochrome b5 reductase (cyto-b5) [142]. (G) Active IDO1 catalyses the oxidative conversion of L-Trp into NFK, necessary for the formation of Kyn and other bioactive downstream metabolites. (H) IDO1-derived Kyn is itself capable of further enhancing IDO1 transcription by binding to and activating the AhR [25]. (I) IDO1 dioxygenase activity can be controlled at the post-translational level by endogenous molecules, such as NO [180,188] or exogenous small-molecule inhibitors, such as 1-L-MT [411]. (J) IDO1 can be post-translationally modified further by Src family kinases that phosphorylate IDO1 at Tyr115 and Tyr253 in response to TGFβ-mediated stimulation of the PI3K pathway or Kyn-mediated AhR activation (not shown) [25,113]. (K) Tyrosine-phosphorylated IDO1 recruits the protein tyrosine phosphatases SHP-1 and SHP-2, leading to the activation of the non-canonical NF-κB pathway, and the sustained production of TGFβ and expression of type I IFNs and other non-canonical NF-κB-dependent genes such as IDO1 [113]. In this manner, IDO1 acts as an intracellular signalling scaffold, independent of its enzymatic activity, to promote downstream signalling events and self-amplify its own gene expression. (L) IDO1 tyrosine phosphorylation can also signal for the binding of SOCS3 to the phosphorylated tyrosine residues and recruitment of ECS, resulting in the ubiquitination of IDO1 and resultant proteasomal degradation [213].

Although it does not have the capacity to induce IDO1 alone, TNFα (tumour necrosis factor α) synergises with IFNγ and other cytokines or stimuli to maximally induce IDO1 activity in several cell types. The binding of TNFα to its receptor stimulates the NF-κB (nuclear factor κB) pathway, enhancing IRF1 transcription and STAT1 activation to up-regulate the expression of IFNγ receptors, thereby potentiating IFNγ signalling [66,7477]. Other stimuli that can induce IDO1 expression in various cell types, either singly or in combination with IFNγ, include type I IFNs (IFNα and IFNβ), IFNλ, IL (interleukin)-1 and IL-6, TLR (Toll-like receptor) 3, 4, 7, 8 and 9 ligands, thymosin α1, oestrogen, PGE2 (prostaglandin E2), muramyl tripeptide, α-galactosylceramide, glucocorticoids and haemoglobin [74,75,7896] and certain infectious pathogens (discussed in detail in the ‘IDO1 and host response to microbial infection’ section below).

Increasing evidence identifies an important regulatory link between COX2 (cyclo-oxygenase 2)-derived PGE2 and IDO1. For example, early studies in mouse lung slices treated with IFNγ reported that COX inhibitors (aspirin and indomethacin) inhibited IDO1 induction [97]. Subsequent studies indicate an important role for COX2-derived PGE2 in elevating IDO1 in different tumour cells (e.g. breast cancer [98]). Activation of human monocyte-derived DCs with TNFα or LPS (lipopolysaccharide) synergise with PGE2 to induce IDO1 expression in an apparent two-step process with PGE2-dependent adenylate cyclase activation inducing IDO1 mRNA expression and TNFα receptor/TLR signalling required for enzyme activation [74].

In DCs, IDO1 expression is also up-regulated upon ligation of CD80/86 with CTLA-4 (cytotoxic T-lymphocyte-associated protein 4) or interaction of the GITRL (glucocorticoid-induced TNF-receptor-related protein) receptor with GITR (glucocorticoid-induced TNF receptor) expressed on Tregs, as well as by engagement of the CD200 receptor [95,99101]. Induction of IDO1 by CTLA-4 is achieved via both IFNγ-dependent (as shown by the lack of Kyn produced by murine IFNγ−/− DCs and STAT1−/− DCs in vitro) [102] and -independent mechanisms (as shown by similar IDO1 expression levels in wild-type and IFNγ−/− mice) [103]. Further emphasising the complexity of the signalling pathways regulating IDO1 expression in different DC subsets, IDO1 induction in splenic CD19+ DCs in response to CpG-ODNs (CpG oligodeoxynucleotides) or CTLA-4 occurs independently of IFNγ and instead requires IFNα-dependent STAT1 activation [104,105]. Recent data indicate that induction of IDO1 in myeloid DCs exposed to DNA (to mimic the cytosolic detection of pathogen-derived DNA) requires a proximal signalling pathway involving the STING (stimulator of IFN genes) adaptor and resultant production of type I IFNs [106].

Activation of the NF-κB pathway is important for the induction of IDO1 in response to stimuli other than IFNγ (Figure 2), including LPS [92], GpC-ODNs [107], viral infection [94,108,109] and CD80/CD86 ligation [101]. For example, induction of IDO1 by LPS and virus infection involves the activation of both NF-κB and p38 MAPK (mitogen-activated protein kinase) [92,94]. NF-κB-dependent induction of IRF3- and TGFβ (transforming growth factor β)-dependent induction of IDO1 in response to TLR9-activation by GpC-ODNs involves novel roles for TRIF [TIR (Toll/interleukin-1 receptor) domain-containing adaptor protein inducing interferon β] and TRAF6 (TNF-receptor-associated factor 6) proteins [107]. With respect to CD80/CD86 signalling, Koorella et al. [101] showed recently that activation of CD80/CD86 in human monocyte-derived DCs with CD28-Ig induces NF-κB-dependent IDO1 expression by activating PI3K/Akt-dependent signalling. Interestingly, NF-κB-dependent IDO1 activation in response to CD80/CD86 ligation also involves Notch and FOXO3 (forkhead box O3) signalling pathways, which differentially regulate PI3K signalling. Koorella et al. [101], studying CD28-Ig-treated human DCs, reported that Notch1 activates the PI3K signalling necessary for IDO1 induction by inactivating PTEN (phosphatase and tensin homologue deleted on chromosome 10) through PIN (peptidyl-prolyl isomerase)-dependent casein kinase activation. Fallarino et al. [110] studying splenic CD8α+ DCs isolated from diabetic mice reported that ligation of CD80/CD86 with CTLA4-Ig induced IDO1 expression in a FOXO3a-dependent manner involving the down-regulation of PI3K signalling. Expression of IDO1 by tumour-infiltrating DCs is also dependent on the activation of the FOXO3 transcription factor [111].

With respect to NF-κB, the non-canonical rather than the canonical pathway plays an important role in IDO1 expression. For instance, reverse signalling through GITRL on murine pDCs (plasmacytoid DCs) stimulates IDO1 expression via the non-canonical pathway [95], as does CD40 ligation and HIV infection of human DCs [109,112] and activation of TLR9 with GpC-ODNs [107]. A recent study identified that stimulating murine pDCs with TGFβ results in the binding of heterodimers of the NF-κB subunits p52 and RelB to the mouse IDO1 promoter [113], a binding site also present in the promoter of human IDO1 [99].

Interestingly, IDO1 enzyme activity can positively influence IDO1 gene transcription through the binding of Kyn to the AhR [25], a cytosolic ligand-activated transcription factor, which has a myriad of functions, depending on the cell type, ligand, duration and location of binding [113,114]. Whether other AhR ligands, such as the Kyn metabolite kynurenic acid [115], can similarly signal for the induction of IDO1 gene transcription or, indeed, has other modulatory effects, requires further studies.

Negative regulators of IDO1 gene expression include IL-4, IL-6 and TGFβ [116118], the transcription repressor BLIMP-1, which is activated in response to L-Trp deficiency [119], and adapter molecules such as DAP12 [(DNAX activation protein of 12 kDa) which is typically bound to activating receptors of myeloid cells and must be down-regulated for functional IDO1 to become apparent] and the Bin1-encoded BAR (Bin/amphiphysin/Rvs) adaptor that blocks STAT1- and NF-κB-mediated IDO1 transcription [120,121] (Figure 2). Many of these negative regulators affect IDO1 gene transcription indirectly, e.g., via modulation of IFN production or the down-regulation of IFNγ receptors. IDO1 expression and activity is also suppressed under hypoxic conditions, with the defect in expression linked to impaired JAK-STAT signalling [122124]. This has important implications for the effectiveness of IDO1’s immune control and host defence roles in in vivo conditions where local tissue O2 tension is reduced (e.g. during infection or within fast-growing tumours).

It should be noted that the actions of many of these IDO1 modulatory stimuli are dependent on the presence of other stimuli for IDO1 induction, are highly cell-specific and may have opposing effects under different circumstances. For example, constitutive IDO1 expression in CD40L-matured CD123+ human DCs is inhibited by IFNγ, while the addition of IL-10 prevents the down-regulation of constitutively expressed IDO1 in these cells [49]. In contrast, IL-10 inhibits LPS-stimulated up-regulation of IDO1 in murine bone marrow-derived DCs [125]. Furthermore, while TGFβ inhibits IFNγ-induced IDO1 mRNA expression in cultured fibroblasts, TGFβ induces IDO1 expression in pDCs [113,117]. Finally, whereas IL-6 inhibits IDO1 in CD8α+ murine DCs, the inflammatory cytokine acts in concert with TNFα to induce IDO1 in virus-infected human monocyte-derived macrophages [94].

The above highlights the complex array of stimuli and cell signalling pathways governing IDO1 gene transcription and expression of relevance to the repression and constitutive or inducible expression of the enzyme in a cell-type-specific manner. In addition to further understanding the fundamental biochemical mechanisms by which different stimuli modulate IDO1 gene expression in different cell types, a key focus of future research is defining the precise molecular events underlying the induction or repression of IDO1 expression during disease conditions where IDO1 is an important determinant of the pathogenesis and severity of disease, e.g. aberrant induction of IDO1 expression in cancer cells or tumour-infiltrating APCs that underlies tumour immune escape [120,126] (see the ‘Cancer’ section below) or impaired IDO1 expression during murine Type 1 diabetes that facilitates the autoimmune destruction of pancreatic β-cells [110,127] (see the ‘Autoimmune diseases’ section below).

IDO1 protein structure and catalytic reaction mechanism

IDO1 gene transcription and translation yields the production of the active IDO1 enzyme containing the haem prosthetic group housed within its active site. IDO1 belongs to a unique family of mammalian haem dioxygenase enzymes. Owing to the increased interest in IDO1, a considerable amount of new information on the structure of IDO1 and its catalytic reaction mechanism has been assembled over the last 10 years.

The first crystal structure of the human IDO1 protein was solved in 2006 by Sugimoto et al. [128]. This structure of IDO1, which had the competitive inhibitor PI (4-phenylimidazole) co-ordinated to the haem iron as a distal ligand, revealed that the enzyme contains two α-helical domains of large and small size. The helices of the large domain create a cavity for the active-site haem, whereas the small domain covers the top of this cavity (Figure 3A). The proximal side of the haem is solely occupied by side chains from the large domain, whereas the distal haem pocket consists of residues from the small domain, the large domain and the loop connecting the two domains.

X-ray crystal structure of human IDO1

Figure 3
X-ray crystal structure of human IDO1

(A) View of the complete protein, showing large (red to green) and small (blue to violet) α-helical domains. Also shown is the haem prosthetic group complexed to the ligand inhibitor 4-phenylimidazole. (B) Detailed view of the haem active site, showing nearby distal and proximal amino acid residues. The 4-phenylimidazole ligand is omitted from this view. The crystal structure shown is derived from [128].

Figure 3
X-ray crystal structure of human IDO1

(A) View of the complete protein, showing large (red to green) and small (blue to violet) α-helical domains. Also shown is the haem prosthetic group complexed to the ligand inhibitor 4-phenylimidazole. (B) Detailed view of the haem active site, showing nearby distal and proximal amino acid residues. The 4-phenylimidazole ligand is omitted from this view. The crystal structure shown is derived from [128].

More recently, a crystal structure of human IDO1 was solved with the thiazolotriazole compound Amg-1 co-ordinated to the haem in the distal pocket with an apparent induced fit to the protein [129]. This study highlighted the possible importance of two distal binding pockets for structure-based design of IDO1 inhibitors: pocket A, which is occupied by PI in the previous structure reported by Sugimoto et al. [128], and pocket B, which is lined by the amino acids Phe226 and Arg231, both of which were previously identified by Sugimoto et al. [128] as crucial for substrate binding (see below). In the structure reported by Tojo et al. [129], Amg-1 spans both pocket A and pocket B (the p-tolyl ring occupying the former and the methylenedioxyphenyl ring the latter), whereas for the IDO1–PI structure solved by Sugimoto et al. [128], pocket B is occupied by two molecules of the buffer Ches [2-(N-cyclohexylamino)ethanesulfonic acid]. Tojo et al. [129] hypothesised that π–π interactions with Phe226 and electrostatic interactions with Arg231 in pocket B are crucial for substrate/inhibitor binding in IDO1. Through a series of structure–activity relationship studies, they designed an imidazothiazole derivative with an urea linker to a 4-cyanophenyl group, a compound that apparently spans both pockets A and B, while co-ordinating with the haem iron, exhibiting an IC50 of 77 nM [129]. Intriguingly, slightly higher inhibitory potencies were previously reported for derivatives of 4-amino-1,2,5-oxadiazole-3-carboximidamide (IC50 ∼60 nM) [130], compounds that appear too small to span both pockets A and B. As such, more work is warranted in the challenging area of structure-based design and discovery of small-molecule IDO1 inhibitors.

Earlier site-directed mutagenesis studies identified His346 and Asp274 as being essential for IDO1 haem binding [131]. The crystal structures substantiate these findings by revealing His346 as the proximal haem ligand and Asp274 being linked to the proximal histidine residue via a hydrogen-bonding network within the proximal haem pocket (Figure 3B). Another highly conserved histidine residue, His303, had been hypothesised as a sixth (distal) haem ligand on the basis of spectroscopic evidence [132,133] and by analogy to globin proteins, which utilise a distal histidine residue to stabilise the bound O2 through H-bonding. However, the crystal structures show that His303 is not positioned near to the distal haem pocket. The nearest polar residue to the haem iron within the distal pocket is Ser167 (Figure 3B), but this residue appears not to be important for O2 or L-Trp binding, or for catalytic activity [128,134]. Further site-directed mutagenesis studies have, however, identified Phe226, Phe227, Arg231 and Ser263 as being critical for maintaining the precise positioning of the substrate within the distal active-site haem pocket, which is a prerequisite for L-Trp to efficiently react with the iron-bound dioxygen upon enzyme activation [128]. This strict shape complementarity requirement between the substrate and the distal pocket of the protein environment was highlighted further in our studies into the mechanisms by which the selenazole drug ebselen [135] and H2O2 inhibit IDO1 dioxygenase activity [30]. CD and resonance Raman spectroscopy analyses indicated that both of these agents inhibit dioxygenase activity by inducing conformational changes to the IDO1 protein, including a widening of the distal haem pocket, which results in the misaligned or non-productive binding of the substrate within the active site [30,135].

Activation of IDO1’s dioxygenase activity requires the single electron reduction of the haem from the resting ferric (FeIII) to the active ferrous (FeII) form, which facilitates the binding of O2 and L-Trp to form an active ternary complex [136]. On the basis of equilibrium binding studies, Sono et al. [136] proposed that the ferrous enzyme binds first with L-Trp and then with O2 [136]. However, a more recent study that combined kinetic and electrochemical measurements showed that, at low concentrations of L-Trp, O2 binds first followed by L-Trp binding, but, at higher L-Trp concentrations, the binding order is reversed [137]. Resonance Raman spectroscopic data for the FeII-O2 complex of human recombinant IDO1 indicated that it is best described as an FeIII-O2•− (i.e. ferric-superoxide) complex [138]. Consistent with this, the superoxide anion radical (O2•−) is capable of reacting directly with FeIII–IDO1 in the presence of L-Trp to produce NFK, albeit inefficiently with respect to supporting a sustained level of catalytic activity [139142] (see the ‘IDO1 enzyme cofactors’ section below).

Another noteworthy aspect of IDO1’s dioxygenase activity is that it exhibits substrate inhibition, i.e. a decrease in the rate of enzyme activity at higher substrate concentrations. First recognised for the rabbit enzyme some decades ago [136], it has been studied in detail more recently for human IDO1 [137,143]. Initially, it was proposed that the substrate inhibition at higher L-Trp concentrations involved the direct binding of the substrate to the ferric form of the enzyme, which interfered with the reduction of the haem iron to the active ferrous state [136]. This scenario now seems unlikely because the dissociation constant of L-Trp for the ferric IDO1 (Kd=0.9 mM) is significantly higher than the substrate inhibition constant (KSI=0.17 mM), and also because the reduction potential of the L-Trp-bound ferric form is higher than that of the substrate-free enzyme [143]. Instead, IDO1’s substrate inhibition is attributed to a second allosteric effector-binding site for indole derivatives [143,144]. Evidence for a second substrate-binding site was also observed in our previous studies measuring L-Trp binding to human recombinant IDO1 [30,135]. In the case of L-Trp, binding of the substrate to the allosteric site causes inhibition, whereas other indole derivatives can activate the enzyme [143]. More recently, another hypothesis was put forward for the mechanism of substrate inhibition that does not rely on a second binding site [137]. Using a combination of kinetic and reduction potential measurements, it was reported that the haem iron reduction potential increases in the presence of L-Trp, thus making the FeIII-O2•− species thermodynamically less favourable relative to the isoelectronic FeII-O2 species. Since the superoxide species was strongly implicated as the first species involved in the catalytic reaction mechanism, the hindrance of the formation of this species in the presence of the substrate was postulated to be the underlying reason for the substrate inhibition [137].

A great deal of progress has been made in the last few years in determining the dioxygenase reaction mechanism affording conversion of L-Trp into NFK and some excellent, more detailed, reviews on this topic have been published [145,146]. We first provide a short historical perspective before addressing the main aspects of the IDO1 catalytic mechanism as they currently stand.

The earliest proposals of the IDO1 dioxygenase reaction mechanism involved a base-catalysed abstraction of the proton from the pyrrole ring nitrogen of the substrate followed by the formation of 3-indolenylperoxo and dioxetane intermediates [147]. The main reasons for this base-catalysed proposal were the presumed presence of a distal histidine residue (i.e. His303) that could act as the base, as well as the compelling observations that 1-methyl-Trp, β-(3-benzofuranyl)-DL-alanine (the furan analogue of tryptophan with O instead of N-H in the 1 position), and β-[3-benzo(b)thienyl]-DL-alanine (the thiophene analogue of tryptophan with S in the 1 position) act as competitive inhibitors [148]. However, since site-directed mutagenesis studies showed that His303 has no effect on the rate of the reaction, we proposed the possibility that the haem-bound dioxygen itself acted as the base that accepted the proton from the substrate in the first step of the reaction [149].

A complete change of perspective for the mechanism of oxidation of L-Trp by IDO1 was brought about in 2009 when rapid-mixing resonance Raman spectroscopic evidence for the formation of a ferryl-oxo (FeIV=O), compound II-type intermediate was first presented [138], and then later confirmed [150]. The detection of the FeIV=O species during the dioxygenase reaction motivated several theoretical and experimental studies [138,145,146,151155] that have led to the presently proposed mechanism shown in Figure 4. The reaction begins with the formation of the ternary complex (FeIII-superoxo-Trp). Radical addition of the ferric iron-bound superoxide to the C2=C3 bond of L-Trp leads to the formation of the ferryl-oxo and indole 2,3-epoxide (L-Trp-epoxide) species via a 2-indolenylperoxo radical transition state. Direct experimental evidence for the formation of the ferryl-oxo species was provided by resonance Raman data as stated above [138], whereas MS data provided indirect experimental evidence for the formation of the L-Trp-epoxide species [151]. The next steps, which are based purely on theoretical calculations [153], are postulated to involve proton transfer from the NH3+ group of L-Trp to the epoxide oxygen, triggering ring opening, followed by nucleophilic attack of the FeIV=O oxygen to C2 of L-Trp. The final step is the concerted C2–C3 bond cleavage and back proton transfer from the C3 oxygen to the NH3+ group of L-Trp. Further experiments are required to confirm the formation of the L-Trp-epoxide intermediate as well as the ensuing steps of this novel dioxygenase mechanism.

Hypothesised reaction mechanism of the oxidation of L-Trp to NFK catalysed by human IDO1

Figure 4
Hypothesised reaction mechanism of the oxidation of L-Trp to NFK catalysed by human IDO1

The first step in this proposed mechanism is radical addition, rather than electrophilic addition, which has also been proposed [146]. Direct experimental evidence for the formation of the ferryl-oxo species was provided by resonance Raman data [138,150], whereas MS data provided indirect experimental evidence for the formation of the Trp-epoxy intermediate [151]. The ensuing steps are based purely on theoretical calculations [153]. The reaction scheme is adapted from reference [153].

Figure 4
Hypothesised reaction mechanism of the oxidation of L-Trp to NFK catalysed by human IDO1

The first step in this proposed mechanism is radical addition, rather than electrophilic addition, which has also been proposed [146]. Direct experimental evidence for the formation of the ferryl-oxo species was provided by resonance Raman data [138,150], whereas MS data provided indirect experimental evidence for the formation of the Trp-epoxy intermediate [151]. The ensuing steps are based purely on theoretical calculations [153]. The reaction scheme is adapted from reference [153].

A further aspect of the catalytic mechanism of IDO1 that warrants further investigation is the relationship between substrate structure and reactivity in IDO1. A variety of L-Trp derivatives have been tested, including derivatives with methyl, hydroxy and fluoro group substitutions in the 1, 4, 5, 6 and 7 positions, as well as the thiophene-Trp and furan-Trp analogues mentioned above [28,156]. These compounds display a range of values for kcat (a measure of enzyme turnover rate). For example, the thiophene-Trp and furan-Trp analogues appear to be unreactive (kcat=0), but act as potent competitive inhibitors and can therefore bind to the active site [148]. 5-Methyl-DL-Trp (kcat=3.8 s1) is more reactive than L-Trp (kcat=1.4 s1), whereas 5-fluoro-DL-Trp (kcat=0.76 s1), 5-hydroxy-DL-Trp (kcat=0.025 s1) and 1-methyl-L-Trp (kcat=0.027 s1) are less reactive than L-Trp as IDO1 substrates [27,28,156]. These differences in reactivity are yet to be reconciled with the proposed catalytic mechanism for IDO1. A comprehensive understanding of IDO1’s structure and catalytic mechanism will not only provide further fundamental insights into haem dioxygenases, but also reveal new and important information useful for the design of selective and potent small-molecule IDO1-inhibitory drugs, which are currently sought after as novel anti-cancer drugs (see the ‘Cancer’ section below).

Post-translational control of IDO1 activity

Considering that uncontrolled IDO1 activity could be harmful to the host through local depletion of an essential and low-abundance amino acid, L-Trp, and overproduction of bioactive and potentially cytotoxic Kyn pathway metabolites, it is logical that the enzyme's activity is tightly regulated at various levels in addition to the transcriptional control of gene expression. Indeed, growing evidence indicates that IDO1 is subject to significant post-translational control. For example, human macrophages can express multiple IDO1 protein species that display different molecular masses and isoelectric points [49], findings consistent with post-translational modifications that modify IDO1 molecular mass and/or protein charge. Also, certain cell populations have been identified to constitutively express IDO1 protein with reduced enzymatic activity [49,157,158]. For example, human DC subsets express inactive IDO1 protein that is activated upon stimulation with IFNγ or the binding of CD80/86 molecules on the cell surface [49]. Similarly, mouse splenic CD8α+ DCs avidly catabolise L-Trp when exposed to IFNγ, whereas CD8α DCs do not, despite both cell subsets expressing similar levels of IDO1 protein [157]. Although the precise biochemical mechanisms responsible for this difference are not known, the transmembrane signalling adaptor molecule DAP12, capable of acting as a co-activator of ITAMs (immunoreceptor tyrosine-based activation motifs), has been identified as a putative determinant controlling the post-translational modifications affording maximal IDO1 enzyme activity [73]. Moreover, CGD (chronic granulomatous disease) mice subject to fungal infection have been reported to express an inactive IDO1 protein within the lung, providing in vivo support for post-translational control of IDO1 [159].

Various modes of post-translational control have been described for IDO1 and include the regulation of enzyme activity at the level of availability of the haem prosthetic group or reducing cofactor(s), control by reduction and oxidation (redox) reactions mediated by different ROS or RNS (reactive nitrogen species), and protein–protein interactions, tyrosine phosphorylation and proteasome degradation.

Haem supply

Once translated, the haem-free IDO1 apoprotein is available for the insertion of the haem prosthetic group, which is essential for the formation of the catalytically active IDO1 holoenzyme (Figure 2). Our previous work in primary cultures of monocyte-derived human macrophages indicates that IDO1 activity is regulated by modulating the proportion of the enzyme present as haem-deficient apoprotein compared with haem-containing holoenzyme [160]. Cellular haem available for incorporation into proteins depends on the relative rates of haem biosynthesis compared with haem degradation. Haem synthesis is governed by the initial and rate-limiting enzyme of the pathway, δ-ALAS (δ-aminolaevulinic acid synthase), whereas degradation is controlled by HO-1 (haem oxygenase 1), which catalyses the oxidation of haem into carbon monoxide (CO), iron and biliverdin (Figure 2). Our studies suggest that the formation of the active IDO1 holoenzyme requires newly synthesised haem. Thus, supplementation of IFNγ-stimulated human macrophages with the haem precursor δ-aminolaevulinic acid enhances cellular IDO1 activity, whereas inhibition of haem biosynthesis with succinyl acetone inhibits both the cellular incorporation of haem into IDO1 and enzyme activity, without altering IDO1 protein expression levels [160].

IDO1 activity can also be modulated by enzymes capable of competing with IDO1 for intracellular haem. For example, increased expression of HO-1 signals for reduced IDO1 activity in cancer cells [161]. We have found a similar inhibitory role of HO-1 for IDO1 activity in human monocyte-derived DCs (A.W.S. Yeung, P. Anand, H. Cai, N.J.C. King and S.R. Thomas, unpublished work). Although these findings are consistent with HO-1-catalysed reduction in cellular haem levels leading to IDO1 inhibition, strict biochemical evidence for this is yet to be provided. Recent data suggest that co-expression of IDO2 with IDO1 inhibits the activity of the latter isoform through competitive haem binding [162].

Other studies, however, have found that HO-1 activity can correlate with enhanced IDO1 enzyme expression. Thus, incubation of murine DCs with the HO-1 inhibitor zinc protoporphyrin or the HO-1 inducer haemin inhibited or augmented IDO1 expression, respectively [163]. Interestingly, others have reported that zinc protoporphyrin can directly and irreversibly inhibit IDO1 enzyme activity by competing with haem (i.e. iron protoporphyrin) for the binding to newly synthesised IDO1 apoprotein [164]. Therefore, whether the actions of zinc protoporphyrin on IDO1 reflect its roles as an HO-1 inhibitor or direct IDO1 inhibitor will depend on the extent to which cells express active IDO1 holoenzyme compared with synthesising new IDO1 apoenzyme at the time of zinc protoporphyrin treatment. In murine macrophages stimulated with IFNγ and LPS, addition of the Kyn pathway metabolite 3-HAA is reported to elevate HO-1, which, via the production of CO, reduces the inducible isoform of NOS (nitric oxide synthase) or NOS2 expression thereby abrogating NO-dependent IDO1 inhibition (see the ‘Nitric oxide’ section below) and hence elevating IDO1 expression and activity [165]. Interestingly, the HO-1 inducer haemoglobin induces NF-κB-dependent IDO1 expression in murine bone marrow-derived DCs involving PI3K/Akt signalling and PKC-dependent production of ROS [96]. The above highlights the potential for regulatory networks of cross-talk between the IDO1 and HO-1 pathways, the extent and nature of which may vary between different cell types, species and/or activation status.

IDO1 enzyme cofactors

As mentioned above, dioxygenase activation requires that the IDO1 active-site haem undergoes a one-electron reduction from the inactive ferric (FeIII) state to the active ferrous (FeII) state. Early studies with purified enzyme identified several physiological reductants as putative cofactors for IDO1, including reduced FMN (flavin mononucleotide), tetrahydrobiopterin and O2•− [139141,166,167]. With respect to flavins or tetrahydrobiopterin, evidence for roles for these intracellular reductants in maintaining IDO1 activity have not been forthcoming. A role for flavins is unlikely given that we have shown that treatment of IFNγ-stimulated human macrophages with the broad-spectrum flavoprotein inhibitor DPI (diphenyleneiodonium) has no effect on cellular IDO1 activity [168]. Interestingly, whereas DPI did not inhibit IDO1, it did selectively inhibit the production of 3-HAA in IFNγ-stimulated human macrophages, consistent with the Kyn 3-mono-oxygenase enzyme being a flavin-dependent enzyme [168].

Traditionally, O2•− has been considered a principal intracellular reducing cofactor for IDO1. Thus, initial pioneering studies by Hayaishi and co-workers with purified rabbit enzyme reported that IDO1 employs O2•− and O2 for activity in a manner sensitive to inhibition by SOD (superoxide dismutase) [139141]. Consistent with this, O2•− reduced the ferric-haem of purified rabbit IDO1 into ferrous-IDO1 such that L-Trp and O2 bind to form the active ternary complex [136]. Moreover, 18O2-labelling studies showed that O2•− is incorporated into L-Trp during the catalytic turnover of the purified rabbit IDO1 enzyme [140]. Together, these results identified O2•− as a reducing cofactor and substrate for the purified IDO1 enzyme.

Evidence of a role for O2•− as a cofactor for cellular IDO1 initially came from studies employing enterocytes, which constitutively express active IDO1 enzyme [169]. Thus, addition of the xanthine oxidase substrate inosine to rabbit enterocytes increased IDO1 activity in a manner inhibited by the addition of the xanthine oxidase inhibitor allopurinol [169]. Notably, however, the actions of inosine were dependent on the co-addition of the redox-active dye methylene blue to the enterocytes, confounding conclusions on the role of xanthine oxidase-derived O2•− as a relevant cofactor for cellular IDO1. More recent data in enterocytes confirmed that, whereas the addition of inosine in the presence of methylene blue increases IDO1 activity, addition of inosine alone to rabbit enterocytes does not activate IDO1 [142], a finding inconsistent with a major role for xanthine oxidase-derived O2•− as an essential intracellular IDO1 cofactor.

Further evidence for the reaction of O2•− with IDO1 is derived from early studies reporting that lung IDO1 levels induced by LPS treatment of rabbits, but not rats and mice [170,171], reached sufficiently high levels to afford protection against oxidative-stress-induced pulmonary damage [170]. This species-specificity correlated with pulmonary IDO1 being induced to a 170-fold greater degree in rabbits than in rats and mice [170,171]. Addition of a SOD inhibitor in order to increase O2•− levels enhanced IDO1 activity 4-fold in lung slices of rabbits, but not in rats and mice. Furthermore, elevating tissue O2•− levels by adding paraquat or inducing hyperoxia enhanced IDO1 activity in rabbit lungs 5-fold in the presence of a SOD inhibitor. These results suggest that lung IDO1 protein may be increased in vivo to a level sufficiently high to react with O2•− and thereby exhibit antioxidant activity [170].

A recent study in a mouse model of CGD provided support for a role for NOX (NADPH oxidase)-derived O2•− as a cellular cofactor in mouse pulmonary neutrophils. Thus, neutrophils isolated from Aspergillus-infected lungs of wild-type mice expressed active IDO1 (measured as cellular Kyn production) after overnight IFNγ treatment. In contrast, neutrophils from the Aspergillus-infected lungs of mice deficient in the NOX regulatory subunit p47phox gene did not form Kyn upon IFNγ treatment, despite expressing similar amounts of IDO1 protein to that of wild-type mouse neutrophils [159]. Similarly, increased Kyn was detected within the lungs of wild-type mice, but not p47phox gene-deficient mice, in response to pulmonary Aspergillus infection, despite elevated expression of IDO1 protein within infected lungs of both mouse genotypes [159]. These findings support that O2•− produced by NOX in mouse lung neutrophils is necessary to support IDO1 activity within these cells. However, a recent study highlights the controversy that exists in this area [172]. In contrast with Romani et al. [159], this recent communication reported that p47phox gene-deficient mice, as well as mice with impaired NOX2 activity, do not exhibit reduced IDO1 activity in response to Aspergillus infection (indeed, IDO1 activity is reportedly elevated in these animals) [172]. Furthermore, human blood leucocytes isolated from human CGD patients have robust IDO1 activity upon IFNγ treatment, findings not in support of NOX-derived O2•− as an essential cofactor of IDO1 in circulating human neutrophils [52]. The reasons for these apparent discrepancies regarding the relationship of O2•− to IDO1 enzyme activity, particularly in the context of infected lungs, warrant further clarification.

At the biochemical level, several studies question the ability of O2•− to maintain maximal steady-state activity of IDO1 and therefore to act as an efficient intracellular reducing cofactor. The conventional and commonly employed in vitro IDO1 enzyme assay requires the addition of the redox-cycling dye methylene blue and ascorbate as electron donors and not O2•− [144]. Accordingly, the ascorbate/methylene blue cofactor system is insensitive to scavenging of O2•− with SOD [144]. Kinetic considerations also argue against a role for O2•− as an intracellular IDO1-reducing agent; SOD reacts with O2•− with a rate constant of 2×109 M−1·s−1, whereas the corresponding rate constant for IDO1 is (2–8)×106 M−1·s−1 [173], findings that suggest that it is unlikely that tissue and cellular levels of O2•− reach the levels required for ferric IDO1 to outcompete SOD for O2•−. Consistent with this conclusion, studies show that O2•− only modestly activates purified rabbit or recombinant human IDO1 and is incapable of maintaining maximal steady-state enzyme turnover [142,144]. Also, reducing intracellular O2•− levels by studying human blood leucocytes isolated from human CGD patients or addition of O2•− scavengers to various cell types, including isolated rabbit enterocytes and IFNγ-stimulated human macrophages or endothelial cells, has no material effect on intracellular IDO1 activity [52,142].

Instead of O2•−, emerging data indicates a key role for cytochrome b5 reductase, a protein expressed in the outer membrane of the endoplasmic reticulum, as an important intracellular reductant for IDO1. Indirect evidence initially came from Vottero et al. [174], employing a tryptophan auxotroph yeast strain transduced with human IDO1, who reported that a deficiency in cytochrome b5 led to increased cell growth, a finding which was interpreted as a requirement for cytochrome b5 for IDO1 activity, depletion of L-Trp and resultant IDO1-dependent growth inhibition. Subsequently, Maghzal et al. [142] reported that cytochrome b5 reductase, in the presence of NADPH and cytochrome P450 reductase, efficiently reduced ferric-IDO1 and maintained the steady-state turnover of recombinant human IDO1 in vitro, in a reaction largely independent of O2•− [142]. Subsequent studies with recombinant IDO1 report that NADPH cytochrome P450 reductase alone also supports in vitro enzyme activity in a manner supported further by cytochrome b5, although with altered enzyme kinetics [175]. Gene silencing of the cytochrome b5 gene in an IDO1-transfected mammalian cell line also inhibited intracellular IDO1 enzyme activity, providing further evidence for cytochrome b5 as an intracellular reductant of IDO1 [142]. As a dioxygenase, IDO1 naturally requires O2 as a substrate. Accordingly, several studies report decreased IDO1 enzyme activity in cells cultured under reduced O2 tension or hypoxia [123,124].

Despite the recent advances identifying cytochromes as cellular IDO1 reductants, the true nature of the reducing cofactor(s) may differ depending on various factors including cell type and activation status, tissue environment, animal species and/or disease state. As such, more detailed systematic studies into the reducing cofactor(s) for IDO1 are clearly warranted, as this will provide important insights into the development of novel strategies aimed at modulating IDO1 activity in vivo. Moreover, because the efficacy and mode of action of certain IDO1 inhibitors may be influenced by the nature of the reducing conditions employed [175], use of relevant physiological reducing conditions (rather than the artificial ascorbate/Methylene Blue reducing system) for supporting recombinant IDO1 activity may be important in the discovery of selective and in vivo potent IDO1-inhibitory drugs.

Redox control of IDO1

Redox signalling

IDO1 is expressed in inflammatory tissue environments characterised by an increase in the local production of ROS and RNS. Although ROS and RNS have been traditionally considered to be cytotoxic tissue-damaging agents, it is now clear that they are also important transducers of cell signalling [176]. Thus, ROS and RNS selectively react with metal centres or amino acids (namely cysteine, methionine and tyrosine residues) of diverse proteins (e.g. protein phosphatases and kinases, transcription factors, ion channels and metabolic enzymes) to modulate their activity. Various ROS and RNS are produced by cells contained within the local inflammatory environment, which are produced from various enzymes including NOX and NOS (Figure 5). Several NOX isoforms exist including NOX 1, 2, 4 and 5 [176]. The primary reaction catalysed by NOX enzymes is the conversion of O2 into O2•−, which dismutates into H2O2 via the action of SOD. The H2O2 produced is either removed by several antioxidant enzymes (e.g. catalase, glutathione peroxidase or peroxidredoxins) or capable of oxidising protein cysteine residues (to form sulfenic, sulfinic or sulfonic acid forms of the amino acid) or oxidising reduced glutathione (GSH) to induce S-glutathionylation of protein cysteine residues (i.e. addition of glutathione to protein cysteine residues) [176]. The ability of H2O2 to selectively oxidise protein cysteine residues is recognised as a fundamental mode of redox signalling, controlling the activities of a wide variety of proteins or enzymes [176]. H2O2 is also an essential cofactor for haem peroxidases (e.g. myeloperoxidase), which, depending on the substrate milieu, catalyse a variety of (patho)-physiologically relevant oxidative reactions, e.g. in the presence of H2O2, haem peroxidases can catalytically consume NO [177] or convert the NO oxidation product nitrite (NO2) into the nitrogen dioxide radical (NO2), which nitrates tyrosine residues on proteins to form 3-nitrotyrosine, a common marker of oxidative stress in inflammatory tissues [178,179]. NOS enzymes oxidise L-Arg (L-arginine) into NO, which signals by binding to haem active sites or reacting with cysteine residues to form protein-S-nitrosothiols via S-nitrosylation (i.e. the covalent attachment of NO to protein cysteine residues) [176]. NO also rapidly reacts with O2•− to form peroxynitrite (ONOO), a potent oxidant capable of oxidising biomolecules (lipids, amino acids and carbohydrates) and forming 3-nitrotyrosine [176]. A growing body of evidence indicates that IDO1, as a haem enzyme rich in redox-active cysteine and tyrosine residues, is subject to redox control by various physiological RNS and ROS, namely NO, peroxynitrite and H2O2.

Redox control of IDO1
Figure 5
Redox control of IDO1

IDO1 is expressed in inflammatory environments that are characterised by the increased local generation of ROS and RNS that are capable of regulating the expression and enzyme activity of IDO1. (A) L-Arg metabolism by NOS generates NO that can (B) suppress IDO1 gene transcription [185], (C) inhibit the dioxygenase activity by reversibly binding to the haem active-site to form an inactive Fe2+-Trp–NO IDO1 adduct [188], as well as (D) enhance proteasomal degradation of IDO1 protein [186]. (E) IDO1 activity and Kyn pathway metabolites can reciprocally control NOS activity by inhibiting NOS expression [184,189,194]. (F) NOX oxidises NADPH to generate superoxide anion radical (O2•−), which can be dismutated to H2O2 by SOD. H2O2 can be (G) consumed by antioxidant enzymes, such as catalase, glutathione peroxidase (GPx) and peroxidredoxin (Prxs) or (H) initiate the peroxidase activity of IDO1 [30]. For the latter, H2O2 reacts with the resting-state ferric IDO1 (IDO1-Fe3+) to initially yield the highly reactive compound I that constitutes a ferryl iron and a highly reactive porphyrin radical cation (IDO1-Fe4+-Por). (I) In the absence of L-Trp or other substrates such as ascorbate (Asc), the porphyrin radical cation (Por) rapidly reacts with amino acid residues in IDO1 to create protein-centred radicals, leading to protein oxidation, changes in IDO1 secondary structure and dioxygenase inhibition. Both L-Trp and ascorbate can block this form of oxidative self-inactivation as L-Trp and ascorbate can rapidly scavenge IDO1 compound I to prevent oxidative protein damage, leading to the formation of IDO1 compound II, which refers to the remaining ferryl-iron [30]. NO or nitrite (NO2) can also react with IDO1 compound I to generate NO2 and the nitrogen dioxide radical (NO2) respectively, with (J) the latter capable of reacting with and nitrating IDO1 protein tyrosine residues to form 3-nitrotyrosine [30]. (K) NOS-derived NO can also rapidly react with NOX-derived O2•− to form the potent oxidant peroxynitrite (OONO), an RNS also capable of nitrating tyrosine residues on IDO1 and suppressing the enzyme's dioxygenase activity [209], as well as inhibiting IDO1 gene transcription by decreasing STAT1 activation [127].

Figure 5
Redox control of IDO1

IDO1 is expressed in inflammatory environments that are characterised by the increased local generation of ROS and RNS that are capable of regulating the expression and enzyme activity of IDO1. (A) L-Arg metabolism by NOS generates NO that can (B) suppress IDO1 gene transcription [185], (C) inhibit the dioxygenase activity by reversibly binding to the haem active-site to form an inactive Fe2+-Trp–NO IDO1 adduct [188], as well as (D) enhance proteasomal degradation of IDO1 protein [186]. (E) IDO1 activity and Kyn pathway metabolites can reciprocally control NOS activity by inhibiting NOS expression [184,189,194]. (F) NOX oxidises NADPH to generate superoxide anion radical (O2•−), which can be dismutated to H2O2 by SOD. H2O2 can be (G) consumed by antioxidant enzymes, such as catalase, glutathione peroxidase (GPx) and peroxidredoxin (Prxs) or (H) initiate the peroxidase activity of IDO1 [30]. For the latter, H2O2 reacts with the resting-state ferric IDO1 (IDO1-Fe3+) to initially yield the highly reactive compound I that constitutes a ferryl iron and a highly reactive porphyrin radical cation (IDO1-Fe4+-Por). (I) In the absence of L-Trp or other substrates such as ascorbate (Asc), the porphyrin radical cation (Por) rapidly reacts with amino acid residues in IDO1 to create protein-centred radicals, leading to protein oxidation, changes in IDO1 secondary structure and dioxygenase inhibition. Both L-Trp and ascorbate can block this form of oxidative self-inactivation as L-Trp and ascorbate can rapidly scavenge IDO1 compound I to prevent oxidative protein damage, leading to the formation of IDO1 compound II, which refers to the remaining ferryl-iron [30]. NO or nitrite (NO2) can also react with IDO1 compound I to generate NO2 and the nitrogen dioxide radical (NO2) respectively, with (J) the latter capable of reacting with and nitrating IDO1 protein tyrosine residues to form 3-nitrotyrosine [30]. (K) NOS-derived NO can also rapidly react with NOX-derived O2•− to form the potent oxidant peroxynitrite (OONO), an RNS also capable of nitrating tyrosine residues on IDO1 and suppressing the enzyme's dioxygenase activity [209], as well as inhibiting IDO1 gene transcription by decreasing STAT1 activation [127].

Nitric oxide

Considerable evidence indicates that L-Trp metabolism by IDO1 is functionally linked to L-Arg metabolism by NOS, the latter catalysing the formation of NO, a critical autocrine and paracrine signalling molecule in physiology and disease. Three NOS isoforms are expressed in mammals: neuronal NOS (NOS1) and endothelial NOS (NOS3), which are constitutively expressed, and inducible NOS (NOS2). Under inflammatory conditions, IDO1 can be co-expressed with these different NOS isoforms depending on the cell type and stimuli involved. For example, IDO1 is co-expressed with NOS2 in murine macrophages treated with IFNγ and LPS [165,180,181], whereas NOS3 is co-expressed with IDO1 in human endothelial cells treated with IFNγ or in vivo within the vascular endothelium of inflammatory blood vessels [142,182,183].

Our study was the first to discover that NO inhibits IDO1’s dioxygenase activity and to show that NOS2-derived NO inhibits IDO1 in murine macrophages [180]. With respect to the latter, inhibition of NOS2 activity in murine macrophages stimulated with IFNγ and LPS resulted in a significant elevation of IDO1 activity [180]. Similarly, in IFNγ-stimulated lung neutrophils isolated from LPS-treated mice, IDO1 activity only becomes apparent after NOS inhibition [172]. Treatment of the human urinary bladder cancer cell line RT-4 with a cytokine cocktail co-induced IDO1 and NOS2, with a NOS inhibitor resulting in elevated L-Trp metabolism [87]. LPS treatment of mouse peritoneal cells in the presence of activated Vα14 NKT (natural killer T-) cells resulted in the co-induction of NOS2 and IDO1 with the inhibition of NO synthesis also resulting in enhanced IDO1 activity [184]. Moreover, treatment of IDO1-expressing cells with chemical NO donors has been consistently shown to inhibit IDO1 activity [87,180,185188].

Mechanistic studies indicate that NO can inhibit IDO1 via various different mechanisms including reversible binding of NO to the active-site haem of the IDO1 protein [188], enhancement of proteasomal degradation of IDO1 [186] and suppression of IDO1 gene transcription [185] (Figure 5). With respect to inhibition of IDO1 transcription by NO, this may be species- or cell-type-dependent, as while NO donors decreased Ido1 gene expression in mouse cells [185], they did not alter IDO1 mRNA expression in IFNγ-stimulated human macrophages [188]. Instead, we have shown in human macrophages that NO donors inhibit IDO1 activity at the post-translational level [188]. Studies with recombinant IDO1 showed that NO directly inhibits IDO1 activity by binding to the active-site haem of IDO1 to form a trapped inactive FeII-Trp-NO IDO1 adduct where O2 is displaced from the quaternary IDO1 species [188]. Notably, inhibition of purified and cellular IDO1 dioxygenase activity by NO is reversible, suggesting that this mode of IDO1 control is of physiological relevance [188]. The ability of NO to control IDO1 in human monocytic cell lines, U-937 or THP-1 cells, was reported to be concentration-dependent, with high concentrations being inhibitory and lower concentrations being stimulatory [187]. The reason for this apparent bimodal action of NO on IDO1 activity in these transformed human monocytic cell lines requires clarification.

A common finding in cultured cells is that cytokine or LPS stimulation of human cells affords robust induction of IDO1, but not NOS2, with the converse true in cultured murine cells [7,87]. Importantly, however, elevated IDO1 and NOS expression are noted in human and murine tissues in vivo, particularly in response to infection and inflammation [7]. In vivo evidence also supports a role for the regulation of IDO1 by NO in the context of allograft transplantation [189] and responses to infection [190]. Thus, Toxoplasma gondii infection induced the co-expression of IDO1 and NOS within the lung and treatment of infected mice with a NOS inhibitor resulted in a further increase in plasma Kyn levels compared with untreated infected mice [190]. Also, CTLA4-Ig-induced heart allograft transplantation tolerance in rats required the co-induction of IDO1 and NOS activities; inhibition of NOS in MLRs (mixed lymphocyte reactions) using splenocytes from CTLA4-Ig-treated rats resulted in an increase in IDO1-catalysed L-Trp consumption [189]. Interestingly, IDO1 inhibition in the same splenocyte MLR cultures resulted in the increased production of NO, indicative of IDO1-mediated inhibition of NOS.

Several other studies also indicate that the IDO1 pathway can negatively and positively cross-regulate the NOS pathway [7]. For example, mouse peritoneal cells derived from IDO1−/− mice showed elevated NO production upon LPS treatment in the presence of activated Vα14 NKT cells [184]. In the mouse macrophage cell line BAC1.2F5, whereas IFNγ induced both NOS2-dependent NO synthesis and IDO1-catalysed L-Trp metabolism, NOS2 expression and NO output was significantly reduced at low L-Trp levels (0–10 μM), indicating the requirement of L-Trp itself or a Kyn pathway metabolite for NOS2 expression in the cells [181]. With respect to the latter possibility, treatment of murine macrophages with the Kyn pathway metabolite picolinic acid synergises with IFNγ to elevate NOS2 transcription in a manner dependent on the hypoxia-sensitive transcription factor consensus site expressed within the NOS2 promoter [191193]. In contrast, 3-HAA inhibits mRNA expression of NOS2 through NF-κB inhibition in RAW264.7 mouse macrophages stimulated with IFNγ and LPS [194]. Whether the relatively high concentrations of these Kyn pathway metabolites required to modulate NOS2 in vitro are achieved in vivo remains unknown.

The above findings indicate that the IDO1-mediated L-Trp and NOS-dependent L-Arg metabolic pathways are characterised by various modes of reciprocal regulation and interconnections. It is important to note, however, that the extent and nature of interplay between the NOS2 and IDO1 pathway may significantly differ between mice and humans, which probably relates to a species differential with respect to the expression of these different enzymes in response to inflammatory stimuli. As indicated above, whereas IFNγ and LPS are potent stimuli of murine NOS2 expression, expression of NOS2 in human innate immune cells is subject to more tight regulation requiring stimulation by combinations of cytokines and inflammatory stimuli or microbial infection, and, when expressed, is generally at considerably lower levels when compared with NOS2 expression in murine counterparts [7,195]. Similarly, human cells express IDO1 in response to IFNγ and LPS to a markedly greater extent than their murine counterparts. Moreover, although there are reports of human APCs expressing NOS isoforms (e.g. [195,196]), controversy exists with respect to the relative levels of production and the biological ramifications of NOS-derived NO produced by human innate and adaptive immune cells [197200]. Therefore, although co-expression of IDO1 and NOS are noted in certain human and murine tissues in vivo in response to infection and inflammation [7], caution is required in readily extrapolating murine data on the interplay between NOS2 and IDO1 to the human situation. Therefore the extent to which IDO1 and NOS cross-talk to regulate inflammatory and immune responses in vivo during physiological and pathological settings where both enzymes coexist and the regulatory modes involved, particularly in human settings, remains an important area of further investigation.

Hydrogen peroxide

It has been recognised for some time that H2O2 inhibits the dioxygenase activity of IDO1. As such, the H2O2-metabolising enzyme catalase is added to the commonly employed ascorbate/methylene blue IDO1 enzyme assay to remove H2O2 formed during the reaction [136]. Traditionally, under these conditions, any H2O2 formed is considered to limit the IDO1 dioxygenase activity by oxidising the active ferrous form back into the inactive ferric enzyme. However, more recent studies show that the treatment of ferric IDO1 with H2O2 also potently inhibits the enzyme's dioxygenase activity when subsequently assayed using the ascorbate/methylene blue system [30,201].

An early study suggested that this involved the direct oxidation of IDO1 cysteine residues by H2O2 [201]. More recently, however, it has become apparent that IDO1 is similar to many other haem proteins in that reaction of the ferric enzyme with H2O2 yields the formation of compounds I/II necessary for IDO1 to catalyse peroxidase and peroxygenase reactions, which may have important implications for the control of IDO1’s dioxygenase activity and biological actions in local cellular or tissue environments where H2O2 is elevated [30,202,203].

An initial study by Lu and Yeh [203] showed that treatment of ferric IDO1 with H2O2 formed a ferryl-oxo haem (FeIV=O) species reminiscent of compound II of haem peroxidase enzymes and that IDO1 could catalyse the one-electron reduction of the synthetic peroxidase substrates. We more recently showed that treatment of IDO1-expressing cells with H2O2 inhibited IDO1 dioxygenase activity in a manner independent of changes to IDO1 protein expression [30]. Similarly, Lactobacillus johnsonii infection has been reported to inhibit IDO1 expression and activity in intestinal epithelial cells due to the bacterial production of H2O2 [42].

Studies with recombinant human IDO1 demonstrated further that IDO1 catabolised H2O2 via its haem peroxidase activity. In the absence of reactive peroxidase substrates, H2O2 metabolism yielded compound I-mediated IDO1 protein oxidation, concomitant alteration in the protein conformation and consequently the misaligned or non-productive binding of L-Trp within the active site, which inhibits dioxygenase activity [30] (Figure 5). Notably, the presence of L-Trp protects against H2O2-induced IDO1 dioxygenase inactivation via two modes of action. First, L-Trp reacts with compound I to inhibit IDO1 protein oxidation and inactivation, a reaction that yields the formation of oxindolylalanine, Kyn and other yet-to-be-characterised L-Trp oxidation products [30]. Secondly, L-Trp, as a poor substrate for IDO1 compound II, retards the turnover of IDO1’s peroxidase catalytic cycle, thereby reducing the ability of H2O2 to promote compound I-mediated IDO1 protein oxidation and inactivation [30] (Figure 5). Ascorbate also inhibits H2O2-induced IDO1 inactivation via its ability to efficiently react with IDO1 compounds I/II, and prevent IDO1 protein oxidation, while accelerating the turnover of the peroxidase cycle of IDO1, thereby promoting H2O2 consumption [30]. Therefore, the extent to which H2O2 inhibits IDO1 activity is likely to be dependent on the local intracellular concentrations of peroxidase substrates, including L-Trp and ascorbate.

The ability of H2O2 to react with the ferric form of IDO1 to promote the formation of a putative compound I significantly broadens the types of substrates with which the enzyme reacts and the types of reactions the enzyme catalyses. Thus, as a peroxidase, IDO1 can catalyse the oxidation of various physiological substrates including L-Trp, ascorbate, tyrosine, melatonin, NADH, nitrite and NO [30,204,205]. Interestingly, IDO1’s peroxidase function catalytically consumes NO [30], verifying that, similar to other haem peroxidases (e.g. myeloperoxidase [177]), IDO1 can exhibit NO oxidase activity. L-Trp inhibits IDO1-catalysed NO consumption, supporting the hypothesis that L-Trp is a preferred substrate for IDO1 compound I compared with NO [30]. The finding that IDO1 is an NO oxidase indicates that IDO1, as a peroxidase, may modulate the signalling actions and bioactivity of NO and adds a further level of interrelationship between the IDO1 and NOS pathways, in addition to the NO-mediated inhibition of human IDO1 dioxygenase activity alluded to in the previous section (Figure 5).

Notably, Kuo and Mauk [202] showed that human IDO1 in the presence of indole and H2O2 exhibits a haem peroxygenase activity, which refers to the concerted insertion of the ferryl oxygen atom of IDO1 compound I into indole to produce oxindoles and analogues of NFK. Our recent data [30] indicate that IDO1’s peroxygenase activity may also be responsible for the H2O2-dependent conversion of L-Trp into oxindolylalanine, although the efficiency of L-Trp as a putative peroxygenase substrate [30] is markedly less when compared with indole as a substrate [202].

Despite catalysing identical haem dioxygenase reactions, IDO1 and TDO differ markedly with respect to the implications of H2O2 for their dioxygenase activity. Thus, whereas compound I-mediated protein oxidation appears to be responsible for the inhibitory action of H2O2 on IDO1 dioxygenase activity [30], Fu et al. [29] reported that compound I-mediated formation of a protein-based radical is a key intermediate step by which H2O2 reactivates the dioxygenase activity of TDO. They proposed that H2O2-induced TDO reactivation involves a two-electron reduction of a FeIV=O intermediate by L-Trp to form a FeII-Trp adduct. In contrast, we and others report that the reaction of IDO1 with H2O2 in the presence of L-Trp forms the FeIV=O intermediate as the major detectable IDO1 haem species [30,203]. Therefore, unlike TDO where L-Trp facilitates dioxygenase reactivation by H2O2, L-Trp is oxidised by IDO1 compound I and retards the reduction of the FeIV=O haem of IDO1 compound II, resulting in the protection of IDO1 from H2O2-induced protein oxidation and dioxygenase inactivation. Whereas various haem enzymes in the presence of H2O2, including TDO [29], can catalyse a catalase-like activity, this is not the case for IDO1 [30,206]. Therefore, H2O2 metabolism by IDO1 proceeds primarily via its peroxidase cycle, with the reduction of IDO1 compound II being the rate-limiting step [30].

Although these recent studies highlight that IDO1 is a versatile haem enzyme capable of reacting with H2O2 and catalysing compound I-dependent peroxidase and peroxygenase activities, key questions remain. At the biophysical and biochemical level, actual detection of IDO1 compound I has yet to be achieved. This may prove challenging given the high reactivity of compound I, particularly in pseudo-peroxidases such as IDO1, where the porphyrin radical probably transfers rapidly to adjacent amino acids to form protein-centred radicals. A further issue is the extent to which H2O2 reacts with intracellular IDO1 to activate its peroxidase activity in order to modulate dioxygenase activity, and how this affects the enzyme's biological actions. In an inflammatory environment where H2O2 levels are elevated, local IDO1-mediated depletion of L-Trp may enhance the susceptibility of IDO1 to peroxidase-mediated self-inactivation. Whether oxidative inactivation of a key immune regulatory protein such as IDO1 (detailed in the ‘Immune regulation’ section below) represents a mechanism by which oxidative stress influences chronic inflammation also requires investigation.

Another issue requiring clarification is the extent to which activation of IDO1’s peroxidase function and resultant H2O2 metabolism by IDO1 alters the intracellular redox status. In support of a potential antioxidant role, overexpression of IDO1 in human umbilical vein endothelial cells and bronchial epithelial cells protects against H2O2-induced necrosis and apoptosis by attenuating intracellular ROS production [207,208] and preserving mitochondrial integrity [208]. Activation of IDO1’s peroxidase function may, however, also elicit pro-oxidant actions via enhancing the consumption of critical intracellular reductants such as ascorbate [30].

Peroxynitrite and tyrosine nitration

In addition to NO and H2O2, IDO1 dioxygenase activity is inhibited by the potent and versatile oxidant peroxynitrite, which is capable of nitrating tyrosine residues in proteins. Several studies report the detection of tyrosine-nitrated IDO1 within cells. For example, exposure of the acute monocytic leukaemia cell line THP-1 to the peroxynitrite generator SIN-1 (3-morpholinosydnonimine N-ethylcarbamide) resulted in the detection of nitrated IDO1 enzyme and inhibition of IDO1 activity independent of changes to IDO1 protein expression levels [209]. Also, Lopez et al. [187] reported the detection of nitrated IDO1 protein in unstimulated U-937 and THP-1 cells.

Studies with recombinant human IDO1 protein reported that exposure to peroxynitrite resulted in the nitration of three of the 12 tyrosine residues (i.e. Tyr15, Tyr345 and Tyr353) present in human IDO1, with nitration of Tyr15 accounting for up to 30% of the inhibitory action of peroxynitrite on IDO1’s dioxygenase activity [209]. Although tyrosine nitration could partially explain peroxynitrite-induced IDO1 inhibition, other yet-to-be-identified redox reactions are likely to be involved. Moreover, how tyrosine nitration may modulate IDO1’s active-site haem to influence dioxygenase activity is currently not clear and worthy of investigation.

We recently reported that the peroxidase function of IDO1 also yields tyrosine nitration, i.e. IDO1 is similar to other haem peroxidase proteins in that exposure of IDO1 to H2O2 in the presence of low micromolar concentrations of nitrite (NO2) results in the formation of NO2 and resultant nitration of IDO1 tyrosine residues [30]. Our recent findings indicate that, rather than enzyme inhibition, tyrosine nitration in the presence of NO2 and H2O2 correlates with the protection of IDO1 against dioxygenase inactivation by H2O2, consistent with NO2 reacting with compound I to attenuate oxidative protein damage (Y.J. Lim, M. Freewan, M.D. Rees, A.W.S. Yeung, E.N. Glaros, P.K. Witting, A.C. Terentis and S.R. Thomas, unpublished work).

In addition to direct inhibition of IDO1 activity, a previous study indicated that peroxynitrite inhibits IDO1 mRNA expression by nitrating and inactivating STAT1 signalling and that this defect in IDO1-mediated signalling contributes to the augmented immune response associated with Type 1 diabetes [127].

Cysteine/glutamate antiporter

Studies indicate that IDO1 is also subject to redox control in human monocyte-derived DCs through modulation of the activity of the cysteine/glutamate antiporter [210,211]. This antiporter acts to preserve a reducing intracellular redox environment by facilitating the intracellular transport of cysteine necessary to support the biosynthesis of glutathione, the most abundant small-molecule reductant contained within cells. The antiporter also influences the local extracellular redox environment through exporting cellular glutathione and cysteine. Blocking antiporter activity in human DCs, which decreases intracellular glutathione levels [212], up-regulated IDO1 mRNA expression and enzyme activity and this correlated with reduced antigen-presentation activity of DCs towards T-cells [210,211]. Our previous data also provided evidence that altering the intracellular redox status through addition of various structurally unrelated redox-active agents modulates IDO1 expression and activity in IFNγ-stimulated human monocyte-derived macrophages [160].

This section highlights that IDO1 is subject to various modes of redox control involving different ROS (H2O2) and RNS (NO, peroxynitrite and nitrogen dioxide) and their reactions with the IDO1 active-site haem and/or amino acids. In addition to redox control of IDO1 dioxygenase activity, the above highlights that IDO1 is a versatile haem enzyme capable of reacting with different redox-active species to catalyse a variety of redox reactions, which involve interplay or cross-talk between different ROS and RNS pathways. For example, whereas NOS-derived NO can bind IDO1’s active-site haem to inhibit its dioxygenase activity, reaction of IDO1 with H2O2 in the presence of NO yields catalytic consumption of NO, potentially reducing its signalling bioactivity (Figure 5). Although there is some evidence for IDO1 redox control in vivo, further studies are required to investigate the extent to which different ROS and RNS modulate IDO1’s dioxygenase and peroxidase activities and how these affect IDO1’s biological actions, particularly during inflammation characterised by the co-expression of IDO1 with ROS- and RNS-producing enzymes, namely, isoforms of NOS and NOX (Figure 5).

Tyrosine phosphorylation and protein–protein interactions

Recent studies indicate that IDO1 is subject to the phosphorylation of two conserved tyrosine residues that, depending on the cellular environment, can impart enzyme activity-independent cell signalling properties to the IDO1 protein [113] (see the ‘IDO1 signalling’ section below) or signal for IDO1 protein proteasome degradation [213] (see the ‘Proteasomal degradation’ section below) (Figure 2). Both of these events involve novel protein–protein interactions between IDO1 and relevant signalling proteins.

IDO1 signalling

In pDCs, recent data indicate that IDO1 can exhibit a novel cell signalling activity independent of its enzyme activity. This involves tyrosine phosphorylation of two putative ITIMs (immunoreceptor tyrosine-based inhibitory motifs), ITIM1 (VPY115CEL) and ITIM2 (LLY253EGV), present in the small domain and interconnecting loop respectively of IDO1 and exposed on the protein surface in close proximity to each other [113] (Figure 2). Treatment of murine pDCs with TGFβ induces tyrosine phosphorylation of IDO1 at Tyr115 and Tyr253 involving the Src family kinase member Fyn [113]. Such tyrosine phosphorylation signals for the recruitment and binding of SHP-1 and SHP-2 (Src homology 2 domain-containing protein tyrosine phosphatase 1 and 2) to the IDO1 protein, which activates the non-canonical NF-κB pathway and the production of type I IFNs that confer a stable immunosuppressive regulatory phenotype on TGFβ-stimulated mouse pDCs (see also the ‘Immune regulation’ section below). A most recent study in LPS-challenged mice indicates that the activation of AhR by TDO-derived Kyn also signals for Src kinase-dependent IDO1 tyrosine phosphorylation in pDCs and immune suppression [25]. These two important studies [25,113] identifying a non-enzymatic signalling action of IDO1 represents a new paradigm in IDO1 research. Various questions remain to be answered including the extent to which these IDO1 tyrosine residues are phosphorylated in cell types other than pDCs and the stimuli/conditions involved, whether other tyrosine, serine and/or threonine residues in IDO1 are subject to phosphorylation and the identity of the kinases involved, and whether phosphorylation can affect IDO1’s various enzyme activities.

Proteasomal degradation

Increasing evidence indicates that IDO1 is also subject to post-translational control by proteolytic degradation of the enzyme via the proteasome, which also requires phosphorylation of Tyr115 and Tyr253. Thus, treatment of murine CD8α+ DCs with IL-6 enhances IDO1 proteasomal degradation [213]. An important requirement linking tyrosine phosphorylation and proteasomal degradation is the binding of SOCS3 (suppressor of cytokine signalling 3) to the tyrosine-phosphorylated ITIM motifs within IDO1 that recruits the ECS (elongin/cullin/SOCS) E3 ligase and targets the IDO1–SOCS3 protein complex for ubiquitination and proteasomal degradation [213]. Other studies also indicate IDO1 degradation via the proteasome. For example, treatment of IDO1-expressing cells with the putative chemopreventive agents sodium butyrate or fludarabine, down-regulate cellular IDO1 protein levels by increasing the ubiquitination of the IDO1 protein, thus enhancing proteasomal degradation of the enzyme [214,215]. Also, exposure of an IDO1-expressing human epithelial cell line to chemical NO donors induces proteasomal degradation of IDO1 [186]. The precise signalling pathways underlying NO-mediated IDO1 proteasomal degradation and its importance for cellular IDO1 control are currently unknown. It has also been proposed recently, considering that ligand-bound AhR associates with the ubiquitin–ligase complex of the proteasome system, that, in addition to governing IDO1 transcription, AhR ligands may also modulate IDO1 activity by influencing protein turnover [216].

This section illustrates the multi-layered biochemical control of cellular IDO1 expression and enzymatic activity involving transcriptional and post-translational events and the action of various signalling modalities including protein kinase and transcription factor pathways, haem biosynthesis and degradative pathways, ROS- and RNS-producing enzymes, tyrosine-phosphorylation-dependent protein–protein interactions and the proteasome. Continued characterisation of the fundamental biochemical mechanisms governing IDO1 expression and activity is critical with respect to the development of novel strategies capable of modulating (i.e. induction or inhibition) IDO1 and its biological actions in vivo, which may represent new therapeutics in treating conditions where IDO1 plays an influential role in determining disease severity and outcome. Below, we discuss the biological roles of IDO1 and their relevance to physiological and disease conditions.

BIOLOGICAL ROLES OF IDO1

The activation of IDO1 and the resultant metabolism of L-Trp and formation of Kyn and its metabolites have considerable biological ramifications, which, depending on the circumstance, can be beneficial or deleterious to the host. As detailed below, IDO1 has been most commonly studied in the context of immune regulation and as a host response to infection.

Immune regulation

The ability of the host immune system to discriminate between self- and non-self-antigens and to respond appropriately via the initiation or suppression of an immune response is dependent on the prevailing immunological milieu. In the last 18 years, a considerable body of work has identified cellular IDO1 expression, particularly in APCs, as having a central role in the creation of a local immunosuppressive microenvironment, whereby a reduction in L-Trp and the production of biologically active Kyn pathway metabolites, as well as tyrosine phosphorylation of IDO1, can potently inhibit T-cell immunity and promote immunosuppressive Treg responses. Notably, the complexities and scope of IDO1’s immune control actions are becoming increasingly appreciated, including catalytically-dependent and catalytically-independent immune regulatory actions, signalling for anti-inflammatory and pro-inflammatory outcomes, and the modulation of the activation status and function of a variety of immune cells in addition to T-cells including NK (natural killer) cells, B-cells, DCs and macrophages. These various aspects of IDO1’s immune control actions are outlined below.

Mechanistic basis of immune control by IDO1

The ability of IDO1 to modulate the immune response was first discovered in 1998 by Munn et al. [217] in pregnant mice, where IDO1 expressed at the fetal/maternal interface was shown to prevent T-cell-mediated rejection of allogeneic fetuses. Efforts since then to understand the mechanisms behind IDO1-mediated T-cell suppression have focused mainly on APCs, such as DCs and macrophages, which, depending on their maturation status and the immune microenvironment, can express active IDO1 in vitro and in vivo.

Expression of active IDO1 is apparent in various types of immune cell types and subsets therein, as detailed below. Briefly, functional IDO1 with important implications for immune control is observed in traditional APCs, including murine CD8α+ DCs and CD19+ pDCs, as well as human CD123+ CCR6+ pDCs, Bregs (regulatory B-cells) and activated monocytes and macrophages (including microglia) [49,51,157,158,218220]. In addition to APCs, other innate immune cells such as NK cells, eosinophils and neutrophils can also be IDO1-competent [50,52,221].

The immune-modulatory action of IDO1 involves several different biochemical or metabolic-regulatory mechanisms primarily acting on T-cells [e.g. promotion of an immunosuppressive milieu via Treg differentiation, induction of apoptosis of particular T-cell subsets or defective TCR (T-cell receptor) activation], NK cells (e.g. down-regulation of activating receptors or cell death), DCs (e.g. control of DC maturation and migration) and macrophages (e.g. control of macrophage phenotype and cytokine production). The various modes of IDO1-mediated immune regulation are summarised in Figure 6 and described in detail herein.

Regulation of immune responses by IDO1
Figure 6
Regulation of immune responses by IDO1

Induction of IDO1 expression and activity in professional APCs, such as DCs and macrophages (Mϕ), has a multitude of effects on the immune system that can be beneficial or deleterious to the host, depending on the context. (A) Active IDO1 enzyme consumes local L-Trp, activating the GCN2 kinase integrated stress response pathway [225] and inhibiting the mTOR and PKCθ pathways [227]. Kyn, derived from IDO1-mediated L-Trp degradation, can activate the AhR, a cytosolic ligand-activated transcription factor [233]. Depending on the cell type and its expression of downstream enzymes in the Kyn pathway, Kyn may be metabolised further. In certain situations, tyrosine residues in the IDO1 protein can become phosphorylated [113]. Together, (I) L-Trp deprivation and activation of GCN2 signalling, (II) Kyn-dependent activation of AhR signalling, (III) synthesis of downstream Kyn pathway metabolites, and (IV) IDO1-Tyr phosphorylation act either alone or in combination to inhibit the T-cell responses via several mechanisms, including (B) the differentiation of naïve T-cells into immunosuppressive Tregs that (C) inhibit effector T-cell function [232,242244]. In turn, Tregs can (D) further induce IDO1 expression in APCs via the binding of CTLA-4 to CD80/86 [99]. (E) Active IDO1 in APC can also directly inhibit the function of effector T-cells by promoting Th2 cell polarisation [237,238], induce defective TCR activation [241], clonal anergy [225], as well as cell cycle arrest and apoptosis in effector T-cells [230]. Notably, this suppression of T-cell function is not confined to the classic αβ T-cell subtype, but also applies to γδ T-cells and invariant NKT cells [159,237]. (F) Kyn metabolite signalling via the AhR can also stabilise the immunosuppressive phenotype of Tregs and prevent their reprogramming to effector helper T-cells by blocking the down-regulation of Eos, a transcription factor that complexes with Foxp3 and is required for the immunosuppressive activity of Foxp3 [245]. (G) Of note, neighbouring IDO1 DCs can take up Kyn and other metabolites from the microenvironment in a paracrine manner, and, if appropriately stimulated to express other Kyn pathway enzymes, can also inhibit effector T-cell function via the production of Kyn metabolites, conferring on these previously immunogenic IDO1 DC regulatory properties and further enhancing the immunosuppressive milieu [248]. (H) Interestingly, IDO1 tyrosine phosphorylation and AhR activation by Kyn can both induce IDO1 gene expression [113,234,235] (see also Figure 2), thereby amplifying and maintaining this immunoregulatory milieu. IDO1 expression also affects the function of other leucocytes. (I) Kyn and downstream metabolites can induce NK cell death [230], as well as the down-regulation of activating receptors, inhibiting NK cell-mediated killing [115]. (J) In macrophages, IDO1 expression can promote polarisation towards an anti-inflammatory M2 phenotype [255,256], or conversely, lead to the induction of pro-inflammatory cytokines by activating the GCN2 kinase stress-response pathway [259]. (K) Further emphasising IDO1’s pro-inflammatory role in certain contexts, IDO1 can also promote the development of autoreactive B-cells [251,330]. (L) IDO1 expression, L-Trp depletion and production of Kyn have also been linked to increased neutrophil apoptosis [263]. (M) DC maturation, migration and immune-stimulatory capacity are also dependent on IDO1 activity under certain conditions [249,250].

Figure 6
Regulation of immune responses by IDO1

Induction of IDO1 expression and activity in professional APCs, such as DCs and macrophages (Mϕ), has a multitude of effects on the immune system that can be beneficial or deleterious to the host, depending on the context. (A) Active IDO1 enzyme consumes local L-Trp, activating the GCN2 kinase integrated stress response pathway [225] and inhibiting the mTOR and PKCθ pathways [227]. Kyn, derived from IDO1-mediated L-Trp degradation, can activate the AhR, a cytosolic ligand-activated transcription factor [233]. Depending on the cell type and its expression of downstream enzymes in the Kyn pathway, Kyn may be metabolised further. In certain situations, tyrosine residues in the IDO1 protein can become phosphorylated [113]. Together, (I) L-Trp deprivation and activation of GCN2 signalling, (II) Kyn-dependent activation of AhR signalling, (III) synthesis of downstream Kyn pathway metabolites, and (IV) IDO1-Tyr phosphorylation act either alone or in combination to inhibit the T-cell responses via several mechanisms, including (B) the differentiation of naïve T-cells into immunosuppressive Tregs that (C) inhibit effector T-cell function [232,242244]. In turn, Tregs can (D) further induce IDO1 expression in APCs via the binding of CTLA-4 to CD80/86 [99]. (E) Active IDO1 in APC can also directly inhibit the function of effector T-cells by promoting Th2 cell polarisation [237,238], induce defective TCR activation [241], clonal anergy [225], as well as cell cycle arrest and apoptosis in effector T-cells [230]. Notably, this suppression of T-cell function is not confined to the classic αβ T-cell subtype, but also applies to γδ T-cells and invariant NKT cells [159,237]. (F) Kyn metabolite signalling via the AhR can also stabilise the immunosuppressive phenotype of Tregs and prevent their reprogramming to effector helper T-cells by blocking the down-regulation of Eos, a transcription factor that complexes with Foxp3 and is required for the immunosuppressive activity of Foxp3 [245]. (G) Of note, neighbouring IDO1 DCs can take up Kyn and other metabolites from the microenvironment in a paracrine manner, and, if appropriately stimulated to express other Kyn pathway enzymes, can also inhibit effector T-cell function via the production of Kyn metabolites, conferring on these previously immunogenic IDO1 DC regulatory properties and further enhancing the immunosuppressive milieu [248]. (H) Interestingly, IDO1 tyrosine phosphorylation and AhR activation by Kyn can both induce IDO1 gene expression [113,234,235] (see also Figure 2), thereby amplifying and maintaining this immunoregulatory milieu. IDO1 expression also affects the function of other leucocytes. (I) Kyn and downstream metabolites can induce NK cell death [230], as well as the down-regulation of activating receptors, inhibiting NK cell-mediated killing [115]. (J) In macrophages, IDO1 expression can promote polarisation towards an anti-inflammatory M2 phenotype [255,256], or conversely, lead to the induction of pro-inflammatory cytokines by activating the GCN2 kinase stress-response pathway [259]. (K) Further emphasising IDO1’s pro-inflammatory role in certain contexts, IDO1 can also promote the development of autoreactive B-cells [251,330]. (L) IDO1 expression, L-Trp depletion and production of Kyn have also been linked to increased neutrophil apoptosis [263]. (M) DC maturation, migration and immune-stimulatory capacity are also dependent on IDO1 activity under certain conditions [249,250].

Lymphocytes

Early in vitro studies were the first to indicate that IDO1-expressing DCs and macrophages induce cell cycle arrest in T-cells in a manner that is dependent on IDO1-mediated L-Trp metabolism [49,51,222]. However, IDO1-expressing immune cells do not necessarily inhibit T-cell responses by preventing TCR activation. Instead, IDO1 or L-Trp starvation induces Fas-mediated cell cycle arrest of T-cells in mid-G1-phase, T-cell apoptosis and clonal anergy and inhibits antigen-specific T-cell responses [103,158,223,224]. For example, ex vivo experiments studying CTLA-4-stimulated and IDO1-expressing murine splenic pDCs and CD8α+ DCs reported that, although T-cell activation could still occur, these activated T-cells rapidly and preferentially underwent apoptosis in response to IDO1 activation such that clonal expansion was prevented [103]. Inhibition of T-cell function was also observed in a mouse melanoma model, where CD19+ pDCs from tumour-draining lymph nodes were responsible for creating local T-cell anergy in an IDO1-dependent manner [158]. A recent in vitro study has also implicated IDO1 expression in the T-cell-suppressive action of CpG-ODN-stimulated Bregs [220]. Interestingly, IDO1 in microvesicles derived from IFNγ-stimulated human amniotic fluid stem cells is also capable of inhibiting T-cell responses, mirroring the effect of the IDO1-expressing stem cells themselves [58].

In many of the early in vitro human and mouse studies, the restoration of effector T-cell activity afforded by the addition of excess L-Trp suggested that the observed immune suppression was primarily the result of L-Trp deprivation, rather than de novo Kyn pathway metabolite synthesis. In accordance with this, L-Trp deprivation of T-cells increases the levels of uncharged tRNA, activating the GCN2 (general control non-derepressible 2) kinase pathway (leading to a generalised blockade of protein synthesis, but also enhanced synthesis of select proteins that counter this amino acid decrease), and inhibiting the mTOR (mammalian target of rapamycin) and PKCθ signalling pathways to promote T-cell autophagy and anergy [225227]. Together, these constitute an integrated stress response induced by withdrawal of an essential amino acid that accounts for some of the observed immunosuppressive effects induced by IDO1-mediated L-Trp deprivation, including cell cycle arrest and anergy in antigen-specific effector CD8+ T-cells in vivo, the activation of immunosuppressive Tregs and the induction of autophagy [225227].

How IDO1-expressing cells (or other cell types within the immediate environment of IDO1-expressing cells), but not T-cells, remain functional/viable in low L-Trp microenvironments despite a requirement for L-Trp for cellular protein synthesis is primarily attributed to TTS (tryptophanyl-tRNA synthetase), a constitutively expressed enzyme that catalyses the association of L-Trp with its tRNA for incorporation into proteins [228]. Differential TTS expression can determine whether a cell is susceptible to IDO1-catalysed L-Trp deprivation, i.e. increased TTS protects cells from the integrated stress response by sequestering sufficient L-Trp for cellular protein synthesis. For example, CTLA-4-Fc-treated human monocyte-derived DCs up-regulate TTS in parallel with IDO1, whereas the enhanced susceptibility of influenza- or phytohaemagglutinin A-activated CD4+ T-cells compared with CD8+ T-cells to IDO1-mediated T-cell inhibition correlates with lower TTS expression in CD4+ T-cells [228].

Although initial studies indicated that L-Trp deprivation was primarily responsible for IDO1-mediated T-cell inhibition, cultured T-cells are suppressed at L-Trp concentrations of <0.5–1 μM [51], which may not be readily achieved in vivo since plasma levels of this amino acid range from ∼40 to 100 μM in mice and humans and, at these concentrations, are likely to diffuse into local tissues with reduced L-Trp levels. This issue has been somewhat reconciled by subsequent in vitro studies indicating that the exogenous addition of relatively high concentrations of Kyn pathway metabolites (∼100–1000 μM), such as Kyn, 3-HK, 3-HAA, QA and picolinic acid, can inhibit the function of activated, but not resting, T-cells, as well as NK cells and B-cells [229231]. Importantly, the T-cell-suppressive action of these metabolites can be achieved at more physiologically relevant concentrations (<∼10–50 μM) when they are added to T-cell cultures in various combinations or under low-L-Trp or L-Trp-free conditions [229,230,232].

How Kyn metabolites mediate their regulatory actions on immune cells has been explained, in part, by the identification of Kyn as a novel endogenous ligand of the AhR transcription factor, which plays essential roles in normal physiology including vascular development, reproduction and immune regulation [114]. Kyn-mediated AhR activation drives the differentiation of naïve CD4+ T-cells following TGFβ stimulation into immunosuppressive Foxp3+ Tregs rather than pro-inflammatory Th17 cells in vitro [233], whereas others report that AhR activation by Kyn can further promote IDO1 expression in DCs [234,235]. The extent to which other AhR ligands similarly induce IDO1 expression to amplify the immunoregulatory microenvironment is currently not known. Interestingly, kynurenic acid, another Kyn pathway metabolite, can act as an AhR ligand, although its signalling outcome can differ from that of Kyn. Thus, kynurenic acid elicits the expression of IL-6 in a human breast cancer cell line (MCF-7 cells) via AhR, and, when in combination with IL-1β, synergistically induces IL-6 [18]. Meanwhile, 3-HAA also contributes to immune suppression by impairing T-cell function and inducing apoptosis in activated T-cells by depleting glutathione [231] and inhibiting TCR engagement-mediated NF-κB activation [236].

The above findings therefore indicate that IDO1-catalysed L-Trp depletion, the synthesis of Kyn pathway metabolites and the activation of AhR and other signalling pathways control the activation status, phenotype and viability of T-cells, B-cells and NK cells in various ways including: (i) induction of cell cycle arrest in T-cells and induction of cell death in T-cells, B-cells and NK cells [51,229231], (ii) impairment of NK cell-mediated killing by preventing the up-regulation of the activating receptors NKp46 and NKG2D [115], (iii) Th2 polarisation of invariant NKT cells [237], (iv) apoptosis of Th1 cells but not Th2 cells [238], (v) interfering with glucose metabolism in effector T-cells to inhibit proliferation [239,240], (vi) defective TCR activation via the interruption of Ca2+ signalling [241], (vii) down-regulation of the ζ-chain in the TCR complex on CD8+ T-cells [232], (viii) acquisition and maintenance of an immunosuppressive Foxp3+ Treg (rather than a pro-inflammatory Th17) phenotype by naïve CD4+ T-cells [232,242244] involving the IDO1- and Kyn/AhR-dependent blockade of the down-regulation of the Eos transcription factor (necessary for reprogramming of Tregs into Th17 inflammatory helper-like cells) [245], and (ix) direct activation of the suppressor activity of Foxp3+ Tregs involving the PD-1 (programmed cell death 1)/PD-L (programmed cell death ligand) pathway [226].

Although it is generally accepted that IDO1 represents a potent inhibitory signal for Th1 responses, the implications of IDO1 and the Kyn pathway for Th2 responses appears more complex, with both inhibitory and stimulatory actions reported. For example, Kyn pathway metabolites such as 3-HAA preferentially induce apoptosis of Th1, but not Th2, cells [238]. Similarly, IDO1-expressing human eosinophils inhibit Th1 cells but support Th2 cells [50]. In vivo, both Th2-supportive and -inhibitory actions of IDO1 and Kyn pathway metabolites have been described. For example, in a murine model of asthma due to OVA (ovalbumin) sensitisation, CpG-ODN (a TLR9 ligand) induced IDO1 expression that inhibited Th2-induced asthma (as determined by experiments with wild-type mice treated with 1-MT) [55]. In contrast, in another murine asthma model, compared with wild-type mice, Ido1 gene-deficient mice exhibited reduced Th2 responses and were protected against allergic disease [246]. These in vivo studies are discussed in more detail in the ‘IDO1 and immune regulation in physiological and disease settings’ section below.

A feature of IDO1’s immune regulatory function involving pDCs and Tregs is the establishment of cross-talk between these cells that can augment the immunosuppressive signal. Thus, whereas IDO1-expressing pDCs convert CD4+ T-cells into Foxp3+ Tregs, these latter cells can further induce IDO1 expression in pDCs through the interaction of CTLA-4 expressed on the Treg surface with its B7 (CD80 or CD86) ligands expressed on the DCs [102,247]. Interestingly, Tregs can influence IDO1 expression and tolerance in cell types other than DCs, e.g. CD3/CD28-activated Tregs have been shown to promote TGFβ and IDO1 expression in neutrophils [53].

Of further interest is the ability of IDO1-expressing immunosuppressive murine CD8α+ DCs (and potentially other IDO1-expressing subsets) to induce a similar tolerogenic state in other naïve or previously immunogenic DC subsets via the paracrine production of Kyn pathway metabolites. Thus, although IFNγ-stimulated murine CD8α and CD8α+ DCs can express several functional Kyn pathway enzymes, only CD8α+ DCs express functional IDO1. When Kyn in the culture medium (either derived from active IDO1 expressed in CD8α+ DCs or provided exogenously) is taken up by CD8α DCs, it can be further metabolised along the Kyn pathway, culminating in the production of QA and conferring tolerogenic potential on normally immunogenic CD8α DCs [248].

Recent important studies have revealed that IDO1 can also act independently of the enzyme's dioxygenase activity to regulate immune responses. In TGFβ-treated murine pDCs, IDO1 is subject to tyrosine phosphorylation on Tyr115 and Tyr253, which forms an intracellular signalling scaffold necessary to bind the protein tyrosine phosphatases SHP-1 and SHP-2 and signal for the maintenance of stable long-term immune tolerance generated via the conversion of CD4+ T-cells into immunosuppressive Tregs [113]. In contrast, IFNγ-stimulated pDC require active IDO1 enzyme and resultant L-Trp metabolism to inhibit the immune response, chiefly by suppressing T-cell proliferation and promoting apoptosis [113].

Together, the above findings indicate that the expression of IDO1 protein and the catabolism of L-Trp along the Kyn pathway can potently control T-cell responses via several modalities. The current challenge is to understand the extent to which these different signalling modalities are employed by IDO1-expressing cells to control immune responses under the diverse physiological and pathological conditions outlined in the ‘IDO1 and immune regulation in physiological and disease settings’ section below.

APCs

In addition to regulating the ability of APCs to control T-cell responses, emerging evidence indicates that IDO1 has important implications for APCs themselves. Thus, depending on the maturation stimuli, IDO1 expression is required for the maturation, migration and immune-stimulatory activity of mouse and human DCs in vitro [249,250]. IDO1 is also required for the differentiation, but not activation or survival, of autoantibody-secreting B-cells in a mouse model of rheumatoid arthritis [251]. Thus, IDO1 is part of a highly complex host response wherein the enzyme's expression in APCs can influence the direction of the immune response by affecting both non-IDO1- and IDO1-expressing cells.

In immature human DCs, IDO1 inhibition with 1-MT treatment decreases constitutive surface expression of the chemokine receptor CXCR4 [250]. Moreover, IDO1 inhibition with 1-MT or methylthiohydantoin-Trp has been reported to down-regulate the expression of co-stimulatory molecules on poly(I:C)-, TNFα- or LPS-activated human monocyte-derived DCs, and, to a lesser extent, DCs transfected to express CD40L [249,250], inhibiting the ability of these cells to stimulate T-cell proliferation [250]. The specificity of action of 1-MT for IDO1 in DCs has, however, been queried by a study reporting that, depending on the maturation signal, 1-MT employed at relatively high concentrations in vitro (i.e. 1 mM) can influence DC function independently of its effects on IDO1 [252]. Importantly, however, a role for IDO1 in directly regulating APC function also comes from experiments employing siRNA gene silencing of IDO1, which, similarly to IDO1-inhibitory drugs, reduces the expression of co-stimulatory molecules in poly(I:C)-, TNFα- or LPS-activated human monocyte-derived DCs [249]. Moreover, murine bone marrow-derived DCs from IDO1−/− mice showed that the maturation of GM-CSF (granulocyte/macrophage colony-stimulating factor)-stimulated, but not Flt3L (Fms-like tyrosine kinase 3 ligand)-stimulated DCs, was impaired, as was their response to LPS, Pam3CSK4 [tripalmitoylcysteinylseryl-(lysyl)4] and CpG-ODN treatment [163,253]. Together, these data support that IDO1 can influence DC maturation and stimulatory activity, although caution needs to be taken with respect to the actions of high concentrations (≥1 mM) of 1-MT, which may alter DC responses independently of IDO1 [252].

Treatment of DCs with Kyn pathway metabolites has also been reported to directly modulate DC function. For example, the redox-active Kyn metabolites 3-HAA and 3-HK contribute to LPS-mediated maturation of human monocyte-derived DCs (as determined by CD80, CD86 and CD83 expression), apparently via ROS production and NF-κB activation [249]. In contrast, treatment of LPS-stimulated murine bone marrow-derived DCs with 3-HAA inhibits inflammatory cytokine production (IL-12, IL-6 and TNFα), the expression of DC maturation markers (CD80, CD86 and I-A) and the capacity of DCs to stimulate T-cells [254]. Moreover, in vivo administration of 3-HAA also inhibits the ability of DCs to activate T-cells [254]. Notwithstanding the above, Ido1 gene-knockout mice do not exhibit altered DC responses towards LPS or Leishmania major infection in vivo, being both phenotypically and functionally comparable with DCs from wild-type mice [253], suggesting a certain level of redundancy in IDO1 function with respect to DC responses in mice.

Emerging evidence indicates that IDO1 can also elicit direct actions when expressed within macrophages. For example, IDO1 expression has been identified as a determinant of macrophage M1 rather than M2 phenotype, with the enzyme signalling in favour of the anti-inflammatory M2 phenotype [255,256]. Studies in the RAW264.7 murine macrophage cell line indicated that incubation of these cells with IDO1-expressing fibroblasts resulted in reduced NOS2 expression and macrophage viability due to L-Trp deprivation [257]. IDO1-expressing fibroblasts similarly reduced macrophage numbers and xenograft survival in a macrophage-dependent pancreatic islet xenograft-rejection model [257]. Further evidence for control of macrophages by IDO1-expressing bystander cells comes from a report with IDO1-expressing endometrial stromal cells impairing macrophage phagocytic activity in a manner dependent on IL-33 production [258]. A recent study identified a role for IDO1 in regulating macrophage inflammatory cytokine production [259]. Thus, IDO1-mediated L-Trp deprivation and resultant activation of the GCN2-dependent metabolic stress-signalling pathway augmented LPS-induced IL-12 and IL-6 production in murine macrophages, providing a novel cellular mechanism by which IDO1 could conceivably promote inflammation [259]. A similar role for IDO1 in signalling for elevated inflammatory cytokine production has been recognised previously in IFNγ-stimulated airway epithelial cells, where IDO1-mediated L-Trp deprivation enhanced IL-6 and IL-8 responses by stabilising IL-6 and IL-8 mRNA [260]. Further evidence for IDO1 regulating gene expression comes from studies in human dermal fibroblasts reporting the ability of IFNγ to inhibit the IL-1β-stimulated mRNA expression of matrix metalloproteinase enzymes (collagenase and stromelysin) involved in IDO1-catalysed L-Trp starvation [261].

The above studies identify significant differences with respect to the implications of IDO1-catalysed L-Trp deprivation and activation of the GCN2 pathway for naïve lymphocytes (signalling for inhibition of activation and immune suppression) [225] compared with macrophages (signalling for cytokine production and inflammation) [259]. As such, the differentiation status of a cell may determine the ultimate effect of L-Trp depletion, with L-Trp deprivation and GCN2 activation applying greater stress to proliferating than to differentiated cells, with the former cell type having a greater demand for newly synthesised proteins. Thus, in more differentiated cells, such as macrophages and Tregs, GCN2 activation has a signalling function [226,259], whereas GCN2-mediated cell cycle arrest adversely affects rapidly proliferating effector T-cells [225]. Studies with human T-cells and fibroblasts have identified that the sensitivity of cells to L-Trp starvation and GCN2-mediated apoptosis relates to the expression level of the protein IMPACT, an inhibitor of GCN2 kinase activity [262]. Accordingly, compared with fibroblasts, T-cells constitutively express low levels of IMPACT and, in response to low-L-Trp environments, exhibit higher levels of GCN2 activation and apoptosis; IMPACT gene overexpression of T-cells protected against L-Trp deficiency, whereas gene knockdown of IMPACT in fibroblasts rendered these cells more sensitive to low L-Trp [262]. Further studies are clearly required into how L-Trp starvation and GCN2 signalling influences cellular function and viability of other IDO1-expressing cell types and the molecular signalling mechanisms involved.

In addition to DCs and macrophages, IDO1 expression in other immune cells is capable of modulating cellular function or viability, including neutrophils and eosinophils. An immune-regulatory function has been described for IDO1 expressed in both cell types. For example, Romani et al. [159] reported that impaired IDO1 activity in lung neutrophils is linked to the excessive immune response and ensuing chronic inflammation in a mouse model of CGD, whereas IDO1-expressing human eosinophils are capable of attenuating Th1 and supporting Th2 responses, although, in this study, the involvement of IDO1 was not directly investigated with an IDO1 inhibitor and/or L-Trp supplementation [50]. IDO1 expression, L-Trp depletion and production of Kyn have also been linked to increased neutrophil apoptosis [263], which may relate to reports of elevated neutrophil tissue infiltration and responses noted in Ido1 gene-silenced or -deficient mice subject to infection in vivo [263,264].

IDO1 as an immune stimulus

In addition to regulating cellular immune responses, several studies have identified IDO1-specific CD4+ and CD8+ T-cell populations in both healthy donors and cancer patients that are capable of targeting and removing IDO1-expressing cells including IDO1-positive DCs and tumour cells [265267]. This anti-IDO1 immune response may represent a counter-regulatory mechanism aimed at quenching IDO1-mediated immune suppression in order to augment antigen-specific immune responses and inflammation. This raises the possibility that IDO1 peptides could conceivably be employed as a vaccine, where the removal of pathogenic IDO1-expressing cells is desirable (e.g. IDO1-positive tumour cells). This novel approach has recently been the subject of a Phase I clinical trial in metastatic lung cancer patients [268].

This section emphasises the remarkable capacity of IDO1 and the Kyn pathway to modulate the activation status, function, phenotype and viability of a variety of innate and adaptive immune cell types involving several different biochemical mechanisms including: (i) L-Trp deprivation and activation of the GCN2-dependent signalling pathway, (ii) production of Kyn and resultant activation of AhR signalling, (iii) synthesis of downstream Kyn pathway metabolites that are also capable of modulating immune cell actions, and (iv) IDO1 tyrosine phosphorylation, which imparts an enzyme activity-independent immune signalling action to IDO1. These effector mechanisms may act either alone or in combination to influence immune cell actions. Continued research is required to define the various modes by which IDO1 affects different immune cell types/subsets and the molecular signalling mechanisms involved under different physiological and pathological settings.

It is important to note that, although great progress has been made in delineating the mechanisms by which IDO1 induces and maintains an immunosuppressive environment, the majority of these data have been derived from studies with murine cells and mouse in vivo models. As such, further clarification of the relative importance of these various immune regulatory mechanisms to human cells is warranted. For instance, the enzyme-independent immune signalling action involving IDO1 tyrosine phosphorylation is yet to be documented in human cells. Notwithstanding, it is clearly evident that IDO1 has a broad-based role as a critical immune-regulatory enzyme during health and disease, as detailed below.

IDO1 and immune regulation in physiological and disease settings

IDO1’s role in immunoregulation has positioned it as a key player in a wide range of physiological and pathophysiological settings, as detailed below and summarised in Table 1.

Table 1
Roles of IDO1 and the Kyn pathway in different physiological and disease settings in experimental animals

NB: commonly studied experimental animal cancer models are detailed in Table 3. Cftr, cystic fibrosis transmembrane conductance regulator; ApoE, apolipoprotein E; LDL, low-density lipoprotein; LPS, lipopolysaccharide; OVA, ovalbumin; STZ, streptozotocin.

Setting Model Role of IDO 
Pregnancy Murine allogenic pregnancy Protective (1-MT [217,269]), non-essential (IDO1−/− mice [272]) 
 Murine syngenic pregnancy Not required [217
Colitis Murine trinitrobenzene sulfonic acid-induced colitis Protective [275,276,281,283
 Murine dextran sodium sulfate-induced colitis Protective [276,282
 Murine T-cell transfer colitis model Protective [282
 Citrobacter rodentium-induced colitis in mice Harmful [44
Multiple sclerosis Murine experimental autoimmune encephalomyelitis Protective [121,284,286,288291
Lupus MRLlpr/lpr mice Protective [277
Autoimmune diabetes Non-obese diabetic mice Protective [127
 STZ-induced diabetes mouse model Protective [292
Transplantation Pancreatic islet transplantation in an STZ-induced diabetes mouse model Protective [102,294,295
 Mouse kidney transplant model Protective [298
 Rat and mouse liver transplant model Protective [299], not involved [305
 Mouse corneal transplant model Protective [300,304
 Rat and mouse lung transplant models Protective [207,241
 Rat and mouse cardiac transplant model Protective [301303
 Mouse bone marrow transplant model Protective [306,307
 Mouse skin transplant model Protective [310
 Chimaeric humanised mouse model of graft arteriosclerosis Protective [347
Asthma OVA-sensitised asthmatic mouse model Protective (1-MT [55,318321]), harmful (IDO1−/− mice [246]) 
Arthritis Collagen type II-induced mouse model Protective [326329
 K/BxN mice Harmful [251,330
Chronic granulomatous disease p47phox-deficient, aspergillosis infection mouse model Protective [159
Cystic fibrosis Cftrtm1Unc (Cftr−/−) mice Protective [331
Allergy Allergic bronchopulmonary aspergillosis mouse model Protective [95
Atherosclerosis LDLr−/− mice Protective [340,341
 ApoE−/− mice Protective [343
Hypertension Mouse hypoxia-induced pulmonary hypertension Protective [54
 Rat monocrotaline-induced pulmonary hypertension Protective [54
 Angiotensin II-induced vascular constriction in rats Protective [41
 Angiotensin II-induced vascular constriction in mice Protective [54], harmful [183
 LPS-induced endotoxaemia Protective [182
Cataracts Mice overexpressing human IDO1 in the lens Harmful [591
Bone remodelling Wild-type mice Protective [592,593
Setting Model Role of IDO 
Pregnancy Murine allogenic pregnancy Protective (1-MT [217,269]), non-essential (IDO1−/− mice [272]) 
 Murine syngenic pregnancy Not required [217
Colitis Murine trinitrobenzene sulfonic acid-induced colitis Protective [275,276,281,283
 Murine dextran sodium sulfate-induced colitis Protective [276,282
 Murine T-cell transfer colitis model Protective [282
 Citrobacter rodentium-induced colitis in mice Harmful [44
Multiple sclerosis Murine experimental autoimmune encephalomyelitis Protective [121,284,286,288291
Lupus MRLlpr/lpr mice Protective [277
Autoimmune diabetes Non-obese diabetic mice Protective [127
 STZ-induced diabetes mouse model Protective [292
Transplantation Pancreatic islet transplantation in an STZ-induced diabetes mouse model Protective [102,294,295
 Mouse kidney transplant model Protective [298
 Rat and mouse liver transplant model Protective [299], not involved [305
 Mouse corneal transplant model Protective [300,304
 Rat and mouse lung transplant models Protective [207,241
 Rat and mouse cardiac transplant model Protective [301303
 Mouse bone marrow transplant model Protective [306,307
 Mouse skin transplant model Protective [310
 Chimaeric humanised mouse model of graft arteriosclerosis Protective [347
Asthma OVA-sensitised asthmatic mouse model Protective (1-MT [55,318321]), harmful (IDO1−/− mice [246]) 
Arthritis Collagen type II-induced mouse model Protective [326329
 K/BxN mice Harmful [251,330
Chronic granulomatous disease p47phox-deficient, aspergillosis infection mouse model Protective [159
Cystic fibrosis Cftrtm1Unc (Cftr−/−) mice Protective [331
Allergy Allergic bronchopulmonary aspergillosis mouse model Protective [95
Atherosclerosis LDLr−/− mice Protective [340,341
 ApoE−/− mice Protective [343
Hypertension Mouse hypoxia-induced pulmonary hypertension Protective [54
 Rat monocrotaline-induced pulmonary hypertension Protective [54
 Angiotensin II-induced vascular constriction in rats Protective [41
 Angiotensin II-induced vascular constriction in mice Protective [54], harmful [183
 LPS-induced endotoxaemia Protective [182
Cataracts Mice overexpressing human IDO1 in the lens Harmful [591
Bone remodelling Wild-type mice Protective [592,593
Pregnancy

As described above, the immunosuppressive role of IDO1 was first discovered in a mouse model of allogeneic pregnancy [217,269]. IDO1 is highly expressed in the placenta, and exposure of pregnant mice to the IDO1 inhibitor 1-MT caused the rejection of foetuses from allogeneic but not syngeneic matings in a manner that was dependent on maternally derived T-cells and complement activation [217,269]. Induction of IDO1 by soluble CTLA-4-Ig in abortion-prone mice improves the pregnancy outcome [270], suggesting that IDO1 expression at the foetal/maternal interface acts to protect the immunologically foreign foetus from maternal immune attack. Accordingly, in humans, active IDO1 is variably expressed during several stages of pregnancy in placental syncytiotrophoblasts, extravillious cytotrophoblasts, macrophages in the villous stroma, vascular endothelium and foetal membranes [48,271]. Notably, however, matings between allogeneic IDO1−/− mice produce offspring at the normal rate and of normal litter size relative to wild-type mice, demonstrating that in mice at least, there appear to be compensatory mechanisms that act in lieu of IDO1 to protect the allogeneic foetus from the maternal immune system [272]. Notwithstanding, pregnant IDO1−/− mice have recently been found to exhibit some signs that are typical of pre-eclampsia, including renal pathology, endothelial dysfunction and foetal growth restriction, but no significant change in blood pressure [273].

Autoimmune diseases

Ido1 gene-knockout mice do not develop spontaneous autoimmune disorders, indicating that the enzyme's immunosuppressive action is not essential for the maintenance of self-tolerance under normal conditions. However, numerous studies with Ido1 gene-deficient mice or IDO1 inhibitors reveal that the enzyme plays an important role in the establishment of acquired immune tolerance upon exposure to new or foreign antigens. Thus, excessive and unregulated immune activation against self-antigens in several chronic inflammatory and autoimmune diseases (e.g. intestinal colitis, primary biliary cirrhosis, uveoretinitis, encephalitis, lupus or Type 1 diabetes) can be dampened by the induction of IDO1 expression and resultant immune suppression [274277]. Below, we discuss certain disease states in more detail.

Inflammatory bowel disease (IBD)

IBD refers to several chronic inflammatory disease states of the colon and small intestine, and includes ulcerative colitis and Crohn's disease. Several studies report that intestinal IDO1 expression and activity are elevated in Crohn's disease and ulcerative colitis patients, with IDO1 detected in epithelial and mononuclear cells [278280]. In murine experimental colitis models, IDO1 expression in intestinal crypt epithelial cells and infiltrating APCs within the lamina propria appear to be crucial for limiting the severity of colitis in mice via Treg-mediated immune suppression [275,276,281283]. For example, pharmacological inhibition of IDO1 with 1-MT or Ido1 gene deficiency enhanced the severity of colitis in mice exposed to TNBS (trinitrobenzene sulfonic acid), which correlated with reduced numbers of Tregs [281,283]. Moreover, inducing IDO1 with a TLR9 agonist [276] or an anti-CTLA-4 monoclonal antibody [275] reduced intestinal inflammation and colitis severity that correlated with increased numbers of Tregs. Of the IDO1-expressing APCs important for protection against colitis, mucosal CD103+ DCs have been identified to play a key role in signalling for elevated Treg responses and reduced inflammatory Th1 and Th17 responses [282].

Interestingly, Ido1 gene deficiency protects against colitis induced by the enteric pathogen Citrobacter rodentium [44]. This apparent contradictory finding is thought to relate to the physiological function of constitutively expressed IDO1 within the small intestine and mesenteric lymph node in governing intestinal immunity through the suppression or maintenance of B-cell responses towards the commensal microflora of the gut, i.e. Ido1 gene deficiency in mice affords an exaggerated B-cell response and elevated levels of natural secretory IgA antibodies that act to recognise and control C. rodentium colonisation, thereby protecting against colitis [44]. Therefore, the apparent discrepant actions of IDO1 in the different models of colitis are in keeping with the paradigm of IDO1 expressed within the gastrointestinal tract being immunosuppressive towards T- and B-cells.

Experimental autoimmune encephalomyelitis (EAE)

EAE, a murine model of multiple sclerosis, is characterised by elevated IDO1 activity during the pre-clinical or remission phases of the disease within the spleen and increases in the CNS with the onset of symptoms, where the enzyme is detected in activated macrophages and microglia [284,285]. 1-MT treatment during EAE exacerbates clinical symptoms and inflammation, and mice are more susceptible to early relapses [284,285]. Similarly, Ido1 gene deficiency augments experimental EAE, characterised by elevated Th1 and Th17 cell responses and reduced Treg responses [286]. Also, inhibition of the negative IDO1 regulator DAP12 promotes the resistance of mice towards EAE in an IDO1-dependent manner [121]. The protective actions of IDO1 during EAE appear to relate to the activation of the GCN2 pathway and the production of Kyn pathway metabolites. Thus, unlike wild-type mice, GCN2-deficient mice did not enter a remission phase, which correlated with elevated CNS inflammation, increased Th1/Th17 cells and reduced IDO1 expression and Treg/pDC frequency [287]. Also, administration of Kyn pathway metabolites (picolinic acid, QA, 3-HAA and 3-HK) or induction of IDO1 in DCs skews the EAE Th1 immune response towards a Th2/Treg response, reduces APC activation, induces cell cycle arrest of myelin-specific CD4+ T-cells and encourages myelin repair, resulting in milder relapses and substantially reduced disease severity [286,288,289]. Interestingly, the ability of disparate treatments, including pluripotent stem cells [288], DNA nanoparticles, cyclic dinucleotides (that activate the STING adaptor pathway) [290] and an oestrogen receptor agonist [291], to induce dominant Treg responses and protect against experimental EAE relates to their shared ability to induce IDO1.

Lupus

In lupus-prone MRLlpr/lpr mice, IDO1 expression is abnormally elevated in the spleen and IDO1 inhibition with 1-D-MT accelerated the progression of this autoimmune disease [277]. Moreover, Ido1 gene deficiency supports the development of a lupus-like pathology upon chronic exposure of mice to apoptotic cells, i.e. elevated serum autoreactivity to dsDNA, kidney pathology and increased mortality [277]. This study emphasises the importance of IDO1 as a key mechanism for immune tolerance towards self-antigens expressed by apoptotic cells [277].

Diabetes

Type 1 diabetes is characterised by the autoimmune destruction of the pancreatic islet cells, critical for the production of insulin and glucose homoeostasis. Grohmann et al. [127] reported in NOD (non-obese diabetic) mice (a common mouse model of autoimmune diabetes) that the inability of IFNγ and CD8α+ DCs to induce tolerance in pre-diabetic mice is related to impaired IFNγ-induced IDO1 expression within DCs involving the inhibition of STAT1 signalling by the oxidant peroxynitrite, the reaction product of NO and O2•− [127] (Figure 5). Accordingly, treatment of NOD mice with a NOS inhibitor and antioxidant scavengers of O2•− or peroxynitrite restored IFNγ-induced IDO1-catalysed L-Trp metabolism and tolerance in NOD mouse DCs [110,127]. Similarly, soluble CTLA-4 treatment programmes immune-tolerant properties in DCs from pre-diabetic NOD mice by inducing autocrine IFNγ production and resultant up-regulation of IDO1-mediated immune suppression [110]. Interestingly, CTLA-4 reportedly protects against peroxynitrite-mediated inhibition of STAT1 signalling through the concomitant activation of the FOXO3a transcription factor and the induction of SOD expression, an enzyme capable of scavenging the O2•− necessary for peroxynitrite production [110]. In the STZ (streptozotocin)-induced model of autoimmune diabetes, increased IDO1 expression was apparent within splenic pDCs and pancreatic draining lymph nodes in diabetic mice and IDO1 inhibition with 1-MT augmented disease severity [292]. Also, activation of TLR9 signalling via the administration of CpG-ODNs induced an IDO1-dependent elevation of Tregs within lymph nodes and the mitigation of the pro-inflammatory milieu within the pancreas of mice rendered diabetic with STZ [292]. Together, the above studies suggest that IDO1 expression and activity are impaired in DCs during Type 1 diabetes and that correcting this impairment affords IDO1-mediated immune suppression and protection against the disease. Accordingly, IDO1 inducers represent a novel immunotherapeutic class of anti-diabetic drugs, the clinical potential of which are yet-to-be determined.

Although the preponderance of mouse studies compared with human studies on IDO1 in autoimmune disease is significant, overall these studies do emphasise that the chronic inflammatory milieu and tissue damage apparent during various autoimmune disorders can be substantially reduced by the immunosuppressive and anti-inflammatory actions of the IDO1 pathway, which acts to dampen the breakdown of peripheral self-tolerance and maintain systemic tolerance to autoantigens. It is important to test whether this paradigm is also relevant to human disease such that the potential therapeutic benefits of enhancing IDO1 expression can be explored further clinically.

Transplantation

The ultimate goal in organ transplantation is the development of donor antigen-specific tolerance, thereby avoiding the use of systemic immunosuppressive drugs that can expose individuals to opportunistic infection. In this regard, the induction of immunological tolerance by IDO1 has made it an increasingly attractive molecule to study. Indeed, numerous experimental studies indicate that IDO1 can create a state of systemic immune tolerance towards alloantigens during tissue transplantation in the absence of other modes of immune suppression.

Although there is currently no cure for Type 1 diabetes, transplantation of human islets into diabetic patients has been trialled, although graft tolerance currently cannot be accomplished without long-term immune suppression [293]. In NOD mice, overexpression of active IDO1 prolongs the survival of transplanted pancreatic islets in recipient diabetic mice in the absence of immunosuppressive drugs [102,294,295]. This was attributed to several different immune factors, including reduced T-cell infiltration into the graft, a shift from a pro-inflammatory Th1 to an anti-inflammatory Th2 response and impaired T-cell priming, although the IDO1-expressing islets did eventually succumb to host immune attack [295,296]. In an effort to further prolong allogenic islet acceptance, Hosseini-Tabatabaei et al. [297] recently developed a stable biomimetic scaffold (to re-establish the extracellular matrix that is lost during islet isolation) housing IDO1-expressing fibroblasts to support islet implantation and survival in mice. This method lengthened the survival of islet allografts 4-fold by enhancing Treg infiltration into the graft itself and the draining lymph nodes [297]. Importantly, insulin staining determined that these islets maintained their functionality.

Spontaneous renal allograft acceptance in mice relies on IDO1-expressing cells (in this case, DC and Tregs) during the later stages to mediate tolerance [298]. Importantly, ‘true’ immunological tolerance develops since ∼30% of the cohorts are capable of accepting donor-matched skin allografts. Furthermore, liver allografts, which tend to be well-tolerated in mice, are rejected following 1-MT treatment [299]. Providing further support for a beneficial role of IDO1 and the Kyn pathway, overexpression of IDO1 or administration of 3-HK or 3-HAA in animal models of allogeneic corneal, lung or cardiac transplants are able to prolong graft survival, attenuate allograft injury and are associated with decreased antigen-specific and non-specific T-cell responses [207,241,300304], although IDO1 overexpression in liver allografts did not prevent ultimate transplant rejection in mice [305].

A potential complication of allogeneic transplantation is GVHD (graft-versus-host disease), which occurs when immunocompetent leucocytes from the donor graft recognise the host as foreign and mount an immune response against the recipient. IDO1−/− mice receiving transplants have aggravated GVHD as a result of enhanced T-cell proliferation and reduced apoptosis, culminating in the increased mortality of IDO1−/− mice in a manner reversed by the administration of Kyn, 3-HAA and 3-HK [306,307]. Furthermore, although allogeneic haematopoietic stem cell transplantation can be used to treat chemorefractory leukaemia (graft-versus-leukaemia activity), it is complicated by the accompanying GVHD. In mice, this GVHD can be minimised by enhancing the IDO1-mediated cross-talk between IFNγ-producing donor T-cells and IDO1-expressing donor pDCs, resulting in the generation of greater numbers of donor Tregs relative to inflammatory T-cells [308]. Interestingly, the ability of HDAC inhibitors to protect against experimental GVHD in mice is linked to their ability to up-regulate the expression of IDO1 in DCs and the resultant suppression of DC-stimulated responses [65]. Use of the HDAC inhibitor vorinostat in patients receiving allo-haematopoietic cell transplants was similarly beneficial. Thus, compared with patients receiving standard-of-care treatment, vorinostat-treated patients had reduced acute GVHD and did not exhibit an increase in relapse, outcomes that correlated with IDO1 induction [309]. IDO1 also appears to contribute to the protective actions of cyclosporin A, an inhibitor of calcineurin and a commonly employed anti-transplant rejection drug in humans. Thus, cyclosporin A treatment of myeloid-derived suppressor cells induced IDO1 expression, which correlated with the ability of these cells to suppress rejection of allogeneic skin transplants in mice [310]. IDO1 inhibition with 1-MT abrogated the suppressive actions of cyclosporin A-treated myeloid-derived suppressor cells [310].

Although the above findings indicate a key role for IDO1 in regulating transplantation immunology in rodents, the relative functional importance of IDO1 in organ transplantation in humans is more difficult to investigate and is hence currently less clear. In patients with renal allografts, increased IDO1 activation correlates with episodes of acute rejection and chronic transplant dysfunction compared with controls [311,312]. Elevated IDO1 activity, as determined by the measurement of the Kyn/L-Trp ratio in serum, has accordingly been posited as a non-invasive early marker of eventual transplant rejection as an alternative to biopsies since elevated IDO1 activity can be detected as early as 1 day post-transplantation [311,313]. However, rather than playing a direct role in rejection of the organ transplant, enhanced IDO1 activity in lung allograft recipients is thought to merely represent a marker of the chronic inflammation associated with organ transplantation [314].

Whether the protective immunosuppressive activity of IDO1 in murine and rat transplant models is also relevant to humans remains to be more fully explored. The association of high IDO1 activation with graft rejection suggests that the general pro-inflammatory environment induced by the transplant may instead overwhelm the immunosuppressive action of IDO1. Nevertheless, carefully timed therapeutic induction of IDO1 may provide an additional approach to enhance graft survival.

A further aspect of IDO1 and its potential protective actions against autoimmune diseases and allograft rejection is the enzyme's contribution to the immunosuppressive activity of mesenchymal stem cells, which are increasingly recognised as a novel mode of cell-based immune therapy [315317]. In addition to their potential in tissue repair and regeneration, mesenchymal stem cells are multipotent immunoprivileged progenitor cells derived from the bone marrow and different organs and tissues, capable of modulating a variety of immune cell (T-cells, B-cells, NK cells and DCs) responses to overall yield anti-inflammatory and immunosuppressive effects. Various reports identify IDO1 as one of a host of immunosuppressive effector molecules expressed by mesenchymal stem cells that also include PGE2, IFNγ, TGFβ, IL-10 and NO [315317]. Current research is focusing on the relative importance or contribution of IDO1 and other immune-modulatory mechanisms and their potential interactions in determining the beneficial actions of mesenchymal stem cells, which have been the subject of numerous pre-clinical animal studies and current human clinical trials [317].

Chronic inflammatory and allergic disorders

Chronic inflammation underlies the pathogenesis of a variety of clinically important disease states. Considerable evidence indicates a modulatory role for IDO1 in a variety of inflammatory disorders including allergic asthma, arthritis, CGD, cystic fibrosis and atherosclerosis.

Asthma

Asthma is an allergen-induced chronic inflammatory disease of the airways involving a prominent pathogenic role for Th2 cells and cytokines produced by these cells (e.g. IL-4, IL-5, IL-9 and IL-13). Several studies employing OVA-sensitised OVA-challenged BALB/c mouse models of asthma indicate a role for IDO1 in suppressing Th2 responses and disease severity. An initial study by Hayashi et al. [55] showed that the administration of the TLR9 ligand CpG-ODNs by various routes induced IDO1 within resident lung epithelial cells, resulting in the suppression of both Th2- and Th1-dependent inflammatory responses and airway hyper-reactivity; 1-MT administration reversed the protective actions of the TLR9 ligands. Subsequent studies examining the mechanisms by which splenic CD8α+ DCs or IL-10-differentiated DCs induce Th2 tolerance and reduced airway hyper-responsiveness in vivo in OVA-challenged asthmatic mice also reported a key role for IDO1 [318,319]. Studies with 1-MT provided further evidence that IDO1 contributes to the beneficial action of allergen immunotherapy in reducing airway eosinophilia and Th2 responses [320]. Kyn pathway metabolites also exhibit Th2 tolerogenic and protective actions in murine asthma. Thus, intratracheal instillation of 3-HAA ameliorates experimental asthma in OVA-challenged mice by impairing the functionality of, and inducing apoptosis in, Th2 cells via inhibiting NF-κB signalling within these cells [236]. In a murine model of mucosal tolerance, although IDO1 inhibition with 1-MT attenuated OVA-induced tolerance upon OVA-sensitisation and rechallenge, treatment with a mixture of Kyn and 3-HAA restored tolerance in 1-MT-treated mice [321]. Similarly, administration of Kyn, 3-HK or xanthurenic acid reduced lung eosinophil and Th2 cytokine levels in allergic mice receiving suboptimal allergen immunotherapy [320].

Despite the above studies indicating a protective action of IDO1 and Kyn metabolites against experimental asthma involving the inhibition of Th2-mediated lung inflammation, a study by Xu et al. [246] employing an OVA-induced chronic asthma model reported that Ido1 gene deficiency attenuated Th2-dependent allergic airway inflammation and hyper-responsiveness and indicated that IDO-expressing DCs in the murine lung can enhance antigen-driven Th2 responses due to the involvement of IDO1 in lung DC maturation. The reasons for the apparent discrepancy with respect to IDO1’s role in controlling Th2 responses during experimental asthma are not known, but may relate to differences in the actions of 1-MT compared with IDO1−/− mice. Further studies are therefore warranted to understand the in vivo role of IDO1 with respect to the control of Th2 immunity and relevance to allergic responses, particularly in humans.

Arthritis

Rheumatoid arthritis is an autoimmune and systemic chronic inflammatory disease that manifests within synovial tissue. Increased activation of the Kyn pathway is a feature of patients with rheumatoid arthritis [322]. Although DCs obtained from the synovial joints of arthritis patients express functional IDO1, autologous or allogeneic synovial fluid T-cells are resistant to IDO1-mediated inhibition of proliferation compared with T-cells derived from healthy donors [323]. This was attributed to the enhanced expression of TTS in T-cells from the arthritis patients, presumably making the cells refractory towards IDO1-mediated L-Trp metabolism and hence capable of promoting inflammation and disease. Similar findings have been made in patients with the autoimmune disorder Graves’ disease, where, although IDO1 expression is elevated in B-cells or DCs, CD4+ T-cells from Graves’ disease patients compared with healthy controls showed elevated TTS expression and resistance to IDO1-mediated inhibition [324]. TTS expression is also elevated in CD4+ and CD8+ T-cells from patients with immune thrombocytopenia relative to cells from healthy donors [325]. Therefore, the expression level of TTS in T-cells is an important determinant of IDO's immunosuppressive actions in certain autoimmune disorders.

A functional role for IDO1 in arthritis has been investigated in two murine models of the disease, with the first involving the injection of collagen type II into susceptible DAB1 mice, which induces CD4+ T-cell-dependent synovial inflammation, and the second, employing K/BxN mice that spontaneously develop the disease. In mice subjected to collagen-induced arthritis, IDO1 inhibition with 1-MT treatment further exacerbates the disease involving a shift to a pro-inflammatory Th1 response [326], a finding more recently replicated in IDO1−/− mice [327]. Accordingly, administration of an agonistic anti-4-1BB antibody or direct transfer of IDO1-competent DCs into arthritic mice decreases inflammation, resulting in less severe arthritis as a result of IDO1 induction in macrophages and DCs [328,329]. These studies support a protective, immunosuppressive and anti-inflammatory role for IDO1 in murine arthritis. Conversely, however, blockade of IDO1 with 1-MT and anti-CD20 treatment to deplete B-cells in spontaneously arthritic K/BxN mice reduced autoantibody titres, the degree of inflammation and disease severity [251,330]. In this case, IDO1 acted independently of T-cell responses to instead stimulate disease by promoting autoreactive B-cell responses [251,330].

The above studies investigating the roles of IDO1 in different mouse models of asthma and arthritis illustrate that while the general view of IDO1 as an immunosuppressive anti-inflammatory enzyme that is protective during Th1-promoted inflammatory disorders holds true, its actions in chronic inflammatory disease are complex. Thus, under certain in vivo conditions, IDO1 is immune-stimulatory, as exhibited by its role as a stimulus for autoreactive B-cells in arthritis [251,330] and Th2 cells during asthma [246]. The reasons for the varying immunomodulatory actions of IDO1 under different disease contexts in vivo are an area of importance requiring further attention and clarification, particularly the extent to which these varied actions can also translate to human patients suffering equivalent disease conditions.

CGD and cystic fibrosis

As mentioned above, CGD involves a defect in the phagocyte NOX2 enzyme. Thus, sufferers are unable to generate an oxidative burst required to defend against pathogens. This results in an exaggerated chronic inflammatory response towards infection and the formation of tissue granulomas. In the lungs of CGD mice deficient in the critical NOX regulatory subunit p47phox, the expression of inactive IDO1 was reported due to the absence of NOX-derived O2•− necessary to support enzyme activity in innate immune cells isolated from the lungs of CGD mice [159]. The expression of inactive IDO1 protein correlated with the loss of IDO1-mediated Treg-dependent immune suppression in Aspergillus fumigatus-infected CGD mice and led to pathogenic increases in IL-17 production, causing chronic lung injury [159]. Combination treatment with IFNγ and Kyn, thereby bypassing the need for IDO1 to generate Kyn metabolites, rescued the CGD mice from this immunopathogenic hyperinflammatory phenotype [159]. Nevertheless, the extent to which NOX-derived O2•− is actually required for active IDO1 in vivo has been recently questioned, with another study using both p47phox-deficient mice, as well as mice with impaired NOX2, reporting that lung IDO1 activity is not decreased in response to A. fumigatus infection [172]. The reasons for this discrepancy on the level of lung IDO1 activity in A. fumigatus-infected mice, its interrelationship with NOX-derived O2•− and impact on CGD are important issues that require further clarification. Despite this controversy, the use of 1-MT and the administration of Kyn to wild-type and CGD mice in the initial report support a key role for IDO1 and the Kyn pathway in this disease [159].

Cystic fibrosis is an inheritable genetic disorder involving mutations in the gene encoding the CFTR (cystic fibrosis transmembrane conductance regulator) protein resulting in excessive build-up of mucus within the lung and increased susceptibility to lung infection. In a mouse model of cystic fibrosis, an imbalance of Th17 and Tregs due to a lack of IDO1 expression and activity is linked to the exaggerated immune response towards A. fumigatus [331]. Thus, bronchial epithelial cells derived from cystic fibrosis patients display lower levels of IDO1 protein compared with control patients both before and after IFNγ or poly(I:C) stimulation or A. fumigatus infection [331]. Importantly, in vivo administration of a mixture of Kyn metabolites (Kyn, 3-HK and 3-HAA) to cystic fibrosis mice reversed their amplified and immunopathogenic response towards A. fumigatus infection. Similarly, dexamethasone-stimulated IDO1 expression also inhibited Th2 cells, as well as increasing Tregs, to protect against allergic bronchopulmonary aspergillosis [95].

Collectively, the studies on CGD and cystic fibrosis mice show that defective IDO1 expression/activity and loss of the enzyme's immunosuppressive activity can lead to damaging inflammatory disorders. The studies also highlight the therapeutic potential of inducing IDO1 (e.g. with glucocorticoids) and administration of immunosuppressive Kyn pathway metabolites.

Atherosclerosis

Atherosclerosis is characterised by the formation of focal lesions within the intima of large and mid-sized arteries, which consist of lipids, fibrotic tissue and inflammatory/immune cells and cause narrowing of the blood lumen. It represents a chronic inflammatory disease of the arterial intima that is orchestrated by cells of both the innate and adaptive immune systems [332,333]. Atherosclerotic lesion rupture and thrombosis leads to stenosis or blockage of blood vessels that can clinically manifest as myocardial infarction or stroke. Increasing evidence indicates a role for IDO1 in regulating immune responses during atherosclerosis and clinical outcomes of the disease.

A variety of clinical studies measuring the ratio of Kyn to L-Trp as an index of in vivo IDO1 activity report a correlation of IDO1 activation and resultant L-Trp metabolism with increased disease severity and clinical cardiovascular risk. For example, the plasma Kyn/L-Trp ratio is increased in coronary artery disease patients compared with age-matched healthy controls [334], with the extent of the increase reported to correlate positively with certain cardiovascular disease risk factors for atherosclerosis in young female adults [335] or in an elderly population with more advanced atherosclerosis [336]. Notably, in the latter study, the Kyn/L-Trp ratio as a measure of IDO1 activity did not represent an independent marker of disease severity. A more recent study reported that an elevated Kyn/L-Trp ratio is also associated with increased risk of acute coronary events in elderly patients without prior coronary heart disease [337]. An elevated Kyn/L-Trp ratio and increased levels of kynurenic acid and 3-HAA in the plasma are also detected in patients resuscitated from cardiac arrest, with the extent of increase in Kyn pathway metabolites correlating with initial hypotension and representing independent predictors of early death and poor longer-term outcome [338]. In human atherosclerotic lesions, IDO1 expression is significantly increased in the atheromatous core and CD68+ monocyte-derived macrophages, with no staining apparent in endothelial cells [336]. Although the above studies indicate that activation of IDO1 and the Kyn pathway correlates with coronary artery disease, whether IDO1 plays a role in human disease or represents a marker of immune activation is unknown. Studies employing murine models of atherogenesis, however, indicate an immunosuppressive and protective role of IDO1 when expressed in APCs.

Experimental evidence for a potential role for IDO1 in atherosclerosis was provided by a study examining the role of pDCs in atherosclerosis-prone LDLr−/− (low-density lipoprotein receptor) mice fed on a high-fat diet. This study reported that, although pDCs are scarce within human or murine atherosclerotic lesions, selective depletion of these cells promoted both lesion formation and the accumulation of T-cells within the lesion [339]. Importantly, pDCs isolated from atherosclerotic mice fed on a high-fat diet expressed elevated levels of IDO1 that was responsible for the greater ex vivo T-cell-suppressive activity of these pDCs relative to those isolated from mice fed on a normal chow diet, findings consistent with an immunosuppressive role of IDO1 expressed in pDCs during murine atherogenesis [339]. More direct experimental evidence for an atheroprotective action of DC-expressed IDO1 was provided by a study reporting that treatment of LDLr−/− mice with the long-chain ω−3 fatty acid eicosapentaenoic acid induced the regression of atherosclerotic plaques [340]. This correlated with increased numbers of immature DCs, increased IDO1 expression within DCs and reduced proliferation of the pro-atherogenic CD4+ T-cell subset [340]. Notably, administration of the IDO1 inhibitor 1-MT attenuated the anti-atherogenic actions of eicosapentaenoic acid, indicative of a key role for IDO1 in facilitating atherosclerotic lesion regression induced by ω−3 fatty acids [340].

In a study examining the relationship between the Kyn pathway and atherogenesis, administration of 3-HAA reduced atherosclerotic lesion formation in LDLr−/− mice, purportedly by inhibiting the uptake of oxidised LDL (low-density lipoprotein) by lesion macrophages, decreasing T-cell-mediated inflammation and altering the atherogenic lipid profile of LDLr−/− mice by reducing plasma cholesterol [341]. Studies in LDLr−/−/IDO1−/− mice indicate that Ido1 gene deficiency results in increases in serum triacylglycerols (but not cholesterol) when these mice are fed on a high-fat diet [342]. How Ido1 gene deficiency affected the development of atherosclerotic lesions in LDLr−/− mice was not, however, detailed. A most recent study reported that inhibition of IDO1 in ApoE−/− (apolipoprotein E) mice by providing 1-DL-MT in the drinking water promoted vascular inflammation and atherosclerotic lesion formation and that the pro-atherogenic effect of IDO1 inhibition was reversed by exogenous 3-HAA administration [343].

Oxidative modification of LDL is considered an important atherogenic event leading to the formation of macrophage-derived foam cells, the hallmark of early atherosclerotic lesions [344]. Activation of IDO1 and the resultant synthesis and extracellular release of 3-HAA underlies the ability of IFNγ to inhibit LDL oxidation by human monocyte-derived macrophages in vitro [168]. Our subsequent studies reported that 3-HAA is an efficient antioxidant capable of inhibiting LDL oxidation induced by a variety of oxidants, involving the ability of 3-HAA to scavenge oxidants (e.g. peroxynitrite) or act as a co-antioxidant, referring to its ability to efficiently reduce the vitamin E radical, ensuring the reduced and antioxidant form of vitamin E is maintained within LDL [345,346].

In atherosclerosis, the intima and adventitia, but not the media, are infiltrated by macrophages and T-cells. In a mouse model of graft arteriosclerosis, the immunoprivileged state of the media has been attributed to the IFNγ-induced expression of IDO1 within medial vascular SMCs, which inhibit T-cell proliferation and activation at this site [347]. Whether IDO1 expression in SMCs or endothelial cells can regulate T-cell responses within atherosclerotic lesions and the implications for the disease remain unclear.

It is clear from the above that, although complex, in general, IDO1-mediated immune suppression serves to protect against a wide range of immunological and inflammatory disorders (Table 1). Accordingly, augmenting the immunosuppressing actions of the IDO1 pathway, either by inducing IDO1 expression and activity or through the direct administration of Kyn metabolites, has proven an effective therapeutic approach in treating experimental animal models of IBD [275,276], diabetes [292], EAE [286,288,289], asthma [55], CGD [159], allergy [320] and atherosclerosis [340]. Indeed, the protective nature of certain stimuli or drugs are often attributed to their ability to induce IDO1 (e.g. treatment of asthma [55] or diabetes [292] with CpG-ODNs, allergy with dexamethasone [95] or atherogenesis with eicosapentaenoic acid [340]). In certain diseases such as CGD [159] and diabetes [127,292], where disease severity is linked to impaired IDO1-mediated immune suppression, exogenous administration of Kyn metabolites represents a useful strategy to bypass the need for IDO1 and provide disease-protecting immunosuppression. In vitro evidence indicates that the efficacy of T-cell suppression by Kyn metabolites is, however, further enhanced in low-L-Trp environments [229,230,232]. Therefore, the induction of IDO1-mediated L-Trp metabolism in combination with the provision of exogenous Kyn pathway metabolites may yield maximal effectiveness in alleviating certain autoimmune or inflammatory disease states. Although inducing IDO1 and/or treatment with Kyn metabolites has proven effective in combating animal models of numerous clinically relevant diseases, the extent to which these approaches translate to the treatment of the corresponding human diseases or the safety surrounding these approaches is currently unknown.

Although bolstering IDO1’s immunoregulatory actions is potentially useful in combating certain autoimmune, allergic or inflammatory disorders (Table 1), the inhibition of IDO1 represents a novel immunotherapeutic approach in the treatment of certain cancers, as outlined below. Indeed, inhibition of IDO1-mediated immune suppression towards tumour antigens has proven beneficial in several experimental animal cancer models, which have prompted several clinical trials testing the actions of small-molecule IDO1 enyme inhibitors in cancer patients.

Cancer

Although the immune system is designed to recognise and remove immunogenic ‘foreign’ cancer cells, tumours can evolve mechanisms that allow them to acquire an abnormal state of immune tolerance, providing a survival advantage aiding cancer survival, progression and malignancy [348]. IDO1 is increasingly recognised as a key mechanism employed by tumour cells to avoid clearance by the host immune response.

Early in vitro studies investigating the role of IDO1 in cancer suggested that the enzyme represented an anti-tumour effector mechanism of IFNγ [349,350]. For example, IFNγ-mediated induction of IDO1 in cultured human cancer cell lines, such as KB oral carcinoma cells, A549 lung carcinoma cells and ME180 lung carcinoma cells, inhibited the proliferation of these cells via L-Trp starvation [351,352]. More recent studies indicate that when human mesenchymal stem cells, which normally promote the growth of follicular lymphoma B-cells, are stimulated with IFNγ, the stem cells induce an IDO1-dependent growth arrest and apoptosis of the follicular lymphoma B-cells in vitro [353]. Also, Fas ligand or TNFα treatment promotes the apoptosis of human tumour cells in an IDO1-dependent manner involving the increased production of ROS [354,355]. Despite these in vitro findings, as detailed below, a significant body of literature indicates that elevated IDO1 is a feature of human and experimental animal tumours with the immunosuppressive action of the enzyme representing an important mechanism involved in tumour cell immune escape.

IDO1 expression in human cancer

Elevated expression of IDO1 is a characteristic of a wide variety of human cancers. Initial surveys reported that a majority of solid human tumours examined (e.g. prostatic, colorectal and pancreatic carcinomas) expressed high concentrations of active IDO1, within the tumour cells themselves, tumour-recruited leucocytes (including macrophages and DCs) or DCs in TDLNs (tumour-draining lymph nodes) [49,158,356]. Numerous studies measuring IDO1 mRNA or protein expression in tumours and/or TDLNs or serum Kyn and L-Trp levels have reported elevated IDO1 in a wide variety of different human cancers (Table 2). Notably, the extent of IDO1 activation frequently correlates with poor prognosis in humans (e.g. increased invasiveness and frequency of metastasis, more advanced clinical staging, and decreased remission rates of disease-free survival, progression-free survival and overall survival) in a variety of malignancies, including solid tumours (ovarian, cervical, breast, endometrial, colon, gastric, lung, melanoma, pancreatic, hepatocellular, colorectal, osteosarcoma, glioma and oesophageal) [357372] and various blood cancers {AML (acute myeloid leukaemia) [373,374], myeloma [375], adult T-cell leukaemia/lymphoma [376] and other lymphomas [377380] (see Table 2 and references therein)}. Elevated IDO1 expression has also been reported to correlate with increased numbers of Tregs in human patients, either within the tumour environment [361,362,366,379382], as noted in human melanoma [383] or in the circulation, as apparent in AML [384]. With respect to AML, Curti et al. [384] reported that elevated IDO1 in AML patients at diagnosis correlated with increased circulating levels of CD4+ CD25+ Foxp3+ Tregs. In vitro and in vivo studies confirmed that IDO1-expressing leukaemia cells increase the number of Tregs in a manner inhibited by 1-MT, providing experimental evidence that IDO1-expressing AML cells can promote tolerance by signalling for a Treg phenotype [384386].

Table 2
Relationship between IDO1 expression in human cancer and disease prognosis
Cancer IDO1 expression and/or activity Prognosis and/or survival Reference(s) 
Acute myeloid leukaemia Varying IDO1 mRNA in AML blasts Higher IDO1 expression correlates with shorter overall survival and relapse-free survival [373
 Increased serum Kyn/L-Trp ratio Increased IDO1 activity correlates with decreased survival [374
Breast cancer Various IDO1 protein levels in stromal cells of all tumours IDO1 expression correlates with late-stage disease and reduced disease-specific and metastasis-free survival [360
 IDO1 mRNA and protein in tumours and lymph nodes IDO1 expression correlates with more advanced clinical staging and more extensive lymph node metastasis [362
 IDO1 protein is expressed in some tumours IDO1 expression correlates with high risk of lymph node metastasis [361
 IDO1 mRNA is expressed in all tumours High IDO1 mRNA expression is associated with medullary features and is an independent prognostic factor for improved survival in basal-like breast carcinoma [388
Cervical cancer IDO1 protein is expressed in some cancers IDO1 correlates with clinical stage, lymph node metastasis and lymph-vascular space invasion. Diffuse expression, but not focal expression at the tumour front, independently predicts impaired overall survival and disease-free survival [359
Colorectal cancer IDO1 protein is expressed in most tumours at the tumour front and/or the tumour centre IDO1 expression at the tumour front is an independent prognostic factor for decreased overall survival and the development of metachronous metastases in pT1-4N1Mx-staged colorectal cancer [594
 IDO1 protein is expressed in some tumours Increased IDO1 expression correlates with the frequency of liver metastasis and poorer overall survival [369
 Decreased serum L-Trp Decreased serum L-Trp correlates with liver metastases and reduced quality of life [595
Cutaneous melanoma IDO1 protein in all tumours and most lymph nodes Increased IDO1 correlates with cancer progression and metastasis [596
Diffuse large B-cell lymphoma Increased serum Kyn Kyn is an independent prognostic factor for reduced overall survival in patients with R-CHOP (rituximab/cyclophosphamide/hydoxydaunorubicin/oncovin/prednisolone) therapy [377
 IDO1 protein in some tumours Increased expression correlates with lower remission rates and lower survival rate [378
Endometrial cancer IDO1 protein in some tumours High IDO1 is an independent prognostic factor for progression-free survival or reduced disease-specific survival. Correlates with myometrial invasion, lymph-vascular space involvement, lymph node metastasis and poorer overall survival [363,597,598
Gastric cancer IDO1 protein is expressed in all tumours Higher IDO1 expression is associated with deeper invasion and more frequent lymph node metastasis [599
Glioma IDO1 mRNA and protein in all tumours High IDO1 expression is associated with malignancy and lower survival rates [371,600
Hepatocellular carcinoma IDO1 mRNA and protein in the tumour-infiltrating cells of some tumours IDO1-positive tumours associate with higher rates of recurrence-free survival relative to IDO1-negative tumours [389
 IDO1 protein in some tumours IDO1 expression is an independent prognostic factor for reduced overall survival and correlates with metastasis and a lower 5-year survival rate. No difference in recurrence rate was observed [368
Hodgkin's lymphoma IDO1 protein in histiocytes, DCs and endothelial cells IDO1 associates with later-stage disease, Epstein–Barr virus infection, poorer prognosis and inferior survival [380
Laryngeal squamous cell carcinoma IDO1 protein in all tumours. Higher IDO1 expression is an independent predictor of decreased overall survival and disease-free survival [601
Lung cancer IDO1 protein in peritumoral stroma eosinophils Higher IDO1 expression correlates with lower overall survival [365
 Increased serum Kyn/L-Trp ratio in serum Higher IDO1 activity associates with more advanced disease [364
Melanoma IDO1 protein in some sentinel lymph nodes Higher expression levels correlate with a worse long-term outcome [158
 IDO1 protein in sentinel lymph nodes with or without malignancy IDO1 expression does not correlate with malignancy [387
 IDO1 protein is expressed in the sentinel lymph nodes High IDO1 expression is an independent prognostic factor for lower overall survival and progression-free survival, particularly in patients with uninvolved lymph nodes [366
 IDO1 protein in some tumours cells, peritumoral endothelial cells and sentinel lymph node cells Peritumoral endothelial IDO1 and lymph node IDO1 expression is a negative prognostic factor for overall survival. Lymph node IDO1 expression is also a negative prognostic factor for relapse-free survival [407
Multiple myeloma Increased serum Kyn Increased Kyn correlates with late-stage disease [375
Non-Hodgkin's lymphoma IDO1 mRNA and protein in some tumours High IDO1 expression correlates with larger tumours, lower remission and survival rates [379
Oesophageal squamous cell carcinoma IDO1 mRNA and protein in the majority of tumours and metastatic and non-metastatic lymph nodes High IDO1 expression correlates with low Bin1 expression and more advanced disease and is an independent predictor of poor prognosis [372
Oral squamous cell carcinoma IDO1 protein in all tumours Higher IDO1 expression is an independent prognostic factor for decreased overall survival only in patients receiving adjuvant radio-chemotherapy or chemotherapy alone [602
Osteosarcoma IDO1 protein in most cancers High IDO1 expression is an independent risk factor for lower overall survival and metastasis-free survival [370
Ovarian cancer IDO1 protein in some tumours High IDO1 expression correlates with poorer overall survival and associates with paclitaxel-resistance in serous-type, but not clear cell, carcinomas [358,603
 IDO1 protein in some cancers High IDO1 expression is associated with impaired overall survival and progression-free survival [357
Pancreatic ductal carcinoma IDO1 protein in all tumours with lymph node metastases IDO1 is expressed in tumours with lymph node metastases, but not in tumours without metastases [367
Renal cell carcinoma IDO1 mRNA and protein in endothelial cells of newly formed blood vessels in a majority of tumours IDO1 expression is an independent prognostic factor for increased overall survival [390
T-cell leukaemia/lymphoma (adult) IDO1 protein in some tumour and non-tumour cells and increased serum Kyn levels and Kyn/L-Trp ratio Higher levels of serum Kyn and an increased Kyn/L-Trp ratio are negative independent prognostic factors, particularly in patients with aggressive variants. [376
Thyroid carcinoma IDO1 mRNA and protein in all tumours Increased IDO1 expression is associated with the aggressiveness of the cancer [381
Uterine cervical cancer IDO1 protein in all tumours at the invasive edge IDO1 expression does not correlate with prognosis [382
Cancer IDO1 expression and/or activity Prognosis and/or survival Reference(s) 
Acute myeloid leukaemia Varying IDO1 mRNA in AML blasts Higher IDO1 expression correlates with shorter overall survival and relapse-free survival [373
 Increased serum Kyn/L-Trp ratio Increased IDO1 activity correlates with decreased survival [374
Breast cancer Various IDO1 protein levels in stromal cells of all tumours IDO1 expression correlates with late-stage disease and reduced disease-specific and metastasis-free survival [360
 IDO1 mRNA and protein in tumours and lymph nodes IDO1 expression correlates with more advanced clinical staging and more extensive lymph node metastasis [362
 IDO1 protein is expressed in some tumours IDO1 expression correlates with high risk of lymph node metastasis [361
 IDO1 mRNA is expressed in all tumours High IDO1 mRNA expression is associated with medullary features and is an independent prognostic factor for improved survival in basal-like breast carcinoma [388
Cervical cancer IDO1 protein is expressed in some cancers IDO1 correlates with clinical stage, lymph node metastasis and lymph-vascular space invasion. Diffuse expression, but not focal expression at the tumour front, independently predicts impaired overall survival and disease-free survival [359
Colorectal cancer IDO1 protein is expressed in most tumours at the tumour front and/or the tumour centre IDO1 expression at the tumour front is an independent prognostic factor for decreased overall survival and the development of metachronous metastases in pT1-4N1Mx-staged colorectal cancer [594
 IDO1 protein is expressed in some tumours Increased IDO1 expression correlates with the frequency of liver metastasis and poorer overall survival [369
 Decreased serum L-Trp Decreased serum L-Trp correlates with liver metastases and reduced quality of life [595
Cutaneous melanoma IDO1 protein in all tumours and most lymph nodes Increased IDO1 correlates with cancer progression and metastasis [596
Diffuse large B-cell lymphoma Increased serum Kyn Kyn is an independent prognostic factor for reduced overall survival in patients with R-CHOP (rituximab/cyclophosphamide/hydoxydaunorubicin/oncovin/prednisolone) therapy [377
 IDO1 protein in some tumours Increased expression correlates with lower remission rates and lower survival rate [378
Endometrial cancer IDO1 protein in some tumours High IDO1 is an independent prognostic factor for progression-free survival or reduced disease-specific survival. Correlates with myometrial invasion, lymph-vascular space involvement, lymph node metastasis and poorer overall survival [363,597,598
Gastric cancer IDO1 protein is expressed in all tumours Higher IDO1 expression is associated with deeper invasion and more frequent lymph node metastasis [599
Glioma IDO1 mRNA and protein in all tumours High IDO1 expression is associated with malignancy and lower survival rates [371,600
Hepatocellular carcinoma IDO1 mRNA and protein in the tumour-infiltrating cells of some tumours IDO1-positive tumours associate with higher rates of recurrence-free survival relative to IDO1-negative tumours [389
 IDO1 protein in some tumours IDO1 expression is an independent prognostic factor for reduced overall survival and correlates with metastasis and a lower 5-year survival rate. No difference in recurrence rate was observed [368
Hodgkin's lymphoma IDO1 protein in histiocytes, DCs and endothelial cells IDO1 associates with later-stage disease, Epstein–Barr virus infection, poorer prognosis and inferior survival [380
Laryngeal squamous cell carcinoma IDO1 protein in all tumours. Higher IDO1 expression is an independent predictor of decreased overall survival and disease-free survival [601
Lung cancer IDO1 protein in peritumoral stroma eosinophils Higher IDO1 expression correlates with lower overall survival [365
 Increased serum Kyn/L-Trp ratio in serum Higher IDO1 activity associates with more advanced disease [364
Melanoma IDO1 protein in some sentinel lymph nodes Higher expression levels correlate with a worse long-term outcome [158
 IDO1 protein in sentinel lymph nodes with or without malignancy IDO1 expression does not correlate with malignancy [387
 IDO1 protein is expressed in the sentinel lymph nodes High IDO1 expression is an independent prognostic factor for lower overall survival and progression-free survival, particularly in patients with uninvolved lymph nodes [366
 IDO1 protein in some tumours cells, peritumoral endothelial cells and sentinel lymph node cells Peritumoral endothelial IDO1 and lymph node IDO1 expression is a negative prognostic factor for overall survival. Lymph node IDO1 expression is also a negative prognostic factor for relapse-free survival [407
Multiple myeloma Increased serum Kyn Increased Kyn correlates with late-stage disease [375
Non-Hodgkin's lymphoma IDO1 mRNA and protein in some tumours High IDO1 expression correlates with larger tumours, lower remission and survival rates [379
Oesophageal squamous cell carcinoma IDO1 mRNA and protein in the majority of tumours and metastatic and non-metastatic lymph nodes High IDO1 expression correlates with low Bin1 expression and more advanced disease and is an independent predictor of poor prognosis [372
Oral squamous cell carcinoma IDO1 protein in all tumours Higher IDO1 expression is an independent prognostic factor for decreased overall survival only in patients receiving adjuvant radio-chemotherapy or chemotherapy alone [602
Osteosarcoma IDO1 protein in most cancers High IDO1 expression is an independent risk factor for lower overall survival and metastasis-free survival [370
Ovarian cancer IDO1 protein in some tumours High IDO1 expression correlates with poorer overall survival and associates with paclitaxel-resistance in serous-type, but not clear cell, carcinomas [358,603
 IDO1 protein in some cancers High IDO1 expression is associated with impaired overall survival and progression-free survival [357
Pancreatic ductal carcinoma IDO1 protein in all tumours with lymph node metastases IDO1 is expressed in tumours with lymph node metastases, but not in tumours without metastases [367
Renal cell carcinoma IDO1 mRNA and protein in endothelial cells of newly formed blood vessels in a majority of tumours IDO1 expression is an independent prognostic factor for increased overall survival [390
T-cell leukaemia/lymphoma (adult) IDO1 protein in some tumour and non-tumour cells and increased serum Kyn levels and Kyn/L-Trp ratio Higher levels of serum Kyn and an increased Kyn/L-Trp ratio are negative independent prognostic factors, particularly in patients with aggressive variants. [376
Thyroid carcinoma IDO1 mRNA and protein in all tumours Increased IDO1 expression is associated with the aggressiveness of the cancer [381
Uterine cervical cancer IDO1 protein in all tumours at the invasive edge IDO1 expression does not correlate with prognosis [382

It is evident that IDO1 expression and/or activity as an adverse prognostic clinical marker does not apply to all cancers, or even within the same cancer type, and is likely to be dependent on factors such as the cohort of patients and their respective treatment regimes. Thus, some studies have found that IDO1 expression does not correlate with prognosis (e.g. melanoma and cervical cancer) [382,387], whereas others report that IDO1 is associated with an improved prognosis (e.g. breast cancer, hepatocellular carcinoma and renal cell carcinoma) [388390], although these reports are in the minority.

Cell types expressing IDO1 during cancer

A key question regarding IDO1 and cancer is the nature of the cell types expressing IDO1 and their relative contribution to the establishment of immune tolerance towards tumour antigens. Depending on the cancer, IDO1 can be expressed by the tumour cells themselves, infiltrating leucocytes (DCs and macrophages) present within the tumour microenvironment in the surrounding stroma and margins of the tumour or DCs within TDLNs [158,391,392]. Although there is evidence that IDO1 expressed within the tumour or tumour-infiltrating APCs exert localised immune tolerance (e.g. [356,393]), the increased expression of IDO1 within DCs present in TDLNs plays a prominent role in establishing systemic tolerance involving the generation of Tregs. Thus, elevated IDO1 expression in DCs isolated from TDLNs is apparent in various tumours, e.g. malignant melanoma, breast, lung, colon and pancreatic carcinomas in humans and mice [49,158,387]. In mice, IDO1-expressing cells in TDLNs have been primarily identified as CD19+ pDCs that, upon transfer into naïve hosts, induce antigen-specific T-cell anergy in a manner abrogated by IDO1 gene silencing or 1-MT [158]. More recent data indicate that IDO1-expressing murine pDCs within TDLNs not only facilitate the generation of Tregs but also directly activate pre-existing Tregs, thereby augmenting their suppressor activity involving the up-regulation of the PD-1/PD-L pathway [226]. IDO1 in pDCs also acts to maintain a stable Treg phenotype, i.e. Tregs within TDLNs normally exhibit functional plasticity such that they can be reprogrammed into inflammatory helper-like Th17 cells upon inhibition of IDO1 activity, which can promote tumour clearance [394,395]. A recent study shows that Treg reprogramming into pro-inflammatory helper-like T-cells involves a select subset of Tregs and is achieved through IL-6-induced down-regulation of the Eos transcription factor without loss of Foxp3+ expression, a process antagonised by the activation of AhR signalling by Kyn produced by active IDO1 expressed in murine pDCs derived from TDLNs of B16-F10 tumours [245]. Mesenchymal stem cells can also express IDO1 in the tumour microenvironment, resulting in immune suppression [396]. Similarly, Zhang et al. [397] identified a subset of circulating IDO1-expressing and immunosuppressive fibrocyte-like mesenchyme-derived stem cells in patients with metastatic cancer. Further characterisation of the nature of IDO1-expressing cells isolated from human tumours, TDLNs and the circulation requires further study [398,399].

A further important issue is the nature of the molecular signalling events or stimuli responsible for the elevated cellular IDO1 expression during cancer. Various candidates have been identified involving factors intrinsic to the tumour or the host immune system. For example, the up-regulation of IDO1 within certain tumour cells has been attributed to the loss of Bin1 expression, a tumour-suppressor gene which encodes a BAR adapter protein that regulates STAT1 and NF-κB transcription [120]. Notably, several human tumours are characterised by Bin1 deletion, aberrant Bin1 splicing or reduced Bin1 expression [120,126,372]. Expression of IDO1 by tumour-infiltrating DCs during prostate cancer is dependent on the activation of the transcription factor FOXO3 [111]. Recent evidence indicates that elevated IDO1 expressed in metastatic human melanoma tissue correlated with increased levels of CD8+ T-cells, PD-L1 and Tregs [383]. Employing a mouse model of melanoma, IFNγ−/− mice and selective CD8+ T-cell depletion, Spranger et al. [383] showed that IDO1 expression within the tumour microenvironment required IFNγ produced by tumour-infiltrating CD8+ T-cells. Interestingly, there is evidence that developing tumours may actively recruit IDO1-expressing leucocytes through their ability to express CCL21 (CC chemokine ligand 21) [392]. COX2-derived PGE2 is a further stimulus in tumours capable of signalling for elevated IDO1 expression. For example, Basu et al. [98] reported that the ability of the selective COX2 inhibitor celecoxib to inhibit IDO1 expression within metastatic spontaneous mammary tumours in mice improved the efficacy of a DC-based cancer vaccine involving enhanced tumour-specific CTLs (cytotoxic T-lymphocytes). Similarly, celecoxib treatment of tumour-bearing mice implanted with a lung cancer cell line inhibited the expression of IDO1 and Treg recruitment, correlating with reduced tumour mass and metastasis [400]. Recent data from Hanks et al. [401] studying mouse breast cancer and melanoma models reported that the absence in expression of type III TGFβ receptor and its shed extracellular domain in tumours augmented TGFβ-dependent expression of IDO1 in pDCs and the resultant infiltration of Tregs into the local tumour environment. Further studies are clearly warranted into the identity of the molecular signalling events and factors promoting IDO1 expression for different cancers characterised by tumour immune escape and whether these are intrinsic to the tumour or derived from host immune cells. Targeting these immunosuppressive signals leading to elevated IDO1 expression during cancer represents a logical immunotherapeutic approach to address tumour immune escape.

Experimental evidence for IDO1 and tumour immune escape

A large body of data derived from various animal models of tumorigenesis emphasise an important role for IDO1 in tumour immune escape (Table 3). For example, Uyttenhove et al. [356] demonstrated that transfection of the IDO1 into an immunogenic mouse mastocytoma cell line protected these cells from T-cell-mediated rejection when injected into mice pre-immunised with the tumour antigen [356]. Similarly, ‘humanised’ murine mesenchymal stem cells that express human IDO1 in place of murine NOS2 are able to promote tumour growth in vivo in a B16-F10 melanoma model, correlating with the inhibition of infiltrating CD8+ T-cells and B-cells [396]. Muller et al. [120] provided compelling evidence that the induction of IDO1 expression and resultant suppression of T-cell responses underscored the in vivo development of tumours characterised by the functional loss of the important tumour-suppressor gene Bin1. Notably, this study also demonstrated that pharmacological inhibition of IDO1 by 1-MT or other structurally diverse small-molecule inhibitors of IDO1 slowed in vivo cancer growth in mice and potentiated the actions of traditional chemotherapeutic drugs (e.g. paclitaxel, cisplatin and cyclosporin) when co-administered into tumour-bearing mice (e.g. in a B16-F10 melanoma and autochthonous MMTV-Neu breast cancer mouse models), resulting in significant tumour regression [120]. Similarly, IDO1 inhibition in combination with chemo-radiation therapy prolonged the survival of mice bearing intracranial glioblastoma tumours, which correlated with a marked increase in complement deposition throughout the tumour environment [402]. These studies highlight that inhibition of IDO1 represents a promising immunotherapeutic adjuvant therapy used in combination with more traditional anti-cancer treatments.

Table 3
Commonly studied IDO1-dependent murine cancer models
Tumour model Mouse strain Localization of IDO1 Approach to determine IDO1 and Kyn pathway involvement Reference(s) 
Autochthonous breast tumour model MMTV-Neu Not evaluated IDO1 inhibitors: 1-DL-MT, zinc protoporphyrin IX, methyl-thiohydantoin-tryptophan, menadione [120,164,411,421
Autochthonous gastrointestinal stromal tumour model KitV558/+ IDO1 expressed in tumour cells IDO1 inhibitors: 1-D-MT, administration of Kyn, 3-HAA and 3-HK [393
Autochthonous KRAS-induced lung carcinoma model Lox-KrasG12D Not evaluated Mice: IDO1−/−Lox-KrasG12D mice [604
Chronic colitis-associated cancer model C57BL/6 IDO1 expressed in tumour epithelial cells and DCs IDO1 inhibitors: 1-L-MT, administration of Kyn and QA; mice: IDO1−/− mice [405
Orthotopic B16-F10 melanoma model C57BL/6 IDO1 expressed in pDCs of tumour-draining lymph nodes Genetic manipulation of IDO1 expression: IDO1 transfection of tumour cells; IDO1 inhibitors: 1-D-MT, 1-L-MT, 1-DL-MT, brassinins, zinc protoporphyrin IX, ethyl pyruvate; mice: IDO1−/− mice [39,158,164,221,245,395,411,418,421,434
Orthotopic Bin1−/− MR KEC cell model (oncogenic transformed keratinocytes) C57BL/6 IDO1 expressed by tumour cells IDO1 inhibitors: zinc protoporphyrin IX, 1-DL-MT, ethyl pyruvate [120,164,434
Orthotopic CT26 colon carcinoma model Balb/c IDO1 expressed in tumour cells and lymph nodes IDO1 inhibitors: hydroxyamidines INCB023843 and INCB024360 [420
Orthotopic GL261 glioblastoma model C57BL/6 IDO1 expressed in tumour cells, in perivascular areas and around the tumour and in astrocyte-like peritumoral cells IDO1 inhibitors: 1-DL-MT, 1-D-MT, NLG919 [371,402
Orthotopic IDO-transfected P815 tumour cell model DBA/2 Tumour cells transfected to express IDO1 Genetic manipulation of IDO1 expression: IDO1 transfection of tumour cells; IDO1 inhibitor: 1-MT (isomer unknown) [356
Orthotopic Lewis lung cancer model C57BL/6 IDO1 expressed in tumour cells and draining lymph nodes IDO1 inhibitors: 1-L-MT, tryptanthrin derivatives, 1-MT (isomer unknown) [427,605
Orthotopic PAN02 pancreatic carcinoma model C57BL/6 IDO1 expressed by tumour cells IDO1 inhibitors: hydroxyamidines INCB023843 and INCB024360; mice: IDO1−/− mice [419,420
Orthotopic primary 4T1 breast carcinoma model with lung metastases Balb/c IDO1 expressed in the native host lung stroma [604] or tumour cells [606Genetic manipulation of IDO1 expression: plasmid expressing shRNA targeting IDO1; IDO1 inhibitors: 1-L-MT, I-DL-MT; mice: IDO1−/− mice [39,411,604,606
PMA-driven DMBA-induced papilloma model C57BL/6 IDO1 expressed in pDCs of tumour-draining lymph nodes IDO1 inhibitor: 1-D-MT; mice: IDO1−/− mice [403
Tumour model Mouse strain Localization of IDO1 Approach to determine IDO1 and Kyn pathway involvement Reference(s) 
Autochthonous breast tumour model MMTV-Neu Not evaluated IDO1 inhibitors: 1-DL-MT, zinc protoporphyrin IX, methyl-thiohydantoin-tryptophan, menadione [120,164,411,421
Autochthonous gastrointestinal stromal tumour model KitV558/+ IDO1 expressed in tumour cells IDO1 inhibitors: 1-D-MT, administration of Kyn, 3-HAA and 3-HK [393
Autochthonous KRAS-induced lung carcinoma model Lox-KrasG12D Not evaluated Mice: IDO1−/−Lox-KrasG12D mice [604
Chronic colitis-associated cancer model C57BL/6 IDO1 expressed in tumour epithelial cells and DCs IDO1 inhibitors: 1-L-MT, administration of Kyn and QA; mice: IDO1−/− mice [405
Orthotopic B16-F10 melanoma model C57BL/6 IDO1 expressed in pDCs of tumour-draining lymph nodes Genetic manipulation of IDO1 expression: IDO1 transfection of tumour cells; IDO1 inhibitors: 1-D-MT, 1-L-MT, 1-DL-MT, brassinins, zinc protoporphyrin IX, ethyl pyruvate; mice: IDO1−/− mice [39,158,164,221,245,395,411,418,421,434
Orthotopic Bin1−/− MR KEC cell model (oncogenic transformed keratinocytes) C57BL/6 IDO1 expressed by tumour cells IDO1 inhibitors: zinc protoporphyrin IX, 1-DL-MT, ethyl pyruvate [120,164,434
Orthotopic CT26 colon carcinoma model Balb/c IDO1 expressed in tumour cells and lymph nodes IDO1 inhibitors: hydroxyamidines INCB023843 and INCB024360 [420
Orthotopic GL261 glioblastoma model C57BL/6 IDO1 expressed in tumour cells, in perivascular areas and around the tumour and in astrocyte-like peritumoral cells IDO1 inhibitors: 1-DL-MT, 1-D-MT, NLG919 [371,402
Orthotopic IDO-transfected P815 tumour cell model DBA/2 Tumour cells transfected to express IDO1 Genetic manipulation of IDO1 expression: IDO1 transfection of tumour cells; IDO1 inhibitor: 1-MT (isomer unknown) [356
Orthotopic Lewis lung cancer model C57BL/6 IDO1 expressed in tumour cells and draining lymph nodes IDO1 inhibitors: 1-L-MT, tryptanthrin derivatives, 1-MT (isomer unknown) [427,605
Orthotopic PAN02 pancreatic carcinoma model C57BL/6 IDO1 expressed by tumour cells IDO1 inhibitors: hydroxyamidines INCB023843 and INCB024360; mice: IDO1−/− mice [419,420
Orthotopic primary 4T1 breast carcinoma model with lung metastases Balb/c IDO1 expressed in the native host lung stroma [604] or tumour cells [606Genetic manipulation of IDO1 expression: plasmid expressing shRNA targeting IDO1; IDO1 inhibitors: 1-L-MT, I-DL-MT; mice: IDO1−/− mice [39,411,604,606
PMA-driven DMBA-induced papilloma model C57BL/6 IDO1 expressed in pDCs of tumour-draining lymph nodes IDO1 inhibitor: 1-D-MT; mice: IDO1−/− mice [403

Initial work using Ido1 gene-deficient mice to identify a central role for IDO1 in supporting tumour immune escape came from Muller et al. [403] studying the traditional two-step inflammatory model of skin carcinogenesis, where tumours are initiated with a mutagen [DMBA (7,12-dimethylbenz[a]anthracene)] and subsequently promoted with phorbol ester (PMA). In this model, IDO1−/− mice were resistant to tumour formation in response to DMBA/PMA exposure. In contrast, wild-type mice were characterised by the IFN-dependent induction of IDO1 in TDLN pDCs, which suppressed T-cell responses and favoured skin tumour formation [403].

Interestingly, the effectiveness of certain anti-cancer drugs or therapeutic approaches currently in clinical use is linked to IDO1. For example, imatinib, a tyrosine kinase inhibitor that is currently used to treat chronic myeloid leukaemia and gastrointestinal stromal tumours, has been retroactively shown to inhibit IDO1-mediated suppression of T-cells [393]. The possibility that imatinib may act by blocking IDO1 tyrosine phosphorylation, thus preventing the ability of tyrosine phosphorylation to signal for IDO1-mediated Treg differentiation, is intriguing and worthy of further investigation. Furthermore, in a mouse melanoma model, the efficacy of ipilimumab, a CTLA-4-blocking antibody currently approved for use in advanced melanoma, was significantly improved by concurrently inhibiting IDO1 activity with D- or L-1-MT [39]. This study highlights the key concept that induction of IDO1 in response to certain immunotherapies attenuates their anti-cancer efficacy and, as such, the co-application of immunotherapy and IDO1 inhibition may prove to be a more effective anti-cancer treatment regimen. Notably, as indicated above, various other potential immune-based anti-cancer treatments can also induce IDO1 including IFNγ, CpG-ODNs and anti-4-1BB (anti-CD137) antibody and the extent to which induction of IDO1 may confound their anti-tumour efficacy requires clarification [398].

In addition to directly abrogating anti-tumour T-cell responses, up-regulation of IDO1 during cancer may influence disease progression in other ways. For example, Muller et al. [403] have proposed that IDO1 expressed in non-haematopoietic cells acts co-operatively in the establishment of a tumour-supportive chronic inflammatory environment. Thus, whereas IDO1 was essential for robust tumour development in a two-step chronic inflammatory model of skin carcinogenesis [403], IDO1 gene deficiency had little effect on tumour formation induced in the absence of an inflammatory tumour-promoting agent [404]. This finding, together with the examples of arthritis and asthma described above (see the ‘Chronic inflammatory and allergic disorders’ section above), emphasises the complex role of IDO1 during inflammation and that a view of IDO1 as an immunosuppressive, anti-inflammatory enzyme is too simplistic. In addition to supporting cancer-associated inflammation, recent data indicate that IDO1-derived Kyn pathway metabolites may also directly support tumour cell proliferation. In a mouse model of colitis-associated tumorigenesis, tumour burden and the proliferation of neoplastic epithelium were not only reduced in 1-MT-treated wild-type and Ido1 gene-deficient mice but also in Rag1−/−/IDO1−/− mice compared with Rag1−/− mice, which lack adaptive immune cells [405]. These findings indicate that IDO1 can also support in vivo tumour growth in a manner independent of its immune regulatory actions [405]. Instead, in vitro and in vivo studies indicate that IDO1-derived Kyn metabolites can directly activate β-catenin signalling and proliferation of colon cancer cells [405]. A study investigating peritoneal dissemination in mice injected with IDO1-overexpressing human ovarian cancer cells also provided in vivo evidence for a potential relationship between IDO1 and angiogenesis [406]. Consistent with this, elevated IDO1 expression has been detected within endothelial cells in tumours isolated from human melanoma and renal cell carcinoma patients [390,407].

Although the most focus has been placed on IDO1 in the context of cancer, recent data also point to a similar role for TDO when expressed in tumour cells. For example, in human gliomas that do not up-regulate IDO1, TDO-derived Kyn acts to promote tumour progression by binding to the AhR [23,24]. The binding of TDO-derived Kyn to AhR in T-cells mediates immune escape via the inhibition of effector T-cells and the induction of Treg differentiation, whereas Kyn-AhR binding in tumour cells promotes tumour cell migration and survival, one aspect of cancer biology that IDO1 has yet-to-be implicated [23,234]. Increased levels of IDO2 mRNA and protein have also been detected in primary human pancreatic ductal adenocarcinomas [408], as well as gastric, renal and colon carcinomas [409], although its relevance to cancer development and progression has to date not been investigated in detail.

IDO1 inhibitors as putative immunotherapeutic anti-cancer drugs

In the light of the promising results of pre-clinical animal studies identifying IDO1 inhibitors as a novel class of immune-targeted adjuvants for more traditional chemotherapeutic agents, Phase I/II clinical trials are currently underway to investigate whether this is also applicable to human malignancy (e.g. identifiers NCT01961115, NCT02042430, NCT02166905, NCT01822691, NCT02118285, NCT01792050, NCT01560923, NCT02077881, NCT02073123, NCT02052648 and NCT01042535 at http://ClinicalTrials.gov). Moreover, pharmacological targeting of IDO1 appears advantageous in that IDO1 is a single-chain monomeric haem enzyme that is structurally dissimilar to other mammalian haem dioxygenases (TDO and IDO2) and the biochemistry of which is becoming increasingly understood (see the ‘IDO1 biochemistry’ section above), features permitting the effective design of selective small-molecule inhibitors. Also, Ido1 gene-deficient mice are in general viable and of overall good health, suggesting that pharmacological targeting of IDO1 should afford minimal side effects [342]. A promising clinical drug candidate targeting IDO1 is the D-isoform of 1-MT (D-1-MT), also referred to as indoximod or NLG8189. The first small-scale Phase I clinical trial for indoximod has been recently published, which reported that twice daily oral provision of indoximod (300–1200 mg) in combination with the chemotherapeutic docetaxel given to patients with metastatic solid tumours was overall well-tolerated with no drug–drug interactions and some degree of anti-tumour activity observed [410].

The focus on the D- over the L-isoform of 1-MT for clinical trials was based on early findings that the former isoform proved more efficacious with respect to abrogating the T-cell-inhibitory actions of IDO1-expressing APCs in vitro [49,411] and acting as an anti-cancer immunotherapeutic adjuvant for chemotherapy in a variety of pre-clinical mouse tumour models [411]. Importantly, the in vivo anti-tumour actions of 1-D-MT are abrogated in tumour-bearing Ido1 gene-deficient mice [411], supporting the specificity of the actions of 1-D-MT for IDO1. The greater in vivo potency of 1-D-MT was initially surprising given that studies with recombinant human IDO1 enzyme and IDO1-expressing cells in culture indicated that 1-L-MT, but not 1-D-MT, effectively inhibits IDO1 dioxygenase activity (e.g. [411]). Mechanistic studies, however, suggest 1-D-MT may instead act to counteract downstream IDO1-dependent signalling pathways. For example, Metz et al. [227] reported that 1-D-MT acts as a Trp-mimetic to reverse the immunosuppressive action of IDO1-catalysed Trp insufficiency, which signals for inhibition of the immune-regulatory proteins mTOR or PKCθ activity, leading to T-cell autophagy. Notably, this Trp-mimetic mode of action of 1-D-MT is reportedly more efficient than L-Trp itself [227]. This study highlights that 1-D-MT may not directly act as an effective inhibitor of IDO1 enzyme activity; rather, it instead acts as an inhibitor of the IDO1 immunosuppressive pathway. Interestingly, 1-D-MT has also been reported to induce the expression of active IDO1 protein in cultured tumour cell lines in vitro via p38 MAPK and JNK (c-Jun N-terminal kinase) signalling [412]. The extent to which this potentially confounding action of 1-D-MT on IDO1 inhibition occurs in vivo is currently not clear, however.

Although a significant body of evidence indicates that 1-D-MT and 1-L-MT act in an IDO1-dependent manner, it is important to note that 1-MT can also target IDO2 under certain circumstances. For example, in vitro studies have found that 1-D-MT inhibits cellular IDO2 activity [32], whereas evidence that 1-D-MT can target IDO2 in vivo derives from a very recent study employing Ido2 gene-deficient mice, which revealed a key role for IDO2 (but not IDO1) in supporting autoreactive antibody responses in a murine autoimmune arthritis model [37]. Other studies report that 1-L-MT is more effective as an IDO2 inhibitor than 1-D-MT [413,414] or conclude that both isomers of 1-MT are inefficient at inhibiting cellular IDO2 [414]. The above emphasises the clear need for more studies with Ido1 and Ido2 gene-deficient mice and the further development and discovery of selective IDO1 and IDO2 inhibitors in order to better define the roles of these two haem dioxygenases and specificity of action of 1-D-MT and other IDO1-targeted inhibitors with in vivo therapeutic potential.

Although the L-isomer of 1-MT more efficiently inhibits IDO1 enzyme activity than 1-D-MT, it is a relatively inefficient competitive inhibitor of IDO1 dioxygenase activity, exhibiting a Ki of 19–34 μM [148]. As such, there is significant interest in the discovery of novel classes of small-molecule inhibitors of IDO1 activity with nanomolar potency. Such agents may represent effective alternatives to 1-MT as new classes of immune-based adjuvants useful in the treatment of cancer [415,416]. Several, structurally diverse classes of potent small-molecule IDO1 dioxygenase activity inhibitors have been reported (reviewed in [417]), e.g. brassinins [418], hydroxyamidines [130,419,420], naphthoquinones [421,422], methyl-thiohydantoin-tryptophan [120], imidazoles (e.g. phenylimadazole and imidazolethiazoles) [129,423,424], exiguamines [425], triazoles [426], indoles [427,428], selenazoles [135] and hydrazines [129,429]. A recent high-throughput screen of major available drug libraries highlighted the potential challenges surrounding the discovery of selective and potent small-molecule inhibitors of IDO1 [424].

In general, these various small-molecule inhibitors are considered to act via their ability to interact within IDO1’s active site, although, in many cases, rigorous biochemical and biophysical studies to confirm this have yet-to-be performed. This is important in the light of recent mechanistic studies into IDO1 inhibition. We have shown that the selenazole drug ebselen, potently inhibited IDO1 activity independent of an interaction with the active-site haem. Instead, ebselen reacted with and formed a covalent adduct with IDO1 cysteine residues, which altered protein secondary structure leading to the non-productive binding of substrate and inhibition of dioxygenase activity [135]. Notably, several other IDO1 inhibitors identified to date (e.g. quinones or compounds containing quinone-like functions) are redox-active and potentially capable of interfering with the reducing cofactors in the IDO1 assay or reacting with IDO1’s amino acid residues [424,430]. More detailed mechanistic biochemical and biophysical studies are therefore warranted, ideally employing physiologically relevant reducing cofactor systems in order to define the precise mechanisms by which different agents with in vivo anti-cancer potential inhibit IDO1 dioxygenase activity. Such mechanistic knowledge will inform the development of more effective and selective in vivo potent IDO1 enzyme inhibitors. Importantly, and in keeping with IDO1’s prominent immunosuppressive role during cancer, several of the IDO1-inhibitory drug classes discovered to date have been reported to exhibit in vivo anti-tumour activity in an IDO1- and T-cell-dependent manner [418,419,421]. Of these, the hydroxyamidine INCB024360, a selective nanomolar potent IDO1 inhibitor capable of inhibiting tumour growth in vivo, is currently being tested in clinical trials in cancer patients [419,420].

The promising findings with 1-D-MT as an in vivo inhibitor of an immunosuppressive effector mechanism downstream of IDO1 activity emphasise that strategies aimed at abrogating IDO1’s immune regulatory actions should not only consider direct inhibition of dioxygenase activity but also target the IDO1 pathway more broadly (e.g. targeting GCN2 or AhR activity). The efficacy of the strategies to inhibit IDO1 enzyme activity may also depend on the nature of the physiological intracellular reducing cofactor(s) necessary for maintaining IDO1 dioxygenase activity. In vivo IDO1 inhibition may also have to take into account recent studies showing that IDO1 is a versatile haem enzyme, which, depending on the local substrate environment, can catalyse several alternative enzyme reactions, e.g. haem peroxidase activity [30]. The emergence of tyrosine-phosphorylated IDO1 as a key signalling event for long-term systemic immune tolerance [25,113], which acts independently of enzyme activity and hence is likely to be refractory to enzyme-based IDO1 inhibitors, will require alternative intervention strategies aimed at inhibiting the relevant protein kinase signalling pathways, namely the Src family non-receptor tyrosine kinases linked to IDO1 tyrosine phosphorylation.

One strategy capable of inhibiting both enzyme activity-dependent and -independent immune regulatory actions of IDO1 is via inhibition of IDO1 gene expression. This could involve the indirect inhibition of key proximal signalling events that elevate IDO1 expression during cancer (e.g. inhibiting COX2/PGE2 [98,400]) or direct inhibition of IDO1 gene expression. Indeed, skin delivery of IDO1-targeted shRNA or siRNA to suppress IDO1 expression promotes tumour immunity in mouse models of bladder, colon or liver cancer [431,432]. Similarly, intra-tumour IDO1 siRNA delivery augmented anti-tumour T-cell responses and tumour cell killing in the B16-F10 mouse melanoma model [433]. Blache et al. [264] reported on a novel immunotherapeutic approach involving the incorporation of IDO1-specific shRNA into Salmonella enterica serovar Typhimurium (a microbial class capable of surviving hypoxic and potentially immunoprivileged tumour environments) as a means of combining IDO1 gene silencing with tumour colonisation of S. enterica Typhimurium that would act in concert to attract ROS-generating neutrophils to the tumour capable of clearing the microbe and promoting tumour cell apoptosis [264]. Systemic delivery of the variant IDO1 shRNA S. enterica Typhimurium strain into mice supported the robust neutrophil-dependent B16-F10 tumour killing [264]. In addition to siRNA, the ability of the anti-inflammatory compound ethyl pyruvate to inhibit tumour growth in mice is linked to its ability to inhibit NF-κB and IDO1 gene expression [434]. A further novel IDO1-targeted anti-cancer therapeutic employed in a recent Phase I clinical trial involved vaccination of metastatic lung cancer patients with an IDO1 peptide, which was tolerated by the patient group and produced long-lasting disease stabilisation [268].

Highlighting the potential complexity and contextual nature of IDO1’s actions, inhibition of the enzyme may not always be advantageous during cancer. A study on clear cell renal cell carcinoma in humans found that IDO1 was mainly expressed within angiogenic endothelial cells of newly formed blood vessels within the tumour and that IDO1 expression correlated with reduced numbers of proliferating tumour cells and patient survival [390]. Thus, whereas the bulk of the data support an important role for IDO1 in promoting tumour immunoprivilege, the paradigm of IDO1 always playing an immunosuppressive role in cancer is perhaps too simplistic as the enzyme, depending on the cell types expressing IDO1, tumour type and local environment, may play other roles including tumour growth retardation and modulation of the local tumour inflammatory microenvironment, capable of influencing tumour survival and progression. Importantly, although inhibiting IDO1 is generally considered to be without substantive toxicity [410], its inhibition in the context of certain cancers and patient groups may elicit unwanted side effects. For example, considering that IDO1 inhibition can exacerbate certain autoimmune and inflammatory diseases (e.g. inflammatory bowel disease; see the ‘Autoimmune diseases’ section above), treatment of cancer patients (e.g. colon cancer) with these ailments with IDO1 inhibitors warrants caution.

Taken together, it is evident that IDO1 expression in host-derived immune cells contained within the local tumour environment or tumour-draining lymph nodes, and IDO1 expressed by the tumour itself has major implications for both local and systemic immune suppression to facilitate immune escape by cancer cells. As such, there is great interest in the outcomes of clinical trials currently investigating the utility of IDO1-based inhibitors as effective immune-adjuvants in the treatment of an expanding list of human cancers where the immunosuppressive action of IDO1 is implicated. As is detailed below, effective IDO1 blockade may also prove beneficial in the treatment of microbial pathogen persistence.

IDO1 and host response to microbial infection

Conventionally, IDO1 is considered a component of the innate immune response against infectious organisms, including viruses, bacteria, fungi and parasites [349,350]. Although some micro-organisms can synthesise L-Trp de novo, it is energetically unfavourable to do so (i.e. >70 molecules of ATP are required per L-Trp compared with 11 ATP per glycine, alanine or serine [435]) and consequently, many microbes have evolved to become auxotrophic and take up L-Trp from the host microenvironment [21]. Accordingly, this requirement for an external source of L-Trp can render certain micro-organisms sensitive to its depletion, particularly those that are closely reliant on the host cell for replication and survival. Therefore, as detailed below, many in vitro studies have identified IDO1 as an important antimicrobial effector mechanism of IFNγ and other cytokines, and there are recent data to support this antimicrobial action of IDO1 for certain pathogens in vivo. However, it is increasingly evident that, in spite of the antimicrobial potential of IDO1, certain micro-organisms can withstand IDO1-mediated L-Trp deprivation in order to hijack the enzyme's immunosuppressive actions, in a manner similar to tumour cells, and evade clearance by the host immune system. These aspects of IDO1 in host defence are discussed in the context of various infectious agents in the following sections.

Bacteria

Bacteria are single-celled micro-organisms that, although not always pathogenic, are opportunistic, causing clinical diseases such as pneumonia, tuberculosis, tetanus and sepsis. Although the majority of bacterial infections can be treated effectively with antibiotics, the rapid emergence and rise of antibiotic-resistant (and multi-antibiotic-resistant) bacterial strains, together with the associated health and economic burden [436,437], have highlighted the urgent need for a more comprehensive understanding of antibacterial host defence mechanisms and the development of novel strategies to combat bacterial infection. The role of IDO1 in the host response to infection with obligate intracellular bacteria (those that cannot survive outside of a host cell), facultative intracellular bacteria (those that invade cells when beneficial for the bacteria) and extracellular bacteria (those that do not invade eukaryotic cells) has received considerable attention (see Tables 4 and 5 for a summary and Supplementary Table S1 for details).

Table 4
Expression and roles of IDO1 in microbial infections in vitro

IDO1 induction is defined as the increased expression of IDO1 mRNA or protein, or enhanced IDO1 activity (i.e. increased Kyn metabolites or functional IDO1) in response to in vitro microbial infection. Inhibition of microbial growth in vitro is defined as infection inhibited by IDO1, regardless of whether the IDO1 is pathogen-induced (*) or up-regulated by other stimuli/methods (e.g. IFNγ±TNFα, transfection of plasmids encoding IDO1) before pathogen exposure. †Growth of bacterial/fungal cultures inhibited by Kyn metabolites. ‡Microbe replication within cells is inhibited by IDO1, and the growth of bacterial cultures is inhibited by Kyn metabolites. See the ‘IDO1 and host response to microbial infection’ section for corresponding references.

Micro-organism Induction of IDO1 in vitro Inhibition of microbial growth in vitro Enhancement of in vitro infection Promotion of immunosuppression No effect on microbial growth 
Bacteria Chlamydia pneumoniae Chlamydia muridarum  Escherichia coli Mycobacterium tuberculosis 
 Escherichia coli Chlamydia pneumoniae  Mycobacterium tuberculosis  
 Listeria monocytogenes Chlamydia psittaci    
 Mycobacterium avium Chlamydia trachomatis    
 Mycobacterium bovis (BCG) Escherichia coli†    
 Mycobacterium tuberculosis Lactobacillus reuteri    
  Listeria monocytogenes   
  Mycobacterium avium    
  Mycobacterium tuberculosis†    
  Pseudomonas aeruginosa†    
  Rickettsia conorii    
  Staphylococcus aureus‡    
  Staphylococcus epidermis†    
  Streptococcus agalactiae    
  Streptococcus pneumoniae    
Fungi Aspergillus fumigatus Aspergillus fumigatus†    
  Candida albicans†    
Parasites Leishmania donovani Encephalitozoon intestinalis Leishmania major Leishmania donovani Cryptosporidium parvum 
 Leishmania major Leishmania donovani  Leishmania major Toxoplasma gondii 
 Trypanosoma cruzi Neospora caninum    
  Toxoplasma gondii    
  Trypanosoma cruzi   
Viruses Dengue virus Cytomegalovirus HIV Epstein–Barr virus Hepatitis C virus 
 HIV Dengue virus Influenza virus HIV Japanese encephalitis virus 
 Influenza virus Hepatitis B virus   West Nile virus 
 Japanese encephalitis virus Herpes simplex virus type 1    
 West Nile virus Herpes simplex virus type 2    
  Measles virus    
  Vaccinia virus    
  West Nile virus    
Micro-organism Induction of IDO1 in vitro Inhibition of microbial growth in vitro Enhancement of in vitro infection Promotion of immunosuppression No effect on microbial growth 
Bacteria Chlamydia pneumoniae Chlamydia muridarum  Escherichia coli Mycobacterium tuberculosis 
 Escherichia coli Chlamydia pneumoniae  Mycobacterium tuberculosis  
 Listeria monocytogenes Chlamydia psittaci    
 Mycobacterium avium Chlamydia trachomatis    
 Mycobacterium bovis (BCG) Escherichia coli†    
 Mycobacterium tuberculosis Lactobacillus reuteri    
  Listeria monocytogenes   
  Mycobacterium avium    
  Mycobacterium tuberculosis†    
  Pseudomonas aeruginosa†    
  Rickettsia conorii    
  Staphylococcus aureus‡    
  Staphylococcus epidermis†    
  Streptococcus agalactiae    
  Streptococcus pneumoniae    
Fungi Aspergillus fumigatus Aspergillus fumigatus†    
  Candida albicans†    
Parasites Leishmania donovani Encephalitozoon intestinalis Leishmania major Leishmania donovani Cryptosporidium parvum 
 Leishmania major Leishmania donovani  Leishmania major Toxoplasma gondii 
 Trypanosoma cruzi Neospora caninum    
  Toxoplasma gondii    
  Trypanosoma cruzi   
Viruses Dengue virus Cytomegalovirus HIV Epstein–Barr virus Hepatitis C virus 
 HIV Dengue virus Influenza virus HIV Japanese encephalitis virus 
 Influenza virus Hepatitis B virus   West Nile virus 
 Japanese encephalitis virus Herpes simplex virus type 1    
 West Nile virus Herpes simplex virus type 2    
  Measles virus    
  Vaccinia virus    
  West Nile virus    
Table 5
Expression and roles of IDO1 in microbial infections in vivo

NB: certain pathogens belong to a number of different (and occasionally seemingly contradictory) categories, reflecting the multiple roles that IDO1 may have in an infectious setting in vivo, depending on, for instance, the particular characteristics of a patient group, the specific animal model employed and/or the pathogen strain and dose. IDO1 induction is defined as the increased expression of IDO1 mRNA or protein, or enhanced IDO1 activity (i.e. increased Kyn metabolites or functional IDO1) in response to in vivo microbial infection. Note that, unlike in Table 4, inhibition of in vivo microbial growth or disease only refers to the antimicrobial role of pathogen-induced IDO1 expression (i.e. IDO1 is not overexpressed in vivo before infection). *Diseases/infections where severity/immunosuppression correlates with increased IDO1 activity, but where IDO1 has yet to be explicitly demonstrated to be the cause, as is often the case with human studies. †Human studies where disease severity/infection correlates with enhanced IDO1 activity, and IDO1 has been shown to be involved in animal models. See the ‘IDO1 and host response to microbial infection’ section for corresponding references.

Micro-organism Induction of IDO1 in vivo Inhibition of in vivo microbial growth or disease Enhancement of infection or disease severity in vivo Promotion of immunosuppression No effect on microbial growth or disease severity 
Bacteria Bartonella henselae Francisella novicida Escherichia coliEscherichia coli Chlamydia trachomatis 
 Escherichia coli (uropathogenic strain) Lactobacillus reuteri Escherichia coli (uropathogenic strain) Mycobacterium tuberculosis Lactobacillus johnsonii 
     Mycobacterium tuberculosis 
 Francisella novicida  Mycobacterium avium  
 Listeria monocytogenes  Staphylococcus aureus  
 Mycobacterium avium  Streptococcus agalactiae  
 Mycobacterium tuberculosis  Streptococcus pneumoniae†   
 Staphylococcus aureus     
 Streptococcus agalactiae     
 Streptococcus pneumoniae     
Fungi Aspergillus fumigatus Aspergillus fumigatus Histoplasma capsulatum Aspergillus fumigatus  
 Candida albicans Candida albicans  Candida albicans  
 Histoplasma capsulatum   Paracoccidioides brasiliensis  
Parasites Eimeria falciformis Toxoplasma gondii Eimeria falciformis Leishmania major Plasmodium berghei ANKA 
 Leishmania guyanensis Trypanosoma cruzi Leishmania major Toxoplasma gondii Trichinella spiralis 
 Leishmania infantum Paracoccidioides brasiliensis Plasmodium falciparum  
 Leishmania major  Toxoplasma gondii   
 Plasmodium berghei ANKA  Trichuris muris   
 Plasmodium berghei K173     
 Plasmodium falciparum     
 Toxoplasma gondii     
 Trichinella spiralis     
 Trichuris muris     
 Trypanosoma cruzi     
Viruses Dengue virus Mouse hepatitis virus Encephalomyocarditis virus Encephalomyocarditis virus Influenza virus 
 Encephalomyocarditis virus Simian immunodeficiency virus Hepatitis B virus Hepatitis B virus LP-BM5 murine leukaemia virus 
 Epstein–Barr virus  Hepatitis C virus* HIV  
 Hepatitis B virus  HIV† Influenza virus  
 Hepatitis C virus  Influenza virus LP-BM5 murine leukaemia virus  
 Herpes simplex virus type 1  LP-BM5 murine leukaemia virus   
 HIV  Simian immunodeficiency virus   
 Influenza virus     
 LP-BM5 murine leukaemia virus     
 Poliovirus     
 Simian immunodeficiency virus     
Micro-organism Induction of IDO1 in vivo Inhibition of in vivo microbial growth or disease Enhancement of infection or disease severity in vivo Promotion of immunosuppression No effect on microbial growth or disease severity 
Bacteria Bartonella henselae Francisella novicida Escherichia coliEscherichia coli Chlamydia trachomatis 
 Escherichia coli (uropathogenic strain) Lactobacillus reuteri Escherichia coli (uropathogenic strain) Mycobacterium tuberculosis Lactobacillus johnsonii 
     Mycobacterium tuberculosis 
 Francisella novicida  Mycobacterium avium  
 Listeria monocytogenes  Staphylococcus aureus  
 Mycobacterium avium  Streptococcus agalactiae  
 Mycobacterium tuberculosis  Streptococcus pneumoniae†   
 Staphylococcus aureus     
 Streptococcus agalactiae     
 Streptococcus pneumoniae     
Fungi Aspergillus fumigatus Aspergillus fumigatus Histoplasma capsulatum Aspergillus fumigatus  
 Candida albicans Candida albicans  Candida albicans  
 Histoplasma capsulatum   Paracoccidioides brasiliensis  
Parasites Eimeria falciformis Toxoplasma gondii Eimeria falciformis Leishmania major Plasmodium berghei ANKA 
 Leishmania guyanensis Trypanosoma cruzi Leishmania major Toxoplasma gondii Trichinella spiralis 
 Leishmania infantum Paracoccidioides brasiliensis Plasmodium falciparum  
 Leishmania major  Toxoplasma gondii   
 Plasmodium berghei ANKA  Trichuris muris   
 Plasmodium berghei K173     
 Plasmodium falciparum     
 Toxoplasma gondii     
 Trichinella spiralis     
 Trichuris muris     
 Trypanosoma cruzi     
Viruses Dengue virus Mouse hepatitis virus Encephalomyocarditis virus Encephalomyocarditis virus Influenza virus 
 Encephalomyocarditis virus Simian immunodeficiency virus Hepatitis B virus Hepatitis B virus LP-BM5 murine leukaemia virus 
 Epstein–Barr virus  Hepatitis C virus* HIV  
 Hepatitis B virus  HIV† Influenza virus  
 Hepatitis C virus  Influenza virus LP-BM5 murine leukaemia virus  
 Herpes simplex virus type 1  LP-BM5 murine leukaemia virus   
 HIV  Simian immunodeficiency virus   
 Influenza virus     
 LP-BM5 murine leukaemia virus     
 Poliovirus     
 Simian immunodeficiency virus     
Obligate intracellular bacteria

Given the prerequisite for obligate intracellular bacteria to reside within the host cell, it is unsurprising to find that this microbe class is particularly susceptible to IDO1-mediated L-Trp depletion. The majority of research on IDO1 and obligate intracellular bacteria pertains to Chlamydia, which exhibits a two-stage cell cycle involving the extracellular infectious form (the elementary body) and the intracellular non-infectious, but metabolically active, form (termed the reticulate body). The reticulate body can replicate within the host cell and differentiate back into elementary bodies, which then exit the host to infect neighbouring cells. Chlamydia can infect both men and women, causing complications such as pelvic inflammatory disease and infertility in women, and severe respiratory and ocular infections. Chlamydia infection and persistence is also implicated in the pathogenesis or exacerbation of certain chronic inflammatory disorders, such as atherosclerosis and asthma [438,439].

Numerous in vitro studies indicate that IFNγ, LPS and TNFα can inhibit the intracellular replication of Chlamydia trachomatis, Chlamydia muridarum, Chlamydia psittaci and Chlamydia pneumoniae within cultured cells, including primary human cells (monocyte-derived and alveolar macrophages, aortic SMCs and monocyte-derived DCs) and various human epithelial cell lines, in a manner that is dependent on the up-regulation of IDO1 and subsequent depletion of L-Trp [57,87,440445]. A notable characteristic of Chlamydia, however, is the microbe's ability to enter a viable, but non-replicative, persistent state under stress conditions, with IFNγ-mediated IDO1 induction and local L-Trp deprivation identified as an important determinant of this process. For example, when faced with severe L-Trp starvation, C. trachomatis cannot multiply or convert into a dormant state, and thus cannot regain its infectivity [446,447]. However, under sub-optimal L-Trp conditions, C. trachomatis enters a non-infectious, but persistent, state within the host cell, with a normal growth cycle restored upon L-Trp supplementation [446,447]. Similar IDO1-dependent persistence has been noted for other strains including C. pneumoniae [448]. Recent data indicate that pharmacological IDO1 inhibition with 1-L-MT enhances the efficacy of the antibiotic doxycycline in the clearance of persistent forms of C. trachomatis [449]. These studies identify a key role for IDO1 in supporting Chlamydia persistence, representing a potential mechanism by which the enzyme could exacerbate chronic inflammation.

Interestingly, it has become increasingly clear that Chlamydia can adapt to high-IDO1 environments where they remain latent, but functional, in part due to the expression of functional L-Trp biosynthesis pathways in the bacterium itself, which are up-regulated under conditions of low L-Trp [450]. Differences in the expression (or the absence of the expression) of enzymes in the L-Trp synthesis pathway in different Chlamydia strains and serovars dictate the nature of the substrates that they are able to use for L-Trp synthesis, their ability to survive in certain host environments, and therefore their virulence [442,451]. Of the four major Chlamydia species, C. psittaci exhibits the most complete and functional L-Trp synthesis pathway, compared with C. trachomatis, C. muridarum and C. pneumoniae [452]. Accordingly, replication of the C. psittaci strain GPIC in HeLa cells is highly resistant to IFNγ-mediated IDO1 induction as it can utilise Kyn, indole or anthranilate as substrates for L-Trp biosynthesis, whereas C. trachomatis serovar L2 can only utilise indole as a substrate [453]. Human, but not murine, genital Chlamydia strains encode tryptophan synthase enzymes [442], enabling human strains to evade IDO1-mediated inhibition of replication, which probably synthesise L-Trp from the indole produced by other vaginal microflora present in the reproductive tract to establish persistent infection [451]. This ability to synthesise L-Trp may account for there being no difference in bacterial shedding between wild-type and IDO1−/− mice infected with human C. trachomatis strains [442], as this bacterium can withstand IDO1. This may also account for the relative inability of mouse C. muridarum strains to infect humans [454]. Given that IFNγ-treated murine epithelial cells do not inhibit C. muridarum infection via IDO1, whereas IFNγ-treated human epithelial cells do inhibit human C. trachomatis strains in an IDO1-dependent manner, Nelson et al. [442] have proposed that the human and mouse Chlamydia strains have evolved in concert with their respective hosts, i.e. in human strains, the development of a strategy to avoid L-Trp depletion is necessary for infection of humans. Further along these lines, ocular C. trachomatis isolates from humans were found to have non-functional tryptophan synthase genes, whereas an ocular serovar isolated from the genital tract did express active versions of these enzymes [451]. Together, these data suggest that the presence and extent of a functional L-Trp biosynthesis pathway dictates not only the host species that the bacteria can infect, but also the tissue tropism of Chlamydia serovars, and is a highly effective survival strategy of certain Chlamydia strains.

Of further interest, Chlamydia infection and replication can induce the expression of IDO1 in cultured cells in the absence of exogenously added cytokines. For example, human macrophages infected with C. psittaci or synovial fibroblasts infected with C. trachomatis induce IDO1 in a manner dependent on the production of type I IFN by infected cells [455,456]. More recent data indicate that C. pneumoniae infection induces IDO1 in human monocyte-derived DCs in a TNFα-dependent manner, which can restrict chlamydial growth in these cells [444].

In addition to the bacterial expression of a functional tryptophan biosynthetic pathway, the local oxygen supply is a further factor governing the effectiveness of IDO1 in combating Chlamydia. Thus, O2 concentrations are low within the urogenital tract microenvironment and Roth et al. [122] showed that hypoxic conditions mitigate IFNγ-induced inhibition of C. trachomatis in vitro and ex vivo. These hypoxic conditions also inhibited IFNγ-induced IDO1 expression and activity, although the failure of IFNγ to control infection was not explicitly shown to be due to the reduced level of functional IDO1 in this study [122].

Although the bulk of research on IDO1 and Chlamydia has focused on IDO1 as a host defence, recent evidence supports an immune regulatory role for Chlamydia-induced IDO1. Thus, in vitro C. pneumonia infection in conjunction with cigarette exposure of murine bone marrow-derived DCs elevated IDO1 expression within a pDC subset, which upon adoptive transfer into the lungs of naïve mice inhibited Th2-mediated inflammation induced by myeloid DCs also pulsed with both C. pneumonia and cigarette extract [457]. These data are consistent with a potential protective role for IDO1-expressing pDCs in suppressing asthma associated with smoking and Chlamydia infection.

IDO1 can also modulate the replication of other obligate intracellular bacteria. For example, the replication of Rickettsia conorii (which can cause diseases such as typhus or Boutonneuse fever) in cultured monocytes treated with IFNγ, TNFα and IL-1β is inhibited in a manner dependent on IDO1-mediated L-Trp depletion [458].

Facultative intracellular bacteria

A number of facultative intracellular bacteria have been linked to IDO1, including species of Francisella, Listeria and Mycobacterium. For these organisms, the ability to synthesise L-Trp also appears important for their ability to withstand IDO1-mediated L-Trp starvation, bestowing a survival advantage, similar to that observed for Chlamydia. For example, the ability to synthesise L-Trp is beneficial for the facultative intracellular bacterium Francisella that can replicate in macrophages and subsequently cause tularaemia, some forms of which can be fatal. In mice, mutant Francisella novicida lacking the tryptophan synthase gene replicate to a similar degree to wild-type F. novicida in the spleen and skin, tissues where IDO1 is not significantly up-regulated in response to infection [459]. However, mutant F. novicida replication is severely attenuated in comparison with wild-type F. novicida in the lung, an organ where IDO1 is highly induced in response to F. novicida infection [459]. This tissue-specific differential between the replication of mutant and wild-type F. novicida was largely absent from IDO1−/− mice, indicating that the ability of these bacteria to synthesise L-Trp confers a survival advantage under conditions of IDO1-mediated L-Trp starvation in vivo.

Another facultative intracellular bacterium, Listeria monocytogenes, is also influenced by IDO1. Although it is ubiquitous throughout the environment, ingestion of Listeria-contaminated food can lead to bacteraemia, infection of the CNS and gastroenteritis in susceptible individuals. IDO1-expressing DCs and macrophages are present in the suppurative granulomas of patients with listeriosis [460] and human monocyte-derived IDO1-expressing DCs have been found to exhibit greater resistance to L. monocytogenes infection and display increased bactericidal activity in vitro [461,462]. This is similar to results observed in human monocyte-derived macrophages, in which L. monocytogenes-induced active IDO1 can also inhibit intracellular infection. This was attributed to the toxicity of Kyn metabolites, as addition of excess L-Trp to cultures (and thus increased Kyn metabolites), as well as exogenous addition of Kyn, enhances macrophage control of L. monocytogenes [462]. In addition, in vitro bacterial cultures of L. monocytogenes were also susceptible to various degrees to exposure to other Kyn metabolites that macrophages can potentially produce including anthranilic acid, 3-HK, 3-HAA, picolinic acid and QA [462]. However, whether the quantities of Kyn metabolites required to alter L. monocytogenes growth in vitro are produced by macrophages in vivo is not clear.

Aside from being present in L. monocytogenes granulomas, IDO1-expressing cells, primarily DCs, are also present within lymph node granulomas induced by Bartonella henselae (a facultative intracellular bacterium that causes cat-scratch disease), tuberculosis (Mycobacterium tuberculosis) and sarcoidosis (of non-infectious origin) [460], and within lung granulomas of M. tuberculosis-infected macaques [463]. This localisation of IDO1-competent DCs and macrophages within the ring-wall formation of granulomas suggests a role for IDO1 in limiting further dissemination or spread of bacterial infection in vivo [460], although this has not yet been explicitly demonstrated. Alternatively, IDO1 expression may contribute to the ongoing chronic inflammation that is associated with granuloma formation by preventing effective clearance of the pathogen and causing continued immune activation at this site.

Mycobacterium tuberculosis is the principal causative agent of tuberculosis, a potentially fatal infectious disease that primarily manifests in the lungs as a granulomatous inflammatory disease, whereas in immunocompromised individuals (e.g. HIV-infected), it can disseminate and infect extra-pulmonary tissues. In the lung, it resides within the endosomal compartment of alveolar macrophages where it evades the cell's antimicrobial actions and is able to persist and replicate. In vitro studies have shown that the growth of M. tuberculosis is suppressed by particular Kyn pathway metabolites (e.g. 100 μM picolinic acid at a pH of 5.5 to mimic the conditions in an acidified phagosome) [464]. However, certain T-cell subsets such as inflammatory Th17 cells appear more sensitive to Kyn metabolites than M. tuberculosis, suggesting that IDO1 may exhibit a more prominent immunosuppressive effect in favour of the mycobacterium. Despite the elevated expression of IDO1 within lung airway epithelial cells, vascular endothelial cells, macrophages and DCs of M. tuberculosis-infected mice [464,465], wild-type and IDO1−/− mice have similar bacterial burdens, inflammatory responses and levels of T-cell activation, with no difference in survival between the two strains [464]. A recent important study [466] provides an explanation for why IDO1−/− mice and wild-type mice have similar bacterial burdens [464]. Zhang et al. [466] established that an important determinant underlying the ability of M. tuberculosis to withstand host immunity is the bacterium's capacity to synthesise L-Trp, thereby making the microbe refractory to IDO1-catalysed L-Trp starvation [466]. Notably, conversion of the microbe into a L-Trp auxotroph via treatment with a pharmacological inhibitor of M. tuberculosis tryptophan synthesis afforded an effective IDO1-dependent antimicrobial host defence [466]. Replication of the mutated M. tuberculosisL-Trp auxotroph in human and mouse macrophages could be restored by exposure to 1-MT, and this mutant bacterium was also able to replicate in macrophages derived from IDO1−/− mice [466].

In addition to being able to survive host cell IDO1 expression, M. tuberculosis may also evade the immune response via IDO1. Thus, mycobacterial α-crystallin antigen (Acr1), which is expressed by M. tuberculosis during latency, enhances M. tuberculosis survival intracellularly and signals for a tolerogenic phenotype in cultured DCs [467]. This tolerogenic phenotype was associated with alteration of PD-L1, IDO1 and IL-10 expression, although these factors were not specifically inhibited to confirm the relative extent of their involvement [467].

The replication of Mycobacterium avium, a bacterium that infects macrophages and is capable of opportunistic infections in immunocompromised individuals, is inhibited in murine bone marrow-derived macrophages treated with ISS-ODNs (immunostimulatory sequence oligodeoxynucleotide analogues of bacterial DNA) in an IDO1-dependent manner [468]. While ISS-ODNs also induce Ido1 mRNA in the lung and spleen of mice, IDO1 was not explicitly shown to be responsible for the anti-M. avium effect of combinatorial ISS-ODN and clarithromycin therapy in vivo.

Infection with the paratuberculosis subspecies of M. avium, which causes a chronic intestinal disease in ruminants, up-regulates Ido1 mRNA in vitro in a human monocyte cell line, as well as induces IDO1 protein expression in vivo within infected sheep, localised to intestinal epithelial cells, granuloma-associated macrophages, and draining lymph node macrophages [469]. Increased IDO1 mRNA expression was also detected in the blood leucocytes from infected sheep and cattle. However, 1-L-MT treatment did not affect M. avium subspecies paratuberculosis-induced proliferation in these cells in vitro when compared with leucocytes from healthy animals, probably due to the finding that this bacterium can also express L-Trp biosynthesis pathway genes and is therefore unaffected by IDO1-catalysed L-Trp depletion [469]. IDO1 expression correlated positively with disease severity in infected sheep, but whether IDO1 has a direct role in limiting the Th1 response that is characteristic of late stage disease is unknown.

Extracellular bacteria

The growth of extracellular bacteria, such as Pseudomonas aeruginosa (which can cause generalised inflammation and sepsis) and Staphylococcus aureus (a commensal bacterium of the skin and mucosa that can cause mild to severe infections, and can also act as a facultative intracellular bacterium), are inhibited in vitro by IDO1-catalysed L-Trp deprivation or upon exogenous addition of Kyn pathway metabolites, namely 3-HK or picolinic acid [186,470472]. Streptococcus agalactiae or group B Streptococcus, a constituent of the normal gastrointestinal flora that can cause septicaemia in newborns, is similarly inhibited by IDO1 in vitro via L-Trp depletion [470,471,473]. The in vitro antibacterial activity of IDO1 translates to an in vivo model of S. agalactiae infection, as IDO1−/− mice exhibited a poorer survival rate when compared with wild-type animals, although the bacterial burden was not specifically analysed [25].

In certain studies, however, IDO1 is linked with the promotion of bacterial infection and replication. For instance, Borrelia burgdorferi, the causative agent of Lyme disease, can induce active IDO1 in human mature conventional DCs and pDCs via IFNα and IFNγ. The relative extents to which different B. burgdorferi genotypes are able to induce IDO1 within these cells correlated with the microbe's ability to cause disseminated infection in vivo, although this was not explicitly shown [474]. IDO1 also contributes to the enhancement of secondary Streptococcus pneumoniae outgrowth in mice initially infected with influenza virus [475]. Furthermore, Escherichia coli-induced IDO1 expression inhibits the transmigration of neutrophils through uroepithelial cells in vitro, perhaps explaining why E. coli-infected IDO1−/− mice exhibit reduced extracellular, but not intracellular, bacterial loads within the bladder [476].

Certain bacteria can also modulate IDO1 expression and activity. L. johnsonii, a non-pathogenic bacterium normally present in the human intestine with probiotic properties, is reported to inhibit constitutive and IFNγ-mediated IDO1 expression due to bacterium-derived H2O2in vitro [42]. Thus, L. johnsonii-infected diabetes-prone rats exhibit reduced IDO1 expression in the intestine and reduced serum Kyn levels [42]. IDO1 does not affect the in vivo growth of L. johnsonii, as the bacterial burden was similar in the stomachs of naïve wild-type and IDO1−/− mice [477]. Whether the ability of L. johnsonii to inhibit IDO1 is responsible for the lack of a role of the enzyme in controlling the growth of the bacterium in vivo is not clear. Interestingly, despite the lack of IDO1 in IDO1−/− animals, AhR can still be activated to promote immune suppression in response to Lactobacillus infection. Thus, Lactobacillus reuteri can generate indole-3-aldehyde (a Trp-indole derivative) under elevated L-Trp conditions apparent in IDO1−/− mice. Indole-3-aldehyde can bind to AhR and elicit IL-22 production, which in turn promotes Candida albicans commensalism, indicating that bacterium–fungus co-operation between commensal microbes is IDO1-dependent [477]. As L. reuteri is a common gut microbe in mammals, this questions whether increased L. reuteri infection and, consequently, increased indole-3-aldehyde can promote infection or commensalism by other micro-organisms.

Clostridium difficile is an obligate anaerobic bacterium that can opportunistically infect the gut and cause colitis. Recent data indicate that C. difficile-infected IDO1−/− mice show increased intestinal tissue destruction and haemorrhage that correlates with elevated numbers of IFNγ-expressing neutrophils, reduced C. difficile burden, but no change in effector or regulatory T-cells [263]. These data indicate that, rather than regulating T-cells and despite an increase in bacterial burden, IDO1 plays a protective role in moderating neutrophil tissue infiltration during C. difficile infection aimed at avoiding excessive inflammatory tissue damage and immunopathology that can be mediated by these innate immune cells.

Helicobacter pylori is a microaerophilic bacterium that resides within the gastrointestinal tract where it is linked to gastritis, gastric and duodenal ulcers and stomach cancer. Very recent data from analyses performed on gastric biopsy samples from human patients suggest an immune-modulatory and protective action of IDO1 against gastritis in H. pylori subjects [478]. This study reported elevated IDO1 expression in H. pylori-infected, but not uninfected, samples that inversely correlates with the severity of gastritis. Moreover, ex vivo inhibition of IDO1 in gastric biopsy cultures with 1-L-MT resulted in the increased expression of Th1-type markers of inflammation (i.e. IFNγ, Tbet and IL-17) and a reduced Th2 marker (IL-4), supporting an anti-inflammatory action of IDO1 [478].

Sepsis

An adverse clinical outcome of bacterial infection is sepsis, a condition characterised by an excessive systemic inflammatory response to generalised infection that can lead to septic shock and organ dysfunction. Clinical and experimental animal studies highlight an important role for IDO1 and the Kyn pathway in response to endotoxic shock. IDO1 expression and activity are increased in septic patients and animals in inflamed cardiac tissue, spleen DCs, bone marrow-derived CD11b+ cells and endothelial cells of resistance vessels [125,182,479,480]. Moreover, IDO1 activity measured as the plasma Kyn/L-Trp ratio is significantly elevated in patients with bacteraemia caused by S. aureus, S. pneumoniae, E. coli and β-haemolytic streptococci, with the extent of increase representing an independent predictor of disease severity and death [481,482].

An initial study in mice challenged with a single lethal dose of LPS (20 mg/kg), a major component of the outer membrane of Gram-negative bacteria, reported that IDO1−/− mice or IDO1 inhibition in wild-type mice with 1-D-MT treatment reduced circulating levels of pro-inflammatory cytokines (TNFα, IL-6 and IL-12), increased anti-inflammatory cytokines (IL-10), and protected against endotoxic shock and lethality [125]. A more recent study reported that IDO1 gene deficiency or 1-D-MT treatment also afforded host protection and reduced mortality in mouse models of bacterial peritonitis or sepsis after caecal ligation and puncture, involving elevated chemokine expression and enhanced recruitment of neutrophils and mononuclear cells into the peritoneum, which correlated with reduced bacterial load [480]. The septic phenotype apparent in wild-type mice was also recapitulated in IDO1−/− mice treated with Kyn or reconstituted with IDO1+/+ bone marrow cells, suggesting that Kyn produced by IDO1-expressing leucocytes plays an important role in promoting sepsis [480]. These studies identify a deleterious pro-inflammatory action for IDO1 during sepsis. A recent study has provided novel insights into the inflammatory signalling pathways employed by IDO1 [259]. Liu et al. [259] reported that IDO1-catalysed L-Trp deprivation and activation of GCN2-dependent signalling in LPS-stimulated macrophages promotes IL-6 and IL-12 production. Moreover, monocytic lineage-specific Gcn2 gene-deficient mice challenged with a lethal dose of endotoxin showed a reduction in IL-6 and IL-12 production and overall mortality [259]. These findings highlight the inflammatory potential of macrophage-expressed IDO1 during sepsis involving the activation of GCN2-dependent metabolic stress signalling. The extent to which the IDO1 and GCN2 signalling in macrophages and other cell types equates to pro-inflammatory actions in other disease states warrants further investigation.

However, although the above studies indicate a pathogenic role for IDO1 in sepsis, a recent important study shows that there is little difference in the survival of wild-type and IDO1−/− mice challenged with a lower LPS dose of 10 mg/kg [25] (cf. 20 mg/kg LPS in [125]). Instead, this LPS model revealed a novel role for IDO1, TDO and AhR cross-talk and signalling in the pre-conditioning of mice against endotoxin shock, where prior exposure to low, non-lethal, doses of LPS induces a protective tolerant response to subsequent exposure to a normally lethal LPS dose. In a raft of studies employing Ido1, Tdo and Ahr gene-deficient mice, Bessede et al. [25] concluded that the initial response to non-lethal low-dose LPS involves the activation of AhR by TDO-derived Kyn, which signals for the down-regulation of pro-inflammatory gene expression. Rechallenge with high-dose LPS results in AhR-induced phosphorylation of IDO1 tyrosine residues and TGFβ expression that together signal for endotoxin tolerance, further minimising the immunopathology that is usually associated with sepsis [25]. These endotoxin-tolerant mice were also protected against subsequent infection with Salmonella enterica Typhimurium (which causes severe invasive salmonellosis) and group B Streptococcus-induced multifocal sepsis [25]. This study not only highlights important roles for the Kyn and AhR signalling pathways in endotoxin tolerance, but also identifies novel cross-talk between the IDO1 and TDO enzymes with the AhR as a critical signalling intermediary.

Therefore, although many animal or human studies linking IDO1 and sepsis show that increased IDO1 activity correlates with sepsis severity, the role of IDO1 in mouse models of sepsis is less clear-cut and appears to be LPS dose- and model-dependent. In accordance with this, a recent study has demonstrated that single exposure of murine conventional DCs (cDCs) to LPS elicits pro-inflammatory IL-6, but not IDO1 expression, whereas consecutive exposure of these cells to LPS up-regulates both IDO1 and TGFβ [483]. Only cDCs receiving consecutive exposure to LPS conferred in vivo protection in mice treated with a lethal LPS dose [483].

The above discussion confirms that IDO1 is a significant response to bacterial infection whose actions are highly contextual and complex. Thus, IDO1 can elicit protective (e.g. inhibition of pathogen infection) and deleterious (e.g. inhibition of the pathogen-specific immune response) actions, depending on the bacterium involved or infection severity (see Tables 4 and 5 for a summary and Supplementary Table S1 for details). Although IDO1-catalysed L-Trp consumption can inhibit the in vivo growth of certain bacteria (e.g. S. agalactiae), other bacteria can withstand the enzyme by synthesising their own L-Trp (e.g. M. tuberculosis) or, in the case of Chlamydia, respond to IDO1 by transitioning into a viable and persistent form that may further exacerbate the degree of inflammation. Furthermore, there is evidence for important roles for IDO1’s immunomodulatory action in bacterial infection, particularly during sepsis, where IDO1 can elicit both beneficial immunosuppressive [25] or deleterious pro-inflammatory [125,259,480] actions. Overall, although it is clear that IDO1 plays a significant role in several clinically relevant bacterial infections or syndromes, including tuberculosis and sepsis, the bulk of the data is either from in vitro cell studies or in vivo mouse models. Accordingly, more human data are needed to clearly establish the relevance of these findings. Notwithstanding the lack of human data, these studies highlight that there is considerable potential for the therapeutic targeting of IDO1 and the Kyn pathway, particularly in combination with inhibition of bacterial L-Trp biosynthesis pathways, in the treatment of certain bacterial infections.

Fungi

Fungi are a diverse group of organisms that include yeasts, mushrooms and moulds and are found ubiquitously in the environment and on human skin. Although less than one-quarter are pathogenic to humans [484], fungi can still cause a variety of diseases, including athlete's foot, candidiasis and aspergillosis, particularly in those that are immunocompromised, e.g. individuals with AIDS or CGD. Research into the role of IDO1 in fungal infection has focused primarily on the Candida and Aspergillus species, both of which do not normally cause disease in immunocompetent individuals. The immune response elicited by both fungi is therefore generally successful at containing the invasiveness of these pathogens.

Candida

The yeast Candida albicans is ordinarily a human commensal of the skin and mucosal surfaces, but can cause opportunistic oral and genital infections (e.g. oropharynhgeal candidiasis and vulvovaginal candidiasis). In vivo, C. albicans infection induces IFNγ-mediated IDO1 expression in the kidneys and stomach of mice, as well as in Peyer's patch DCs and neutrophils [485]. IDO1 inhibition with 1-MT in mice infected with C. albicans intragastrically culminates in increased fungal burden and dissemination of the fungus to organs not normally infected in untreated mice, resulting in enhanced mortality rates [485]. However, when 1-MT treatment is delayed until 3 days post-infection, the deleterious effect of 1-MT is abolished, suggesting that there is a narrow timeframe during which the IDO1 pathway is protective. In contrast with 1-MT treatment, however, a separate study reported that IDO1−/− mice that were infected intravaginally were resistant to C. albicans infection during the first week of infection, but were unable to control the fungal burden after 2 weeks, unlike wild-type mice that showed the inverse effect (i.e. higher fungal burden early, lower burden later) [64]. This supports the idea that the antifungal effect of IDO1 is only evident within a specific timeframe and that, in the later stages, the fungus is able to overcome the host response. The reasons for these apparently disparate findings on IDO1 may reflect different infection administration routes and hence site of infection (in addition to factors such as the dose and yeast strain used). They may also reflect differences in the specificity of action of 1-MT and compensatory gene changes that can occur in Ido1 gene-knockout mice. In C. albicans-infected mice, IDO1 also exhibits immunoregulatory actions including the induction of tolerogenic DCs, Tregs and inhibition of IL-17 production, allowing the establishment of fungal persistence and hence commensalism [244]. Together, these studies highlight the multiple intricate roles that IDO1 may have during fungal infection.

Consistent with a potential beneficial role for IDO1 against human Candida infection, particular IDO1 gene polymorphisms in humans, which are associated with increased IDO1 expression/activity in the vaginal fluid as determined by measuring IDO1 protein levels and the Kyn/L-Trp ratio, correlate with increased protection against recurrent vulvovaginal candidiasis [64]. Nevertheless, how these polymorphisms correspond to local immune or inflammatory status has yet-to-be determined.

Aspergillus

A. fumigatus spores seldom cause disease, despite being ubiquitous in the air, although they can be responsible for significant invasive lung infections in immunocompromised patients. Intranasal infection of mice with A. fumigatus elicits IDO1 expression within neutrophils and DCs in the lung and the draining lymph node [486]. Treatment of A. fumigatus-infected wild-type mice with 1-MT produces increased chitin levels and allergic responses, as well as enhanced IL-17 production by V1γ+ γδ T-cells and reduced numbers of Tregs and IL-10- and TGFβ-producing γδ T-cells, decreasing survival rates [159]. Moreover, mice with chronic granulomatous disease (i.e. mice with defective NOX that are unable to produce an oxidative burst necessary to combat microbes) are reported to express inactive or defective IDO1 protein within the neutrophils isolated from infected lungs, thought to relate to the absence of NOX-derived O2•− necessary for IDO1 activity [159]. Administration of IFNγ and Kyn to A. fumigatus-infected CGD mice reversed this immunopathogenic phenotype [159], indicating that the loss of the immunosuppressive activity of IDO1 and production of Kyn pathway metabolites due to the expression of inactive IDO1 protein underlies the excessive and unrestrained immune and inflammatory response against A. fumigatus in CGD in mice hampering fungal clearance. However, as mentioned above, there is considerable controversy as to whether O2•− is required for active IDO1 in vivo during A. fumigatus lung infection [172], and more research is needed to elucidate the specific cofactors required for optimal IDO1 activity in vivo.

As alluded to in the ‘Chronic inflammatory and allergic disorders’ section above, defective IDO1 expression is also linked to an exaggerated inflammatory response within the lung towards A. fumigatus in a murine model of cystic fibrosis [331]. Accordingly, restoration of active IDO1 with IFNγ or administration of a mixture of Kyn, 3-HK and 3-HAA balanced the inflammatory Th17 response with the immunosuppressive Treg response and was protective in murine cystic fibrosis [331].

A further model supporting a key role for IDO1 in limiting an unnecessarily heightened inflammatory airway response within the infected lung is the mouse model of allergic bronchopulmonary aspergillosis. In this model, the protective actions of dexamethasone were attributed to the ability of glucocorticoids to up-regulate the expression of IDO1, which subsequently inhibited Th2 cells while increasing Tregs to protect against immunopathology [95]. Employing a model of allergic airway disease involving exposure of mice to the A. fumigatus antigen, Paveglio et al. [487] showed that selective overexpression of IDO1 within the airway epithelium reduced lung CD4+ T-cell numbers and their ability to produce cytokines, although this did not reduce disease.

Collectively, it is evident that IDO1 activity during Aspergillus infection within the airways is important in co-ordinating the actions of inflammatory γδ T-cells and Th17 cells, as well as the anti-inflammatory and tolerogenic aspects of the host response [159,244,331,486,487]. Moreover, prominent inflammatory lung disorders such as CGD and cystic fibrosis involving Aspergillus infection are linked to impaired IDO1 expression/activity, suggesting that treatments aimed at elevating functional IDO1 or addition of relevant Kyn pathway metabolites may represent novel strategies useful in the treatment of these diseases.

Recent data indicate that IDO1 is also protective against infection with Paracoccidioides brasiliensis, a fungus that can cause severe systemic granulomatous disease, characterised by a dominant Th2 response, with the milder form of the disease associated with Th1 activation. Mouse strains that show differing susceptibility to P. brasiliensis infection nonetheless display similar extents of increased IDO1 expression within the lung [488]. Inhibition of IDO1 through 1-DL-MT treatment increased early fungal burden in the lung, promoted inflammation and inhibited Th1/Treg immunity, while increasing IL-17 production. However, only in B10.A mice, which are more susceptible to infection than A/J mice, did 1-DL-MT treatment lead to sustained fungal growth, enhanced tissue pathology and increased mortality, indicating that IDO1 appears to have a greater role in susceptible hosts for limiting infection and tissue damage.

Whereas the above indicate an in vivo antifungal role for IDO1, it appears that IDO1 can, however, support the growth of Histoplasma. Thus, in mice infected with Histoplasma capsulatum, the causative agent of the pneumonia-like disease histoplasmosis, IDO1 expression is localised to cells located within the periphery of lung granulomas that are positively correlated with fungal burden and TNFα and IFNγ levels [489]. IDO1 inhibition through 1-MT treatment of H. capsulatum-infected mice reduced lung Kyn, inflammation and fungal burden [489].

Although it is evident that IDO1 induced within host cells can exhibit multifactorial actions in regulating the immune response to fungal infection, fungi themselves can also express IDO1, which is also likely to influence the course of infection. For example, C. albicans expresses fungal IDO1 with 31% homology with mammalian IDO1. In vitro, 1-MT enhances hyphal formation and inhibits C. albicans IDO1 expression, as does L-Trp exposure, suggesting that fungal germination and morphology may be related to IDO1 function [485]. Three IDO1 homologues have similarly been identified in A. fumigatus, and again, exposure to 1-MT promotes fungal germination [490]. The specific role of fungal IDO1 in disease pathogenesis, however, has not been investigated and needs to be taken into account, particularly with studies employing 1-MT, which may inhibit both host-derived and fungal IDO1.

Taken together, although it is clear that IDO1 is important for controlling fungal infection and determining commensalism in vivo in mice (see Tables 4 and 5 for a summary and Supplementary Table S2 for details), more research into how this relates to human fungal infection is required and how IDO1 expression and activity can be therapeutically manipulated to benefit the host.

Parasites

Parasites, such as parasitic worms, mites and protozoa, are organisms that rely on other living host organisms (i.e. plants and animals, including humans) for food and survival. Parasitic infections, which can cause severe and debilitating diseases such as malaria, disproportionally affect those in low-income tropical and subtropical regions [491]. The impact of these infections is exacerbated by the lack of effective vaccines, coupled with difficulties in vector control and the increased drug resistance of various parasites. Hence, there is a need for a more thorough understanding of the host response to parasitic infections that can inform the development of new anti-parasitic agents.

Growing in vitro and in vivo data indicates that IDO1 is an important response to parasitic infection (see Tables 4 and 5 for a summary and Supplementary Table S3 for details). Various in vitro studies have shown that the anti-parasitic actions of certain cytokines relate to their ability to induce IDO1. For example, IFNγ and TNFα have been reported to inhibit Neospora caninum (a parasite causing abortion in livestock) in human cell lines, primary cells and bovine fibroblast-like cells [492,493], Encephalitozoon intestinalis (an opportunistic diarrhoeal parasite associated with AIDS) in mouse enterocyte cells and human epithelial colorectal adenocarcinoma cells [494], Leishmania donovani (an intracellular protozoan that causes leishmaniasis) in human macrophages [441], as well as the growth of T. gondii (which can cause severe toxoplasmosis, particularly in the immunocompromised) in primary human cells and human cell lines, primarily via IDO1-mediated L-Trp starvation [441,471,495500]. The effectiveness of IFNγ-induced IDO1 at inhibiting T. gondii infection in vitro is, however, impaired in cells cultured under low O2 tension [124], i.e. physiological O2 levels similar to that apparent in tissues that normally vary between 3 and 5%, but can fall to <1% in infected tissues. Parasite infection itself may also act to impair antimicrobial signalling. For example, IFNγ-induced IDO1 expression in enterocytes is down-regulated by Cryptosporidium parvum (the cause of cryptosporidiosis of the intestine) via the impairment of IFNγ signalling [501].

In vivo studies identify a more complex role for IDO1 in response to parasite infection, with both anti-parasitic and immunosuppressive activities described; the latter activity is capable of supporting the growth of certain parasites. For example, whereas several studies have consistently found that T. gondii infection of C57BL/6 mice induces IFNγ-dependent IDO1 expression and activity in the lungs and brain [14,15,190], the role of IDO1 in these mice is not clear-cut. Some studies have found that blocking IDO1 function in T. gondii-infected mice with 1-DL-MT [15] or using IDO1-deficient mice [15,502] has no effect on lethality compared with wild-type mice [15,502], despite the reduction in inflammatory cytokine mRNA expression (IL-6 and IFNγ; no change in TNFα) [15]. In contrast, another study found that survival was clearly impaired in mice treated with 1-D-MT or 1-L-MT, which exhibited higher parasite burdens and no change in pro- or anti-inflammatory cytokine expression (TNFα, IFNγ or IL-10) [503]. As both studies employed the same mouse strain, administration route and dose of T. gondii cysts [15,502] the reasons for the difference may relate to the use of different T. gondii strains (i.e. ME49 [503] compared with Fuyaka [15,502]) and other yet-to-be defined factors.

A beneficial anti-parasitic action of IDO1 is observed in vivo following infection of mice with Trypanosoma cruzi, which is responsible for Chagas’ disease in humans. Intraperitoneal inoculation of mice with T. cruzi induced IDO1 expression and activity in the heart, skeletal muscle and spleen, and blockade of IDO1 activity by 1-MT increased parasite burden, pathology and mortality in mice [504].

An immunosuppressive role for IDO1 is apparent in L. major-infected mice where IDO1 expression in pDCs results in the inhibition of T-cell proliferation, inflammatory cytokine release and the development of Th17 cells, thus promoting parasite replication in vivo [505]. Similarly, L. major or L. donovani infection of human monocyte-derived DCs in vitro induces IDO1 expression involving yet-to-be identified autocrine/paracrine factors, which inhibit both polyclonal and antigen-specific T-cell proliferation [506].

The up-regulation of IDO1 in response to Leishmania infection is also observed in human patients with cutaneous leishmaniasis (Leishmania guyanensis, which causes skin sores), who have elevated levels of IDO1 mRNA within the lesions [507], and in those that suffer from visceral leishmaniasis (Leishmania infantum, which usually affects the spleen, liver and bone marrow), who have increased plasma Kyn/L-Trp ratios [508]. Interestingly, this increased plasma Kyn/L-Trp ratio decreased significantly in patients that were successfully treated, although Kyn levels remained above that of the controls [508]. Although there is no direct evidence revealing an in vivo role for IDO1 in human parasitic infection, intralesional IDO1 mRNA expression correlates with FOXP3 mRNA expression, a marker of immunosuppressive Tregs, in chronic cutaneous leishmaniasis [507], whereas the decrease in IDO1 activity in successfully treated visceral leishmaniasis patients correlates with a switch from a Th1 to a Th2 response [508], together supporting a potential immunosuppressive function of IDO1 in human leishmaniasis.

Schistosoma parasite infection causes schistosomiasis, a chronic inflammatory disorder resulting in granuloma formation within the intestine and liver. Studies examining the mechanisms responsible for the increased immunopathology apparent in Schistosoma mansoni-infected IL-4 receptor α−/− mice revealed that, in the absence of M2 macrophages, active IDO1 expressed within IFNγ-stimulated M1 macrophages acts to suppress destructive tissue inflammation [255], highlighting an immunoregulatory role for IDO1 when expressed in macrophages.

Of interest, there is evidence that IDO1 can also promote parasitic infection in an immune-independent manner. For instance, infection of mice with the roundworm Trichuris muris (a model of the human intestinal disease trichuriasis) induces IDO1 in goblet cells of the caecum of immune-deficient SCID (severe combined immunodeficiency) mice and studies with 1-MT provided evidence that IDO1 lowers the turnover rate of colonic epithelial cells, allowing the parasite to persist at the base of the caecal crypts [509]. Eimeria falciformis, which infects mouse intestinal epithelial cells, also takes advantage of IDO1 expression within the host; infection with this parasite induces IDO1 in the caecal epithelium, resulting in the formation of xanthurenic acid that is necessary for normal E. falciformis development [510]. In contrast, other Kyn pathway metabolites can suppress parasite function in vitro, e.g. 3-HK stimulates morphological changes and interferes with the motility of T. cruzi [504]; the implications of this in vivo, however, are unknown. IDO1 induced in response to parasite infection does not always equate to parasite clearance and/or immune regulation; thus, whereas Trichinella spiralis infection induces IDO1 in mononuclear leucocytes of the diaphragm, it is not involved in the eosinophil-mediated suppression of NOS2 and Th2 polarisation of the immune response that reportedly supports infection [511].

Malaria is a severe mosquito-borne disease caused by the Plasmodium parasite. Typical symptoms include fever, chills and malaise that can lead to respiratory distress, organ failure, neurological complications and coma if left untreated. Various lines of evidence link IDO1, L-Trp metabolism and malaria. For example, xanthurenic acid, which is present in the mosquito blood meal of Plasmodium falciparum, acts as a gametocyte-activating factor critical for the development of the malaria parasite in the mosquito vector [512].

In murine models of malaria, inoculation of mice with Plasmodium berghei ANKA and P. berghei K173, which cause cerebral and non-cerebral malaria in mice, respectively, stimulates systemic IFNγ-dependent IDO1 expression and activity in the vascular endothelium and Ido1 mRNA expression in the spleen and liver [513,514]. IDO1 enzyme activity and tissue levels of Kyn, 3-HK, kynurenic acid and QA are also elevated within the brains of mice with either cerebral or non-cerebral malaria. Notably, the ratio of the neuro-excitoxin QA relative to kynurenic acid (an antagonist of QA's neurotoxic actions) was selectively elevated in mice with cerebral malaria, accompanying the appearance of cerebral symptoms [515], findings consistent with a role for QA as a mediator of the neuropathology apparent during cerebral malaria. However, a subsequent study reported no differences in cerebral pathology or survival between wild-type and IDO1−/− mice infected with P. berghei ANKA [516]. Importantly, pharmacological inhibition of Kyn 3-mono-oxygenase (the enzyme responsible for the conversion of Kyn into 3-HK and further metabolism into QA and picolinic acid; Figure 1) yielded partial protection against the development of cerebral malaria in a proportion of infected wild-type mice [516]. This correlated with a reduction in brain levels of picolinic acid, but not QA or kynurenic acid [515518], implying a role for picolinic acid in the neuropathology of cerebral malaria. The alteration in Kyn metabolism apparent during murine models of malaria infection is mirrored in humans, with elevated levels of Kyn, QA, picolinic acid and kynurenic acid detected in the CSF (cerebrospinal fluid) of adults and children with malaria [519,520]. The levels of Kyn metabolites correlate with mortality (QA and picolinic acid), convulsions (QA) and the duration of fever (QA, picolinic acid and kynurenic acid) [520].

To summarise, it is evident that the role of IDO1 in parasitic infection is not straightforward and is parasite-dependent (see Tables 4 and 5 for a summary and Supplementary Table S3 for details). The effect of IDO1 and Kyn pathway metabolites on infection, the immune response and the parasite itself must be considered, as well as the potential impact of infection on IDO1 expression. Moreover, the relationship between IDO1, host defence and immune regulation during parasitic infection in humans warrants further investigation.

Viruses

Viruses are responsible for a wide spectrum of diseases, ranging from the relatively mild common cold to influenza, to cancer and to diseases with extremely high mortality rates such as Marburg haemorrhagic fever and Ebola [521]. Viruses are only able to replicate within living cells; outside these host cells, viruses, which are essentially nucleic acids in a protein coat that may or may not have a membrane, are metabolically inert. This renders viruses particularly susceptible to IDO1-mediated L-Trp deprivation in vitro given that viruses are reliant on the host cell protein synthesis machinery for replication (see Tables 4 and 5 for a summary and Supplementary Table S4 for details). As discussed below, the role of IDO1 in modulating viral replication and immune responses in vivo, however, is more complex, with several examples of viruses employing IDO1’s immunosuppressive activity to their advantage.

Expression of IDO1 in vitro and in vivo in response to viral infection

An increasing number of studies report that infection of cells in vitro with virus alone induces IDO1 expression and activity. For example, IDO1 induction has been reported in human monocyte-derived macrophages infected with HIV [522], the flaviviruses West Nile virus and Japanese encephalitis virus [108,523] and Epstein–Barr virus [94], in human pigment retinal epithelial cells infected with West Nile virus [523], and influenza virus-infected primary human airway epithelial cells [81,524] and a mouse lung epithelial cell line [81].

Studies are beginning to unravel the stimuli and cell signalling pathways responsible for virus-induced IDO1 expression in cultured cells. Whereas the ability of HIV infection to induce IDO1 in human monocyte-derived macrophages requires IFNγ [522], HIV up-regulates IDO1 in human PBMCs (peripheral blood mononuclear cells) and murine organotypic hippocampal slice cultures in an IFN-independent but p38 MAPK-dependent manner [525,526]. IDO1 induction in HIV-infected human pDCs involves non-canonical NF-κB activation [109]. Similarly, we have shown that West Nile and Japanese encephalitis flaviviruses require NF-κB, together with the cellular production of TNFα, to induce robust IDO1 expression in human monocyte-derived macrophages in an IFNγ-independent manner [108]. Of note, TNFα alone does not induce IDO1 in human macrophages [108], indicating that, for flavivirus infection, other yet-to-be-identified factors such as viral components and/or other cytokines are required for IDO1 induction. With respect to this, Liu et al. [94] recently reported that the ability of Epstein–Barr virus to induce IDO1 in human monocyte-derived macrophages required TNFα and IL-6 synthesis, both of which up-regulated IDO1 expression via NF-κB and p38 MAPK signalling pathways. IFNλ, expressed by human and mouse lung epithelial cells following influenza infection, also elicits IDO1 expression and activity [81].

Numerous studies have described the induction of IDO1 in response to in vivo viral infection. An initial study in 1979 by Yoshida et al. [527] reported the robust induction of IDO1 and the Kyn pathway in the lungs of influenza virus-infected mice, a finding subsequently replicated by others [475,528530]. Recent data indicate that lung influenza infection elicits IFN-independent IDO1 expression in haematopoietic cells of the draining lymph node, whereas IDO1 expression in the non-haematopoietic epithelial cells of the lung is IFNγ-dependent [530].

Various other viruses also induce IDO1 in vivo in mice, humans and other primates. Thus, IDO1 is up-regulated in vivo following infection with HSV-1 (herpes simplex virus type 1) [531], poliovirus (which causes poliomyelitis) [532], hepatitis C [533,534], dengue virus [535], Epstein–Barr virus (which causes glandular fever) [536], encephalomyocarditis virus (which causes encephalitis and myocarditis primarily in pigs) [537,538], LP-BM5 murine leukaemia virus (murine AIDS) [502], SIV (simian immunodeficiency virus) [539] and HIV [540543].

These studies suggest that the stimuli responsible for IDO1 expression in response to infection are not only dependent on the virus, but also cell-dependent. More studies into how virus infection induces IDO1 are warranted, e.g. whether it is via autocrine/paracrine cytokine expression and/or direct induction by viral nucleic acids. Given that viral infections often provoke an IFN/TNFα response in vivo, it is not unlikely that these and other related cytokines such as IL-6 are required for virus-mediated IDO1 induction, despite some viruses having evolved to evade antiviral IFN/TNFα responses [544]. Therefore, whereas the host may up-regulate IDO1 to control viral infection, viruses may circumvent this by using IDO1 as a component of their immune evasion strategy, as discussed below.

Role of IDO1 in virus infection in vitro

In vitro studies have consistently shown that viruses are particularly susceptible to IDO1-mediated L-Trp degradation, which is perhaps not surprising given that viruses must employ the host cell protein synthesis machinery in order to replicate. IDO1-mediated inhibition of viral replication in vitro is particularly apparent when cells are treated with IFNs before infection, meaning that active IDO1 protein is expressed before virus exposure. The first case in which the antiviral nature of IDO1 was documented was in IFNα- and IFNβ-treated primary human retinal epithelial cells infected with cytomegalovirus [545], which commonly causes a mild flu-like disease in the normal population, yet immunocompromised individuals can have more significant consequences such as encephalitis. This inhibition was due to the depletion of L-Trp, since replication was restored upon supplementation of L-Trp to the culture medium. Likewise, L-Trp depletion due to IDO1 activity is also responsible for the IFNγ-mediated, but not IFNα- or IFNβ-mediated, suppression of HSV-1 and -2 in vitro [78,546], although the IDO1-mediated inhibition of HSV-1 infection is abrogated during hypoxia [124], which inhibits IDO1 expression and activity [124]. In vitro replication of vaccinia (used in smallpox vaccination) virus, dengue virus, measles virus and HBV (hepatitis B virus) are all limited by IFNγ, in a manner that is either wholly or partially dependent on L-Trp deprivation [535,547549]. We recently showed that West Nile virus replication is strongly inhibited in human monocyte-derived macrophages cultured in L-Trp-free medium or in HeLa cells transfected with the human IDO1 gene that affords L-Trp deprivation [108].

Currently, the vast majority of studies examining the role of IDO1 as an antiviral mechanism in vitro have focused on IDO1 as an antiviral effector mechanism of IFNs. It is clear from these studies that IFNγ-stimulated induction of IDO1 and L-Trp metabolism before infection is antiviral in vitro. What is less apparent is the role of IDO1 in controlling viral replication when it is induced in response to viral infection alone. For example, we have shown that, although IDO1 is robustly induced within 48 h of infection, IDO1 is non-essential for human monocyte-derived macrophages to effectively control the replication of established flavivirus infections, which is instead controlled by other yet-to-be defined antiviral mechanisms [108]. This is despite flaviviruses being sensitive to IDO1-mediated L-Trp starvation [108]. This raises the important question of the potential role for IDO1 up-regulated subsequent to control of the virus by the host cell. Our recent data noted that in human monocyte-derived macrophages exposed to West Nile virus, IDO1 was primarily expressed in non-infected cells, suggesting that the delayed up-regulation of IDO1 may act to limit viral spread by promoting local L-Trp depletion and thus creating an antiviral microenvironment surrounding the infectious foci [108]. Alternatively, the relative absence of IDO1 in West Nile virus-infected cells may also reflect that the replicating virus itself may inhibit the up-regulation of IDO1 in infected cells via an as-yet unknown mechanism; how this relates to the pathology of disease is also unknown. A precedent for this, however, is apparent with other viruses; although the replication of cytomegalovirus is susceptible to IDO1-mediated L-Trp depletion, the virus itself can block IDO1 expression in human fibroblasts and mesenchymal stromal cells, thus inhibiting the IDO1-mediated immunosuppressive and antimicrobial effects of these cells [550,551], illustrating the complex interactions that can occur between IDO1 and viral infection.

In contrast, recent in vitro data from a murine alveolar epithelial cell line (MLE-15 cells) and primary human bronchial epithelial cells show that influenza induces IDO1 in these cells via IFNλ (type III IFN) and that IDO1 induced in this manner acts to promote infection [81]; whether this is a reflection of the in vivo situation, however, is unknown.

Virus-induced IDO1 may also elicit immune suppression in vitro. For example, IDO1 induced in human monocyte-derived macrophages in response to infection with Epstein–Barr virus effectively inhibits T-cell proliferation and impairs the cytotoxic activity of CD8+ T-cells in vitro [94]. Such IDO1-mediated immune suppression could conceivably act to protect against further inflammatory tissue damage and immunopathology once the virus has been controlled by alternative earlier-acting antiviral defence mechanisms. Alternatively, however, the virus may employ the immunosuppressive action of IDO1 to protect itself from the host immune defence. As discussed below, in vivo evidence for the latter scenario is evident in several viral infections.

Role of IDO1 in virus infection in vivo

Despite the clear antiviral effect of IDO1 when the enzyme is induced before viral infection in many in vitro studies, the situation in vivo differs in that viral infection commonly precedes the up-regulation of IDO1. Under these in vivo conditions, emerging data indicate that IDO1 and Kyn pathway metabolism primarily function to control the immune response to virus infection. For example, infection of mice with LP-BM5 murine leukaemia virus, which can lead to murine AIDS, stimulates IDO1 expression and activity in the spleen, lung, liver and brain [502]. Ido1 gene-deficient mice exhibit reduced LP-BM5 viral load due to an enhanced type I IFN production [502], although a more recent study reports no difference in viral burden between wild-type and IDO1−/− mice [552]. In encephalomyocarditis virus infection, which induces IDO1 in microglia, Purkinje cells [538], spleen and heart, Ido1 gene deficiency affords milder disease, linked to increased IFNα and IFNβ expression and a corresponding reduced viral load [537]. Administration of Kyn, 3-HAA and 3-HK to IDO1−/− mice restored disease severity, with these mice exhibiting more severe myocardial damage and succumbing to infection to a greater degree than wild-type mice [537], demonstrating that elevating Kyn metabolites alone significantly exacerbates encephalomyocarditis. How the concentrations of the Kyn metabolites exogenously administered to mice compare with those produced in vivo in infected wild-type mice requires clarification.

Influenza virus

Although influenza virus was reported to induce the robust expression of IDO1 within the lung of infected animals back in 1979 [527], insights into the potential role of the enzyme in influenza infection have only recently been forthcoming. In murine influenza, inhibition of IDO1 activity in wild-type mice with 1-DL-MT or infection of IDO1−/− mice, enhanced the primary immune response to infection by increasing the production of macrophage-derived pro-inflammatory cytokines, promoting CD4+ and CD8+ T-cell function, as well as Th1 responses [529,530]. Secondary immune responses are also affected, as 1-DL-MT treatment or Ido1 gene deficiency also resulted in elevated effector CD4+ Th1 and Th17 responses, increased the number of effector memory CD8+ T-cells, altered the memory CD8+ T-cell repertoire and decreased numbers of CTLA-4-expressing Tregs [529,530,553]. IDO1 inhibition is also associated with improved tissue repair in influenza [553]. However, despite these clear enhancements of primary and secondary immune responses, there is no apparent difference in lung viral burden or morbidity recorded between untreated and 1-DL-MT- or 1-D-MT-treated wild-type mice [529,530,553], although infected IDO1−/− mice exhibit improved survival and more rapid recovery from influenza virus infection [530]. These discrepant lethality results between the 1-MT and IDO1−/− studies may be reconciled by the fact that, although the level of replicating virus in the lung declines by day 6 and is below the level of detection by day 7, 1-MT treatment does not inhibit IDO1 activity, as indexed by the Kyn/L-Trp ratio, until day 10 post-infection [529]. It may also reflect partial inhibition of IDO1 with 1-MT compared with complete inhibition of IDO1 expression in IDO1−/− mice.

Interestingly, the immune-modulatory activity of IDO1 induced during influenza infection also plays a critical role in the enhanced susceptibility of mice to S. pneumoniae infection, which causes pneumonia. Thus, 1-MT treatment of mice previously infected with influenza resulted in a marked reduction in secondary pneumococcal outgrowth [475]. Providing further support for the ability of IDO1 to suppress immune function during influenza infection, administration of the adjuvant α-galactosylceramide in conjunction with inactivated influenza A virus inhibited virus-specific CD8+ T-cell responses via NKT-cell-dependent IDO1 expression [93]. These studies imply an overall deleterious role of IDO1 during influenza infection and secondary bacterial pneumonia linked to the immunosuppressive activity of IDO1 in vivo.

Hepatitis virus

IDO1 is increasingly linked to hepatitis infection, the most common cause of liver cirrhosis and hepatocellular carcinoma. Increased levels of IDO1 mRNA are expressed in the liver of patients with chronic hepatitis C, who have higher Kyn/L-Trp ratios in the serum than healthy controls [533], as do chronic hepatitis B patients [533,534]. This increase in serum Kyn level correlates positively with the extent of liver fibrosis, inflammation and insulin resistance and negatively with the platelet count, findings indicating that elevated IDO1 activity is associated with disease progression in humans [534,554].

From a mechanistic point of view, in vitro evidence suggests that, whereas active IDO1 has no role in IFNγ-mediated inhibition of hepatitis C RNA replication in Huh7 cells [533], IDO1-mediated L-Trp depletion inhibits HBV replication in a human hepatocyte cell line [549]. In vivo studies, however, point to IDO1 exhibiting a more dominant role in suppressing hepatitis immunity, particularly during the chronic phase of infection, which correlates with weak CD8+ T-cell responses in human patients [555,556]. Higher levels of IDO1 activity (as determined by plasma Kyn/L-Trp ratios) in haemodialysis patients vaccinated with the recombinant HBV vaccine were associated with an inadequate immune response to the vaccine (i.e. low serum-specific antibody levels) when compared with patients with lower levels of IDO1 activity [557]. More direct evidence for IDO1’s hepatitis-specific suppressive activity was shown in a mouse model where joint administration of α-galactosylceramide and hepatitis B antigen to IDO1−/− mice elicits enhanced production of HBV-specific antibody compared with wild-type mice [558].

IDO1 activation may also contribute to tissue pathology in a murine fulminant hepatitis model, in which HBV-specific CD8+ T-cells are transferred into HBV-transgenic mice [559]. In this model, 1-D-MT treatment, Kyn administration and use of HBV-transgenic/IDO1−/− mice determined that IDO1 induction in the liver and the subsequent production of Kyn metabolites was responsible for the increased liver damage. Conversely, in a non-viral α-galactosylceramide-induced hepatitis mouse model using 1-D-MT-treated wild-type mice and IDO1−/− mice, IDO1 has been shown to limit liver inflammation and injury via the suppression of TNFα-expressing CD49b+ and CD11b+ cells [560]. This is similar in mice infected with mouse hepatitis virus, where 1-L-MT treatment increased liver damage and mortality [561]. Therefore, it is evident that IDO1 can display both protective and damaging qualities depending on the hepatitis animal model, indicating once more the complexity of IDO1’s function in vivo.

HIV

The pathogenesis of HIV is associated with the induction of immune deficiency involving the depletion of CD4+ T-cells and unrestrained chronic T-cell activation, which can eventually manifest as AIDS. A role for IDO1 as an immunosuppressive enzyme during HIV infection capable of dampening antiviral T-cell responses and contributing to virus persistence has received considerable attention. Several studies report that HIV patients (or rhesus macaques infected with the SIV equivalent) up-regulate IDO1 expression and activity in the CNS, resulting in decreased L-Trp and increased Kyn, QA, kynurenic acid and 3-HK in the serum, CSF, the frontal cortex, lungs, lymph nodes and colon [539543,562,563]. This elevated IDO1 activity correlates with progressively poorer disease outcomes such as increased neuropsychological deficits, cognitive and motor dysfunction, and the development of AIDS-associated dementia and mortality [539543,563].

Similar to other viruses discussed above, experimental studies indicate that HIV can take advantage of the elevated IDO1 in vivo, employing the enzyme to suppress the virus-specific immune response. For example, a feature of HIV-positive subjects is an imbalance of antiviral Th17 cells and immunosuppressive Tregs in favour of the latter cell type, which correlates with the increased expression of IDO1 within myeloid DCs [564]. In vitro studies in human T-cells show that exogenous 3-HAA treatment enhances the numbers of Tregs while decreasing the proportion of Th17 cells, indicating that this Kyn metabolite has the potential to signal for this T-cell phenotype imbalance that is associated with the maintenance of the chronic inflammatory state and the development of AIDS [564].

Further evidence for an immunosuppressive action of IDO1 in response to HIV are in vitro studies where HIV-infected human pDCs express active IDO1, eliciting the differentiation of naïve CD4+ T-cells into functional Tregs that suppress the maturation of human monocyte-derived conventional, but not plasmacytoid, DCs [525,565,566]. Potula et al. [567] found that SCID mice reconstituted with human peripheral blood leucocytes and injected intracranially with HIV-1-infected human monocyte-derived macrophages develop HIV encephalitis. These mice fail to control infection despite the recruitment of HIV-1-specific CD8+ T-cells. However, inhibition of the IDO1 pathway with 1-MT treatment enhances the trafficking of these CD8+ T-cells to the site of infection, correlating with greater clearance of virus and infected monocyte-derived macrophages, and thus provides evidence of an additional IDO1-mediated avenue by which HIV can evade the virus-specific immune response [522,567].

Correspondingly, inhibition of the IDO1 pathway with 1-D-MT in SIV-infected macaques synergised with anti-retroviral therapy to reduce the in vivo viral burden, supporting the therapeutic potential of targeting IDO1 in combination with standard anti-retroviral therapy during HIV infection (treatment of SIV-infected macaques with 1-D-MT alone was without effect) [568]. Interestingly, this study also reported that 1-D-MT elevated IDO1 and TGFB mRNA expression within lymph nodes [568], providing in vivo evidence for a compensatory up-regulation of IDO1 in response to 1-D-MT, similar to that noted in 1-D-MT-treated cancer cells in vitro [412]. CTLA-4 expressed on Tregs appear to be an important signal for elevating IDO1, as treatment of SIV-infected macaques receiving anti-retroviral therapy with a neutralising anti-CTLA-4 antibody reduced the in vivo expression of IDO1 and TGFβ, correlating with reduced viral levels and increased virus-specific effector T-cells [569]. A more recent study, however, raises caution with respect to the safety of combinatorial targeting of IDO1 and CTLA-4: co-administration of 1-D-MT and an anti-CTLA-4 antibody in SIV-infected macaques also receiving anti-retroviral therapy and vaccination resulted in the development of severe pancreatitis and diabetes in all animals [570]. The above studies highlight that while targeting the IDO1 pathway is of potential therapeutic benefit, complexity surrounds IDO1 pathway inhibition in combination with other immunotherapeutic and antiviral therapies during HIV infection that warrants further investigation.

To conclude, it appears that the induction of IDO1 is a common feature of several clinically relevant viral infections, including influenza, flaviviruses and HIV. Induction of IDO1 in vitro before infection is predominantly antiviral due to its ability to deplete L-Trp (and if induced later may act to limit further viral spread). However, this is not necessarily borne out in in vivo animal models where virus-induced IDO1 and resultant L-Trp metabolism along the Kyn pathway exhibits an immunosuppressive role benefiting the virus (see Tables 4 and 5 for a summary and Supplementary Table S4 for details). This also follows for other microbial pathogens, as described in the previous sections, where IDO1 has a largely antimicrobial role in vitro that is less commonly observed in vivo. Although there is in vivo evidence for an antimicrobial role for IDO1 (e.g., Chlamydia [441], Leishmania [441] and HSV [78,546]), these reports are in the minority, which may reflect that the extent of IDO1 induction is insufficient to reduce in vivoL-Trp to levels necessary to suppress local microbial growth, which in some cases enables microbial persistence, as noted for Chlamydia. The ability of certain pathogens themselves to synthesise L-Trp from indole and other substrates available in the host further mitigates the efficacy of IDO1-catalysed L-Trp starvation in inhibiting microbial growth in vivo. In microbes capable of withstanding IDO1-mediated L-Trp metabolism, considerable evidence supports a predominant immunosuppressive action of the enzyme in vivo. Depending on the pathogen, this inhibition of the immune response can benefit the pathogen (e.g. microbial persistence and commensalism that could elicit sustained immune activation whereby the host attempts to clear infection, albeit ineffectively, leading to chronic inflammation and immunopathology or evasion of the pathogen-specific immune response, similar to tumour cells evading a tumour-specific response) or the host (e.g. dampening of the immune response once infection is resolved to avoid immunopathology).

The discrepancies between the in vitro and in vivo studies are reflective of the isolated nature of cell culture models compared with animal models where the intricate pathogen–host interplay and influence of other pathways capable of modulating both IDO1 and microbial growth (e.g. NOS-derived NO) can be taken into consideration. Therefore, as a key response to microbial infection and control mechanism in host immunity, it is evident that in vivo models are critical to uncover the specific role(s) of IDO1 in response to different microbial infections, particularly as the function of IDO1 in infectious disease remains largely unclear in humans. A more complete understanding of the role of IDO1 during specific infectious diseases can inform the utility of relevant therapeutic strategies aimed at modulating IDO1 in an attempt to eradicate the microbe and at the same time maintaining appropriate and effective antimicrobial host immune responses.

Other biological actions of IDO1 and the Kyn pathway

Vascular tone and blood pressure

The regulation of vascular tone by IDO1 during inflammation has recently emerged as a new area of study. Recent reports indicate that certain inflammatory conditions are associated with the IFNγ-dependent expression of IDO1 within the vascular endothelium and that this can influence endothelial function, viability and vascular tone. Wang et al. [182] reported that local production of Kyn as a result of IDO1 expression within the endothelium of resistance vessels following malaria infection or LPS-induced endotoxaemia in mice induces the relaxation of vessels via the activation of the adenylate and soluble guanylate cyclase pathways, and, consequently, hypotension. IDO1 expressed in the endothelium also mitigates hypoxia-induced pulmonary hypertension in rats [54]. Furthermore, active IDO1 is expressed in the brown, and to a lesser extent white, adipose tissue surrounding rat arteries. Fat-mediated protection against vascular constriction induced by angiotensin II is inhibited by 1-MT in the thoracic aorta (surrounded by brown fat), but not the mesenteric artery (surrounded by white fat). Vasorelaxation, in this case, is thought to be the result of IDO1-mediated QA production rather than Kyn, as QA was more efficient at attenuating vasoconstriction in this study [41].

In humans, IDO1 is induced within the vascular endothelium of disease states such as pre-eclampsia [571], as well as in resistance vessels and cardiac tissue during septic shock in response to various infections including S. aureus, E. coli, Enterobacter cloacae, Serratia marcescens, Legionella pneumophila, Enterococcus spp., Streptococcus spp., and coagulase-negative Staphylcoccus and Klebsiella [479]. In human sepsis, IDO1 expression and activity positively correlates with the degree of hypotension [479]. These human findings, together with the observations of a hypotensive role for IDO1 in septic mice [182], suggest that endothelium-expressed IDO1 contributes to hypotension that is a feature of sepsis. Elevated circulating Kyn pathway metabolites also correlate with the extent of hypotension apparent in patients resuscitated from a heart attack over the initial 24 h period post-cardiac arrest [338]. Furthermore, obese patients exhibit greater hypotension than lean subjects, correlating with higher Kyn/L-Trp ratios and higher expression of IDO1 in the liver and white adipose tissue, presumably reflecting chronic low-grade immune activation in obese patients [572].

A recent study found that IDO1 induction and intracellular production of 3-HK in endothelial cells activates NOX and the resultant production of ROS, oxidative events leading to endothelial apoptosis [183]. 3-HK is thought to act via its ability to form oxidative adducts with endothelial proteins, which in the case of NOX, is considered to activate the enzyme resulting in increased O2•− production and endothelial dysfunction [183]. IDO1 is therefore causally linked to NOX-dependent endothelial apoptosis and dysfunction in mice infused with angiotensin II [183]. However, the induction of apoptosis by endothelium-expressed IDO1 is not necessarily always detrimental; overexpression of human IDO1 in the lung endothelial cells in a mouse model of pulmonary hypertension, characterised by the excessive proliferation and impaired apoptosis of pulmonary arterial SMCs, attenuates the disease as IDO1 promotes the apoptosis and inhibits the proliferation of SMCs [54]. Notably, however, this is in contrast with the rat lung allograft and isograft model, where targeted overexpression of human IDO1 in the rat lung enhanced the resistance to apoptosis and necrosis due to the inhibition of ROS formation [207].

Therefore, the above highlights the potential of IDO1 when expressed in the endothelium or periadventitial fat of arteries to regulate vascular tone and blood pressure indicating that targeting IDO1 activity and the Kyn pathway represents a plausible strategy for modulating vascular function during several inflammatory conditions such as sepsis, hypertension and potentially obesity. Further work is necessary, however, to define further the extent and molecular mechanisms by which endothelium- or adipose tissue-expressed IDO1 influence vascular function and homoeostasis during different inflammatory vascular disorders and the utility of targeting IDO1 in order to improve the vascular dysfunction apparent in these disorders.

Whereas the above indicate a functional role for vascular-expressed IDO1, a recent study reported that despite elevated IDO1 expression in cerebral arterioles in mice subjected to transient cerebral ischaemia/reperfusion, 1-MT treatment or Ido1 gene deficiency did not significantly influence the degree of cerebral injury or function [573].

Neurological disorders

IDO1 and the generation of neuroactive Kyn pathway metabolites have been implicated in the pathogenesis of several neurodegenerative and psychiatric disorders, including depression, bipolar mania, multiple sclerosis, Parkinson's disease, Huntington's disease and Alzheimer's disease. This area has been the subject of several comprehensive reviews [574576] and is thus only described briefly in the present review.

QA, generated via the Kyn pathway in microglia, is a highly potent NMDA receptor agonist and, under steady-state conditions, is counterbalanced by the presence of NMDA receptor antagonists such as kynurenic acid (which also binds to the α7-nicotinic acetylcholine receptor [20]), formed in astrocytes [577]. Thus, dysregulated activation of the Kyn pathway, causing an imbalance of QA, kynurenic acid and other neuroactive L-Trp metabolites in the CNS, is linked to a spectrum of neurological effects ranging from cognitive deficits (e.g. high kynurenic acid levels in schizophrenia) to neurodegeneration (high QA levels in Huntington's disease and Alzheimer's disease) [575].

Low kynurenic acid levels are also observed during major depression, an area of research that is growing rapidly [576]. Increased IDO1 expression may contribute further to depression by reducing the quantity of L-Trp available for serotonin synthesis by diverting L-Trp for metabolism along the Kyn pathway. Moreover, IDO1 itself can degrade serotonin into formyl-5-hydroxykynuramine, further decreasing the level of serotonin, which is critical for optimal serotonergic signalling in the brain [578].

Therefore, although it is clear that IDO1 is involved in neurological disorders, the level of Kyn in the CNS (either produced endogenously or originating from the periphery), together with the relative expression levels of the Kyn pathway enzymes responsible for QA and kynurenic acid production, represent important determinants of disease severity and pathogenesis [575].

Age-related cataracts

A cataract is a clouding of the lens that can impair vision and ultimately lead to blindness. IDO1 and Kyn metabolites have been implicated in both protecting the lens from UV damage, and, conversely, in age-related cataract formation [579].

Active IDO1 protein is expressed in lens epithelial cells of the anterior cortex [580,581], with lenses from donors 26–80 years of age having similar levels of IDO1 activity [580]. IDO1-derived Kyn and 3-HK, as well as 3-HKG (3-hydroxy-L-kynurenine glucoside) and 4-(2-amino-3-hydroxypheny)-4-oxobutanoic acid glucoside derived from 3-HK, are present in human lenses to filter out and protect the eye from damaging 300–400 nm UV light [579,580,582,583].

However, during the normal aging process, the formation of these normally protective UV filters are implicated in the development of age-related cataracts by oxidising and forming adducts with the lens crystallin proteins (called kynurenilation) following exposure of the filters to UV light or in the presence of oxygen [584586]. In addition, UVA-excited free and lens protein-bound kynurenines (NFK, Kyn or 3-HK) can oxidise the antioxidant ascorbate in human donor lenses and lenses from transgenic mice overexpressing the human ascorbate transporter and IDO1 [587]. The resulting ascorbate oxidation products in turn promote the formation of advanced glycation end-products that contribute to cataract development [587].

Although the antioxidant glutathione (GSH) can protect against the kynurenilation of the lens by forming Kyn–GSH adducts in vitro with calf lens proteins [588] and 3-HKG–GSH adducts in human lenses [589], the protection conferred by GSH decreases with age due the reduced diffusion of GSH to the centre of human lenses from older donors compared with younger donors [590]. Furthermore, GSH is not effective at preventing or reversing NFK-, Kyn- and 3-HK-induced oxidation of ascorbate in vitro in human lenses [587].

The anti-proliferative effect of IDO1 activity is also evident in vitro in mouse lens epithelial cells overexpressing human IDO1, with the induction of cell cycle arrest correlating with increased Kyn production and formation of Kyn-modified proteins [581]. This inhibition of proliferation was reversed by 1-DL-MT treatment [581]. In contrast, when compared with lens epithelial cells, mouse lens fibre cells overexpressing human IDO1 are more susceptible, with IDO1-mediated fibre cell apoptosis evident even in pre-cataract lenses [591].

Of note, these mice overexpressing human IDO1 exhibit significant morphological changes in their lenses, with 3-month-old mice exhibiting dense cataracts, abnormal fibre cell differentiation and smaller overall eye size [591].

Therefore, even though IDO1-derived Kyn metabolites protect against UV-related damage in the lens, these metabolites over time can form adducts with lens proteins and induce the formation of advanced glycation end-products, both contributing to the development of cataracts. Whether IDO1 inhibitors mitigate cataract formation has yet to be studied in animal models or investigated in human trials.

Bone remodelling

The activity of osteoclasts, cells that resorb bone, must be tightly controlled to avoid excessive bone loss. IDO1 has been recently identified as having a novel and key role in bone metabolism; Tregs and soluble CTLA-4-Ig inhibit the differentiation of osteoclast precursors in an IDO1-dependent manner by inducing apoptosis of these cells [592]. Conversely, IDO1 promotes the differentiation of osteoblasts, cells that are responsible for bone formation [593]. Accordingly, IDO1−/− mice have more osteoclast precursors in the spleen and bone marrow, as well as increased numbers of osteoclasts and decreased numbers of osteoblasts in the bone perimeter, contributing to the osteopenic (low bone density) phenotype of IDO1−/− mice [592,593].

Importantly, these data are reflected in humans: rheumatoid arthritis patients also have increased numbers of osteoclast precursors compared with healthy individuals, an increase that is reversed by abatacept, a CTLA-4-Ig-fusion protein, but not methotrexate or TNFα inhibition, each of which are arthritis treatments [592]. Conversely, melanoma patients undergoing ipilimumab treatment, a CTLA-4 inhibitor, have greater numbers of osteoclast precursors than the healthy controls [592]. These findings have clear implications not only for osteoporosis treatment, but also for the use of CTLA-4-modulating therapies in autoimmune diseases and cancer, in which the potential link of IDO1 with the dysregulation of bone remodelling should be considered.

CONCLUSION

The present review clearly illustrates that IDO1 is a fascinating multifunctional haem enzyme of clear importance to health and disease. Although IDO1 was originally recognised as an in vitro antimicrobial and anti-tumorigenic effector mechanism of IFNs due to its ability to starve pathogens or malignant cells of L-Trp and inhibit their proliferation, the importance of IDO1’s immune regulatory functions has been firmly established within the last 18 years, ranging from beneficial immune suppression in various chronic inflammatory disorders, autoimmune diseases, transplantation and pregnancy to deleterious IDO1-dependent suppression of tumour- or pathogen-specific immune responses.

Not surprisingly, it is becoming increasingly apparent that the immune-regulatory actions of IDO1 are complex. Thus, IDO1 is not always necessarily immunosuppressive, with several reports of an immune-stimulatory or pro-inflammatory role of IDO1 in certain in vivo contexts, including cancer [403], asthma [246] and arthritis [251,330]. Moreover, IDO1 engages several molecular effector mechanisms to mediate immune control that may act either alone or in combination, including enzyme activity-dependent (i.e. L-Trp deprivation-induced activation of the GCN2 stress pathway or Kyn-dependent AhR activation) and enzyme-activity-independent (i.e. via IDO1 tyrosine phosphorylation) signals. Although the initial focus was on how IDO1-expressing APCs suppressed T-cell activation, it is now clear that IDO1 controls the function and viability of various immune cell types/subsets including B-cells, NK cells, DCs, macrophages and neutrophils. In some cases, IDO1 can have seemingly opposing effects in different cell types, e.g. whereas IDO1-dependent GCN2 activation suppresses T-cells [225], it has recently been reported to activate inflammatory cytokine production in macrophages [259].

In the light of this complexity, it is critical that future research understands various important factors surrounding IDO1 immune control in vivo and how these differ depending on disease type and stage. Thus, for different pathologies, it is important to define the cell type(s) expressing IDO1, the immune stimuli and transcriptional events modulating IDO1 expression (e.g. loss of expression of the transcriptional repressor Bin1 is linked to elevated IDO1 in tumour cells [120]), the IDO1-dependent effector mechanisms controlling the immune response, the identity of the IDO1-responding or target immune cells and the relevant respondent signalling events altering immune cell function and phenotype.

Importantly, to date, many of these factors have been identified in murine cells and experimental animal models of disease, and as such, their relevance to human disease requires clarification. Unravelling these various complexities regarding IDO1-mediated immune control will inform the design of relevant therapeutic strategies to suppress or enhance IDO1’s immunomodulatory actions in vivo.

Also critical to the further development of more efficacious and selective IDO1-modulatory drugs is the continued advancement of our understanding of the fundamental biochemistry of IDO1 including enzyme protein structure, dioxygenase reaction mechanism and the nature of the intracellular cofactor(s) supporting IDO1 dioxygenase activity. Recent advances show that IDO1 is not only a dioxygenase, but also capable of catalysing other reactions, namely haem peroxidase [30] and peroxygenase [202] reactions. The implications of these and other novel enzyme activities for IDO1’s biological actions, however, require clarification. Further adding to the complexity of targeting IDO1 is the recent discovery of enzyme-activity-independent signalling actions involving IDO1 tyrosine phosphorylation [113], which is likely to be refractory to inhibitors of IDO1’s enzyme activities and instead require inhibition of the kinases responsible for IDO1 tyrosine phosphorylation.

Although the majority of research has focused on IDO1’s immune-regulatory function, IDO1 can also elicit various non-immune actions. For example, IDO1 induction and the production of Kyn pathway metabolites are linked to neurological disorders [574576], formation of cataracts [584587], regulation of vascular tone and blood pressure [182] and bone remodelling [592]. Future studies will no doubt uncover further novel roles for IDO1 and the Kyn pathway.

Although IDO1 has gained considerable attention as an immune-control mechanism influencing health and disease, it is important to recognise that the bulk of findings stem from studies with experimental animals. As such, the importance of IDO1-mediated immune regulation in the corresponding human situations is largely unknown, to date, most human studies on IDO1 are correlative. Thus, whereas induction of the IDO1 pathway is a feature of numerous human diseases, the extent to which IDO1 represents an activation marker of immune/inflammatory systems or plays an active role of disease pathogenesis and severity is not clear. Notwithstanding, manipulating IDO1 and the Kyn pathway to treat disease has proved successful in a broad spectrum of pre-clinical models of disease. Accordingly, targeting IDO1 clinically is a significant area of interest. Current focus is on clinical trials in cancer patients testing the safety and effectiveness of small-molecule inhibitors of IDO1’s dioxygenase activity or downstream effector mechanisms in order to break tumour-based immune suppression. The outcome of these studies will begin to inform the importance of IDO1-mediated immune suppression in human cancer and the utility of targeting the enzyme therapeutically. Inhibition of IDO1 may also prove beneficial in certain infectious diseases that also employ the enzyme to avoid the host's immune response (e.g. HIV [567]). However, although targeting the IDO1 pathway is of potential therapeutic benefit, caution is required regarding the safety of targeting the IDO1 pathway in combination with other immunotherapeutic and antiviral therapies during HIV infection [570]. Similarly, inhibiting IDO1 in cancer patients suffering certain chronic inflammatory or autoimmune disorders, where IDO1-mediated immune suppression is of benefit, could also prove problematic. In such cases, the development of more targeted IDO1-based therapies is required.

As we await the outcomes of the clinical trials testing IDO1 inhibitors in human cancer patients, it is interesting to note that the effectiveness of certain therapeutics already in clinical use or the subject of clinical trials may relate to their ability to influence IDO1 expression and activity. For example, the tyrosine kinase inhibitor imatinib used to treat certain cancers can also block IDO1-mediated T-cell suppression [393]. Also, IDO1 induction by glucocorticoids or CpG-ODNs may contribute to the drug's protective anti-inflammatory properties in allergic, autoimmune and chronic inflammatory disorders [95,292]. In contrast, induction of IDO1 by ipilimumab confounds its effectiveness in treating melanoma [39], whereas clinically relevant CTLA-4 inhibitors have been linked to inhibition of IDO1 and the dysregulation of bone remodelling [592]. Therefore, given IDO1’s broad implications for disease, it is prudent to consider potential roles of IDO1 in modulating the therapeutic actions of various immune- or inflammatory-based drugs employed in the clinic and their associated side effects.

In addition to inhibition of IDO1-mediated immune suppression, a raft of pre-clinical studies in a range of animal models of inflammatory and autoimmune disorders highlights that augmenting IDO1 expression and activity or direct administration of Kyn pathway metabolites represents a potential therapeutic avenue. The extent to which targeted IDO1 induction or treatment with Kyn metabolites proves efficacious and safe in treating the corresponding human diseases is an area of future interest.

Although IDO1 is the focus of the present review, it is important to remember that, along with IDO2 and TDO, it forms a unique family of mammalian haem dioxygenases capable of converting L-Trp into NFK. These enzymes are capable of exhibiting similar [23,24,35], opposing [37,38] or co-operative [25] biological actions. Furthermore, although it is clear that targeting IDO1 activity with the D- or L-isomers of 1-MT underlies the ability of the drug to modulate numerous physiological and pathophysiological conditions in experimental animals, questions regarding the specificity of 1-MT (and other IDO1-targeted small-molecule inhibitors) for IDO1 compared with IDO2 both in vitro and in vivo remain to be clarified. Although continued study of IDO1, IDO2 and TDO gene knockouts will be useful in defining the physiological and pathological roles of these enzymes, gene-knockout mice can be subject to redundancy and compensatory changes. As such, the generation of conditional, tissue-specific genetically modified mice will help address these potentially confounding issues associated with gene-knockout mice in order to provide further important insights into the roles of these different haem dioxygenases in biology and disease.

When studying IDO1, it is also important to consider the impact that other enzymes related to L-Trp biosynthesis and metabolism have on the actions of IDO1. For example, elevated TTS expression within mammalian cells (e.g. synovial T-cells isolated from rheumatoid arthritis patients [323]) or L-Trp synthesis by certain infectious microbes (e.g. M. tuberculosis [466]) can render these cells refractory to IDO1-mediated L-Trp deprivation. In the latter case, inhibition of microbe L-Trp synthesis can reveal IDO1’s antimicrobial action in vivo [466].

Studies into the regulation of IDO1 expression and enzyme activity have revealed that the biological actions of the enzyme are also linked to the actions of various co-regulatory enzymes, which themselves have immune/inflammation modulatory signalling properties (e.g. NOS, NOX, HO-1, COX, Src kinases). For example, NOS-derived NO represents a potent inhibitory mechanism of IDO1 activity [180,185,186,188]. The extent to which cross-talk between IDO1 and these different enzyme pathways leads to modulation of immune and inflammatory responses under different disease settings represents an important area for further investigation. Moreover, roles for these co-regulatory enzymes also need to be taken into account in the development of IDO1-targeted therapeutic strategies (e.g. inhibiting NOS may elevate IDO1 activity within inflammatory tissues while addition of NO-donor drugs could inhibit IDO1).

To conclude, it is clear that the last 20 years have seen a significant increase in interest in the biochemistry of IDO1 and its role as a central immunoregulatory enzyme in health and disease. It is expected that this interest will only continue to intensify, with studies on the effectiveness of IDO1-based therapeutics in human disease of considerable importance. Continued understanding of all aspects of IDO1, from its catalytic action to its biochemical control and signalling mechanisms, through to its roles in health and disease is crucial in order to effectively modulate IDO1 expression and activity for therapeutic gain.

We apologize to those colleagues whose work was not cited due to space constraints.

FUNDING

This work was supported by an Australian National Health and Medical Research Council (NHMRC) project grant [grant number APP1058508 to S.R.T., P.K. Witting and N.J.C.K.].

Abbreviations

     
  • AhR

    aryl hydrocarbon receptor

  •  
  • δ-ALAS

    δ-aminolaevulinic acid synthase

  •  
  • AML

    acute myeloid leukaemia

  •  
  • L-Arg

    L-arginine

  •  
  • APC

    antigen-presenting cell

  •  
  • BAR

    Bin/amphiphysin/Rvs

  •  
  • Breg

    regulatory B-cell

  •  
  • cDC

    conventional DC

  •  
  • CGD

    chronic granulomatous disease

  •  
  • CNS

    central nervous system

  •  
  • COX

    cyclo-oxygenase

  •  
  • CpG-ODN

    CpG oligodeoxynucleotide

  •  
  • CSF

    cerebrospinal fluid

  •  
  • CTLA-4

    cytotoxic T-lymphocyte-associated protein 4

  •  
  • DAP12

    DNAX activation protein of 12 kDa

  •  
  • DC

    dendritic cell

  •  
  • DMBA

    7,12-dimethylbenz[a]anthracene

  •  
  • DPI

    diphenyleneiodonium

  •  
  • EAE

    experimental autoimmune encephalomyelitis

  •  
  • ECS

    elongin/cullin/SOCS

  •  
  • FOX

    forkhead box

  •  
  • GCN2

    general control non-derepressible 2

  •  
  • GITRL

    glucocorticoid-induced TNF-receptor-related protein

  •  
  • GVHD

    graft-versus-host disease

  •  
  • 3-HAA

    3-hydroxyanthranilic acid

  •  
  • HBV

    hepatitis B virus

  •  
  • HDAC

    histone deacetylase

  •  
  • 3-HK

    3-hydroxykynurenine

  •  
  • 3-HKG

    3-hydroxy-L-kynurenine glucoside

  •  
  • HO-1

    haem oxygenase 1

  •  
  • HSV

    herpes simplex virus type

  •  
  • IBD

    inflammatory bowel disease

  •  
  • IDO

    indoleamine 2,3-dioxygenase

  •  
  • IFN

    interferon

  •  
  • IL

    interleukin

  •  
  • IRF

    IFN regulatory factor

  •  
  • ISS-ODN

    immunostimulatory sequence oligodeoxynucleotide analogues of bacterial DNA

  •  
  • ITIM

    immunoreceptor tyrosine-based inhibitory motif

  •  
  • JAK

    Janus kinase

  •  
  • Kyn

    kynurenine

  •  
  • LAT

    system L amino acid transporter

  •  
  • LDL

    low-density lipoprotein

  •  
  • LDLr

    low-density lipoprotein receptor

  •  
  • LPS

    lipopolysaccharide

  •  
  • MAPK

    mitogen-activated protein kinase

  •  
  • MLR

    mixed lymphocyte reaction

  •  
  • 1-MT

    1-methyltryptophan

  •  
  • mTOR

    mammalian target of rapamycin

  •  
  • NF-κB

    nuclear factor κB

  •  
  • NFK

    N-formylkynurenine

  •  
  • NK

    natural killer

  •  
  • NKT

    natural killer T-

  •  
  • NMDA

    N-methyl-D-aspartate

  •  
  • NOD

    non-obese diabetic

  •  
  • NOS

    nitric oxide synthase

  •  
  • NOX

    NADPH oxidase

  •  
  • OVA

    ovalbumin

  •  
  • pDC

    plasmacytoid DC

  •  
  • PD-1

    programmed cell death 1

  •  
  • PD-L

    programmed cell death ligand

  •  
  • PGE2

    prostaglandin E2

  •  
  • PI

    4-phenylimidazole

  •  
  • PI3K

    phosphoinositide 3-kinase

  •  
  • PKC

    protein kinase C

  •  
  • QA

    quinolinic acid

  •  
  • RNS

    reactive nitrogen species

  •  
  • ROS

    reactive oxygen species

  •  
  • SCID

    severe combined immunodeficiency

  •  
  • SHP

    Src homology 2 domain-containing protein tyrosine phosphatase

  •  
  • SIV

    simian immunodeficiency virus

  •  
  • SNP

    single nucleotide polymorphism

  •  
  • SOCS

    suppressor of cytokine signalling

  •  
  • SOD

    superoxide dismutase

  •  
  • STAT

    signal transducer and activator of transcription

  •  
  • STING

    stimulator of IFN genes

  •  
  • STZ

    streptozotocin

  •  
  • TCR

    T-cell receptor

  •  
  • TDLN

    tumour-draining lymph node

  •  
  • TDO

    tryptophan 2,3-dioxygenase

  •  
  • TGFβ

    transforming growth factor β

  •  
  • TLR

    Toll-like receptor

  •  
  • TNFα

    tumour necrosis factor α

  •  
  • Treg

    regulatory T-cell

  •  
  • L-Trp

    L-tryptophan

  •  
  • TTS

    tryptophanyl-tRNA synthetase

References

References
1
Higuchi
K.
Hayaishi
O.
Enzymic formation of D-kynurenine from D-tryptophan
Arch. Biochem. Biophys.
1967
, vol. 
120
 (pg. 
397
-
403
)
[PubMed]
2
Yamamoto
S.
Hayaishi
O.
Tryptophan pyrrolase of rabbit intestine. D- and L-tryptophan-cleaving enzyme or enzymes
J. Biol. Chem.
1967
, vol. 
242
 (pg. 
5260
-
5266
)
[PubMed]
3
Oxenkrug
G.F.
Metabolic syndrome, age-associated neuroendocrine disorders, and dysregulation of tryptophan-kynurenine metabolism
Ann. N.Y. Acad. Sci.
2010
, vol. 
1199
 (pg. 
1
-
14
)
[PubMed]
4
Armstrong
M.D.
Stave
U.
A study of plasma free amino acid levels. II. Normal values for children and adults
Metabolism
1973
, vol. 
22
 (pg. 
561
-
569
)
[PubMed]
5
De Marte
M.L.
Enesco
H.E.
Influence of diet on plasma tryptophan and brain serotonin levels in mice
Experientia
1985
, vol. 
41
 (pg. 
48
-
50
)
[PubMed]
6
Rivera
S.
Lopez-Soriano
F.J.
Azcon-Bieto
J.
Argiles
J.M.
Blood amino acid compartmentation in mice bearing Lewis lung carcinoma
Cancer Res.
1987
, vol. 
47
 (pg. 
5644
-
5646
)
[PubMed]
7
Thomas
S.R.
Stocker
R.
Redox reactions related to indoleamine 2,3-dioxygenase and tryptophan metabolism along the kynurenine pathway
Redox Rep.
1999
, vol. 
4
 (pg. 
199
-
220
)
[PubMed]
8
Khan
J.A.
Forouhar
F.
Tao
X.
Tong
L.
Nicotinamide adenine dinucleotide metabolism as an attractive target for drug discovery
Expert Opin. Ther. Targets
2007
, vol. 
11
 (pg. 
695
-
705
)
[PubMed]
9
Grant
R.S.
Naif
H.
Espinosa
M.
Kapoor
V.
IDO induction in IFN-γ activated astroglia: a role in improving cell viability during oxidative stress
Redox Rep.
2000
, vol. 
5
 (pg. 
101
-
104
)
[PubMed]
10
Kudo
Y.
Boyd
C.A.
The role of L-tryptophan transport in L-tryptophan degradation by indoleamine 2,3-dioxygenase in human placental explants
J. Physiol.
2001
, vol. 
531
 (pg. 
417
-
423
)
[PubMed]
11
Seymour
R.L.
Ganapathy
V.
Mellor
A.L.
Munn
D.H.
A high-affinity, tryptophan-selective amino acid transport system in human macrophages
J. Leukoc. Biol.
2006
, vol. 
80
 (pg. 
1320
-
1327
)
[PubMed]
12
Verrey
F.
System L: heteromeric exchangers of large, neutral amino acids involved in directional transport
Pflugers Arch.
2003
, vol. 
445
 (pg. 
529
-
533
)
[PubMed]
13
Bhutia
Y.D.
Babu
E.
Ganapathy
V.
Interferon-γ induces a tryptophan-selective amino acid transporter in human colonic epithelial cells and mouse dendritic cells
Biochim. Biophys. Acta
2014
, vol. 
1848
 (pg. 
453
-
462
)
[PubMed]
14
Silva
N.M.
Rodrigues
C.V.
Santoro
M.M.
Reis
L.F.
Alvarez-Leite
J.I.
Gazzinelli
R.T.
Expression of indoleamine 2,3-dioxygenase, tryptophan degradation, and kynurenine formation during in vivoinfection with Toxoplasma gondii: induction by endogenous γ interferon and requirement of interferon regulatory factor 1
Infect. Immun.
2002
, vol. 
70
 (pg. 
859
-
868
)
[PubMed]
15
Murakami
Y.
Hoshi
M.
Hara
A.
Takemura
M.
Arioka
Y.
Yamamoto
Y.
Matsunami
H.
Funato
T.
Seishima
M.
Saito
K.
Inhibition of increased indoleamine 2,3-dioxygenase activity attenuates Toxoplasma gondii replication in the lung during acute infection
Cytokine
2012
, vol. 
59
 (pg. 
245
-
251
)
[PubMed]
16
Kaper
T.
Looger
L.L.
Takanaga
H.
Platten
M.
Steinman
L.
Frommer
W.B.
Nanosensor detection of an immunoregulatory tryptophan influx/kynurenine efflux cycle
PLoS Biol.
2007
, vol. 
5
 pg. 
e257
 
[PubMed]
17
Uwai
Y.
Honjo
H.
Iwamoto
K.
Interaction and transport of kynurenic acid via human organic anion transporters hOAT1 and hOAT3
Pharmacol. Res.
2012
, vol. 
65
 (pg. 
254
-
260
)
[PubMed]
18
DiNatale
B.C.
Murray
I.A.
Schroeder
J.C.
Flaveny
C.A.
Lahoti
T.S.
Laurenzana
E.M.
Omiecinski
C.J.
Perdew
G.H.
Kynurenic acid is a potent endogenous aryl hydrocarbon receptor ligand that synergistically induces interleukin-6 in the presence of inflammatory signaling
Toxicol. Sci.
2010
, vol. 
115
 (pg. 
89
-
97
)
[PubMed]
19
Kaszaki
J.
Erces
D.
Varga
G.
Szabo
A.
Vecsei
L.
Boros
M.
Kynurenines and intestinal neurotransmission: the role of N-methyl-D-aspartate receptors
J. Neural Transm.
2012
, vol. 
119
 (pg. 
211
-
223
)
[PubMed]
20
Hilmas
C.
Pereira
E.F.
Alkondon
M.
Rassoulpour
A.
Schwarcz
R.
Albuquerque
E.X.
The brain metabolite kynurenic acid inhibits α7 nicotinic receptor activity and increases non-α7 nicotinic receptor expression: physiopathological implications
J. Neurosci.
2001
, vol. 
21
 (pg. 
7463
-
7473
)
[PubMed]
21
Mellor
A.L.
Munn
D.H.
IDO expression by dendritic cells: tolerance and tryptophan catabolism
Nat. Rev. Immunol.
2004
, vol. 
4
 (pg. 
762
-
774
)
[PubMed]
22
Wu
W.
Nicolazzo
J.A.
Wen
L.
Chung
R.
Stankovic
R.
Bao
S.S.
Lim
C.K.
Brew
B.J.
Cullen
K.M.
Guillemin
G.J.
Expression of tryptophan 2,3-dioxygenase and production of kynurenine pathway metabolites in triple transgenic mice and human Alzheimer's disease brain
PLoS One
2013
, vol. 
8
 pg. 
e59749
 
[PubMed]
23
Opitz
C.A.
Litzenburger
U.M.
Sahm
F.
Ott
M.
Tritschler
I.
Trump
S.
Schumacher
T.
Jestaedt
L.
Schrenk
D.
Weller
M.
, et al. 
An endogenous tumour-promoting ligand of the human aryl hydrocarbon receptor
Nature
2011
, vol. 
478
 (pg. 
197
-
203
)
[PubMed]
24
Pilotte
L.
Larrieu
P.
Stroobant
V.
Colau
D.
Dolusic
E.
Frederick
R.
De Plaen
E.
Uyttenhove
C.
Wouters
J.
Masereel
B.
Van den Eynde
B.J.
Reversal of tumoral immune resistance by inhibition of tryptophan 2,3-dioxygenase
Proc. Natl. Acad. Sci. U.S.A.
2012
, vol. 
109
 (pg. 
2497
-
2502
)
[PubMed]
25
Bessede
A.
Gargaro
M.
Pallotta
M.T.
Matino
D.
Servillo
G.
Brunacci
C.
Bicciato
S.
Mazza
E.M.
Macchiarulo
A.
Vacca
C.
, et al. 
Aryl hydrocarbon receptor control of a disease tolerance defence pathway
Nature
2014
, vol. 
511
 (pg. 
184
-
190
)
[PubMed]
26
Forouhar
F.
Anderson
J.L.
Mowat
C.G.
Vorobiev
S.M.
Hussain
A.
Abashidze
M.
Bruckmann
C.
Thackray
S.J.
Seetharaman
J.
Tucker
T.
, et al. 
Molecular insights into substrate recognition and catalysis by tryptophan 2,3-dioxygenase
Proc. Natl. Acad. Sci. U.S.A.
2007
, vol. 
104
 (pg. 
473
-
478
)
[PubMed]
27
Rafice
S.A.
Chauhan
N.
Efimov
I.
Basran
J.
Raven
E.L.
Oxidation of L-tryptophan in biology: a comparison between tryptophan 2,3-dioxygenase and indoleamine 2,3-dioxygenase
Biochem. Soc. Trans.
2009
, vol. 
37
 (pg. 
408
-
412
)
[PubMed]
28
Basran
J.
Rafice
S.A.
Chauhan
N.
Efimov
I.
Cheesman
M.R.
Ghamsari
L.
Raven
E.L.
A kinetic, spectroscopic, and redox study of human tryptophan 2,3-dioxygenase
Biochemistry
2008
, vol. 
47
 (pg. 
4752
-
4760
)
[PubMed]
29
Fu
R.
Gupta
R.
Geng
J.
Dornevil
K.
Wang
S.
Zhang
Y.
Hendrich
M.P.
Liu
A.
Enzyme reactivation by hydrogen peroxide in heme-based tryptophan dioxygenase
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
26541
-
26554
)
[PubMed]
30
Freewan
M.
Rees
M.D.
Plaza
T.S.
Glaros
E.
Lim
Y.J.
Wang
X.S.
Yeung
A.W.
Witting
P.K.
Terentis
A.C.
Thomas
S.R.
Human indoleamine 2,3-dioxygenase is a catalyst of physiological heme peroxidase reactions: implications for the inhibition of dioxygenase activity by hydrogen peroxide
J. Biol. Chem.
2013
, vol. 
288
 (pg. 
1548
-
1567
)
[PubMed]
31
Ball
H.J.
Sanchez-Perez
A.
Weiser
S.
Austin
C.J.
Astelbauer
F.
Miu
J.
McQuillan
J.A.
Stocker
R.
Jermiin
L.S.
Hunt
N.H.
Characterization of an indoleamine 2,3-dioxygenase-like protein found in humans and mice
Gene
2007
, vol. 
396
 (pg. 
203
-
213
)
[PubMed]
32
Metz
R.
Duhadaway
J.B.
Kamasani
U.
Laury-Kleintop
L.
Muller
A.J.
Prendergast
G.C.
Novel tryptophan catabolic enzyme IDO2 is the preferred biochemical target of the antitumor indoleamine 2,3-dioxygenase inhibitory compound D-1-methyl-tryptophan
Cancer Res.
2007
, vol. 
67
 (pg. 
7082
-
7087
)
[PubMed]
33
Ball
H.J.
Yuasa
H.J.
Austin
C.J.
Weiser
S.
Hunt
N.H.
Indoleamine 2,3-dioxygenase-2; a new enzyme in the kynurenine pathway
Int. J. Biochem. Cell Biol.
2009
, vol. 
41
 (pg. 
467
-
471
)
[PubMed]
34
Austin
C.J.
Mailu
B.M.
Maghzal
G.J.
Sanchez-Perez
A.
Rahlfs
S.
Zocher
K.
Yuasa
H.J.
Arthur
J.W.
Becker
K.
Stocker
R.
, et al. 
Biochemical characteristics and inhibitor selectivity of mouse indoleamine 2,3-dioxygenase-2
Amino Acids
2010
, vol. 
39
 (pg. 
565
-
578
)
[PubMed]
35
Trabanelli
S.
Ocadlikova
D.
Ciciarello
M.
Salvestrini
V.
Lecciso
M.
Jandus
C.
Metz
R.
Evangelisti
C.
Laury-Kleintop
L.
Romero
P.
, et al. 
The SOCS3-independent expression of IDO2 supports the homeostatic generation of T regulatory cells by human dendritic cells
J. Immunol.
2014
, vol. 
192
 (pg. 
1231
-
1240
)
[PubMed]
36
Fukunaga
M.
Yamamoto
Y.
Kawasoe
M.
Arioka
Y.
Murakami
Y.
Hoshi
M.
Saito
K.
Studies on tissue and cellular distribution of indoleamine 2,3-dioxygenase 2: the absence of IDO1 upregulates IDO2 expression in the epididymis
J. Histochem. Cytochem.
2012
, vol. 
60
 (pg. 
854
-
860
)
[PubMed]
37
Merlo
L.M.
Pigott
E.
DuHadaway
J.B.
Grabler
S.
Metz
R.
Prendergast
G.C.
Mandik-Nayak
L.
IDO2 is a critical mediator of autoantibody production and inflammatory pathogenesis in a mouse model of autoimmune arthritis
J. Immunol.
2014
, vol. 
192
 (pg. 
2082
-
2090
)
[PubMed]
38
Metz
R.
Smith
C.
DuHadaway
J.B.
Chandler
P.
Baban
B.
Merlo
L.M.
Pigott
E.
Keough
M.P.
Rust
S.
Mellor
A.L.
, et al. 
IDO2 is critical for IDO1-mediated T-cell regulation and exerts a non-redundant function in inflammation
Int. Immunol.
2014
, vol. 
26
 (pg. 
357
-
367
)
[PubMed]
39
Holmgaard
R.B.
Zamarin
D.
Munn
D.H.
Wolchok
J.D.
Allison
J.P.
Indoleamine 2,3-dioxygenase is a critical resistance mechanism in antitumor T cell immunotherapy targeting CTLA-4
J. Exp. Med.
2013
, vol. 
210
 (pg. 
1389
-
1402
)
[PubMed]
40
Dai
X.
Zhu
B.T.
Indoleamine 2,3-dioxygenase tissue distribution and cellular localization in mice: implications for its biological functions
J. Histochem. Cytochem.
2010
, vol. 
58
 (pg. 
17
-
28
)
[PubMed]
41
Watts
S.W.
Shaw
S.
Burnett
R.
Dorrance
A.M.
Indoleamine 2,3-diooxygenase in periaortic fat: mechanisms of inhibition of contraction
Am. J. Physiol. Heart Circ. Physiol.
2011
, vol. 
301
 (pg. 
H1236
-
H1247
)
[PubMed]
42
Valladares
R.
Bojilova
L.
Potts
A.H.
Cameron
E.
Gardner
C.
Lorca
G.
Gonzalez
C.F.
Lactobacillus johnsonii inhibits indoleamine 2,3-dioxygenase and alters tryptophan metabolite levels in BioBreeding rats
FASEB J.
2013
, vol. 
27
 (pg. 
1711
-
1720
)
[PubMed]
43
Onodera
T.
Jang
M.H.
Guo
Z.
Yamasaki
M.
Hirata
T.
Bai
Z.
Tsuji
N.M.
Nagakubo
D.
Yoshie
O.
Sakaguchi
S.
, et al. 
Constitutive expression of IDO by dendritic cells of mesenteric lymph nodes: functional involvement of the CTLA-4/B7 and CCL22/CCR4 interactions
J. Immunol.
2009
, vol. 
183
 (pg. 
5608
-
5614
)
[PubMed]
44
Harrington
L.
Srikanth
C.V.
Antony
R.
Rhee
S.J.
Mellor
A.L.
Shi
H.N.
Cherayil
B.J.
Deficiency of indoleamine 2,3-dioxygenase enhances commensal-induced antibody responses and protects against Citrobacter rodentium-induced colitis
Infect. Immun.
2008
, vol. 
76
 (pg. 
3045
-
3053
)
[PubMed]
45
Takikawa
O.
Tagawa
Y.
Iwakura
Y.
Yoshida
R.
Truscott
R.J.
Interferon-γ-dependent/independent expression of indoleamine 2,3-dioxygenase: studies with interferon-γ-knockout mice
Adv. Exp. Med. Biol.
1999
, vol. 
467
 (pg. 
553
-
557
)
[PubMed]
46
Ryu
Y.H.
Kim
J.C.
Expression of indoleamine 2,3-dioxygenase in human corneal cells as a local immunosuppressive factor
Invest. Ophthalmol. Vis. Sci.
2007
, vol. 
48
 (pg. 
4148
-
4152
)
[PubMed]
47
Sarkar
S.A.
Wong
R.
Hackl
S.I.
Moua
O.
Gill
R.G.
Wiseman
A.
Davidson
H.W.
Hutton
J.C.
Induction of indoleamine 2,3-dioxygenase by interferon-γ in human islets
Diabetes
2007
, vol. 
56
 (pg. 
72
-
79
)
[PubMed]
48
Sedlmayr
P.
Blaschitz
A.
Wintersteiger
R.
Semlitsch
M.
Hammer
A.
MacKenzie
C.R.
Walcher
W.
Reich
O.
Takikawa
O.
Dohr
G.
Localization of indoleamine 2,3-dioxygenase in human female reproductive organs and the placenta
Mol. Hum. Reprod.
2002
, vol. 
8
 (pg. 
385
-
391
)
[PubMed]
49
Munn
D.H.
Sharma
M.D.
Lee
J.R.
Jhaver
K.G.
Johnson
T.S.
Keskin
D.B.
Marshall
B.
Chandler
P.
Antonia
S.J.
Burgess
R.
, et al. 
Potential regulatory function of human dendritic cells expressing indoleamine 2,3-dioxygenase
Science
2002
, vol. 
297
 (pg. 
1867
-
1870
)
[PubMed]
50
Odemuyiwa
S.O.
Ghahary
A.
Li
Y.
Puttagunta
L.
Lee
J.E.
Musat-Marcu
S.
Ghahary
A.
Moqbel
R.
Cutting edge: human eosinophils regulate T cell subset selection through indoleamine 2,3-dioxygenase
J. Immunol.
2004
, vol. 
173
 (pg. 
5909
-
5913
)
[PubMed]
51
Munn
D.H.
Shafizadeh
E.
Attwood
J.T.
Bondarev
I.
Pashine
A.
Mellor
A.L.
Inhibition of T cell proliferation by macrophage tryptophan catabolism
J. Exp. Med.
1999
, vol. 
189
 (pg. 
1363
-
1372
)
[PubMed]
52
De Ravin
S.S.
Zarember
K.A.
Long-Priel
D.
Chan
K.C.
Fox
S.D.
Gallin
J.I.
Kuhns
D.B.
Malech
H.L.
Tryptophan/kynurenine metabolism in human leukocytes is independent of superoxide and is fully maintained in chronic granulomatous disease
Blood
2010
, vol. 
116
 (pg. 
1755
-
1760
)
[PubMed]
53
Lewkowicz
N.
Klink
M.
Mycko
M.P.
Lewkowicz
P.
Neutrophil–CD4+CD25+ T regulatory cell interactions: a possible new mechanism of infectious tolerance
Immunobiology
2013
, vol. 
218
 (pg. 
455
-
464
)
[PubMed]
54
Xiao
Y.
Christou
H.
Liu
L.
Visner
G.
Mitsialis
S.A.
Kourembanas
S.
Liu
H.
Endothelial indoleamine 2,3-dioxygenase protects against development of pulmonary hypertension
Am. J. Respir. Crit. Care Med.
2013
, vol. 
188
 (pg. 
482
-
491
)
[PubMed]
55
Hayashi
T.
Beck
L.
Rossetto
C.
Gong
X.
Takikawa
O.
Takabayashi
K.
Broide
D.H.
Carson
D.A.
Raz
E.
Inhibition of experimental asthma by indoleamine 2,3-dioxygenase
J. Clin. Invest.
2004
, vol. 
114
 (pg. 
270
-
279
)
[PubMed]
56
Curran
T.A.
Jalili
R.B.
Farrokhi
A.
Ghahary
A.
IDO expressing fibroblasts promote the expansion of antigen specific regulatory T cells
Immunobiology
2014
, vol. 
219
 (pg. 
17
-
24
)
[PubMed]
57
Pantoja
L.G.
Miller
R.D.
Ramirez
J.A.
Molestina
R.E.
Summersgill
J.T.
Inhibition of Chlamydia pneumoniae replication in human aortic smooth muscle cells by γ interferon-induced indoleamine 2,3-dioxygenase activity
Infect. Immun.
2000
, vol. 
68
 (pg. 
6478
-
6481
)
[PubMed]
58
Romani
R.
Pirisinu
I.
Calvitti
M.
Pallotta
M.T.
Gargaro
M.
Bistoni
G.
Vacca
C.
Di Michele
A.
Orabona
C.
Rosati
J.
, et al. 
Stem cells from human amniotic fluid exert immunoregulatory function via secreted indoleamine 2,3-dioxygenase1
J. Cell. Mol. Med.
2015
 
doi:10.1111/jcmm.12534
59
Burkin
D.J.
Kimbro
K.S.
Barr
B.L.
Jones
C.
Taylor
M.W.
Gupta
S.L.
Localization of the human indoleamine 2,3-dioxygenase (IDO) gene to the pericentromeric region of human chromosome 8
Genomics
1993
, vol. 
17
 (pg. 
262
-
263
)
[PubMed]
60
Grohmann
U.
Bronte
V.
Control of immune response by amino acid metabolism
Immunol. Rev.
2010
, vol. 
236
 (pg. 
243
-
264
)
[PubMed]
61
Tone
S.
Kadoya
A.
Maeda
H.
Minatogawa
Y.
Kido
R.
Assignment of the human indoleamine 2,3-dioxygenase gene to chromosome 8 using the polymerase chain reaction
Hum. Genet.
1994
, vol. 
93
 (pg. 
201
-
203
)
[PubMed]
62
Arefayene
M.
Philips
S.
Cao
D.
Mamidipalli
S.
Desta
Z.
Flockhart
D.A.
Wilkes
D.S.
Skaar
T.C.
Identification of genetic variants in the human indoleamine 2,3-dioxygenase (IDO1) gene, which have altered enzyme activity
Pharmacogenet. Genomics
2009
, vol. 
19
 (pg. 
464
-
476
)
[PubMed]
63
Tardito
S.
Negrini
S.
Conteduca
G.
Ferrera
F.
Parodi
A.
Battaglia
F.
Kalli
F.
Fenoglio
D.
Cutolo
M.
Filaci
G.
Indoleamine 2,3 dioxygenase gene polymorphisms correlate with CD8+ Treg impairment in systemic sclerosis
Hum. Immunol.
2013
, vol. 
74
 (pg. 
166
-
169
)
[PubMed]
64
De Luca
A.
Carvalho
A.
Cunha
C.
Iannitti
R.G.
Pitzurra
L.
Giovannini
G.
Mencacci
A.
Bartolommei
L.
Moretti
S.
Massi-Benedetti
C.
, et al. 
IL-22 and IDO1 affect immunity and tolerance to murine and human vaginal candidiasis
PLoS Pathog.
2013
, vol. 
9
 pg. 
e1003486
 
[PubMed]
65
Reddy
P.
Sun
Y.
Toubai
T.
Duran-Struuck
R.
Clouthier
S.G.
Weisiger
E.
Maeda
Y.
Tawara
I.
Krijanovski
O.
Gatza
E.
, et al. 
Histone deacetylase inhibition modulates indoleamine 2,3-dioxygenase-dependent DC functions and regulates experimental graft-versus-host disease in mice
J. Clin. Invest.
2008
, vol. 
118
 (pg. 
2562
-
2573
)
[PubMed]
66
Robinson
C.M.
Shirey
K.A.
Carlin
J.M.
Synergistic transcriptional activation of indoleamine dioxygenase by IFN-γ and tumor necrosis factor-α;
J. Interferon Cytokine Res.
2003
, vol. 
23
 (pg. 
413
-
421
)
[PubMed]
67
Gough
D.J.
Levy
D.E.
Johnstone
R.W.
Clarke
C.J.
IFNγ signaling: does it mean JAK–STAT?
Cytokine Growth Factor Rev.
2008
, vol. 
19
 (pg. 
383
-
394
)
[PubMed]
68
Jeong
Y.I.
Kim
S.W.
Jung
I.D.
Lee
J.S.
Chang
J.H.
Lee
C.M.
Chun
S.H.
Yoon
M.S.
Kim
G.T.
Ryu
S.W.
, et al. 
Curcumin suppresses the induction of indoleamine 2,3-dioxygenase by blocking the Janus-activated kinase–protein kinase Cδ–STAT1 signaling pathway in interferon-γ-stimulated murine dendritic cells
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
3700
-
3708
)
[PubMed]
69
Chon
S.Y.
Hassanain
H.H.
Pine
R.
Gupta
S.L.
Involvement of two regulatory elements in interferon-γ-regulated expression of human indoleamine 2,3-dioxygenase gene
J. Interferon Cytokine Res.
1995
, vol. 
15
 (pg. 
517
-
526
)
[PubMed]
70
Konan
K.V.
Taylor
M.W.
Importance of the two interferon-stimulated response element (ISRE) sequences in the regulation of the human indoleamine 2,3-dioxygenase gene
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
19140
-
19145
)
[PubMed]
71
Pine
R.
Convergence of TNFα and IFNγ signalling pathways through synergistic induction of IRF-1/ISGF-2 is mediated by a composite GAS/κB promoter element
Nucleic Acids Res.
1997
, vol. 
25
 (pg. 
4346
-
4354
)
[PubMed]
72
Mounayar
M.
Kefaloyianni
E.
Smith
B.
Solhjou
Z.
Maarouf
O.H.
Azzi
J.
Chabtini
L.
Fiorina
P.
Kraus
M.
Briddell
R.
, et al. 
PI3Kα and STAT1 interplay regulates human mesenchymal stem cell immune polarization
Stem Cells
2015
, vol. 
33
 (pg. 
1892
-
1901
)
[PubMed]
73
Orabona
C.
Puccetti
P.
Vacca
C.
Bicciato
S.
Luchini
A.
Fallarino
F.
Bianchi
R.
Velardi
E.
Perruccio
K.
Velardi
A.
, et al. 
Toward the identification of a tolerogenic signature in IDO-competent dendritic cells
Blood
2006
, vol. 
107
 (pg. 
2846
-
2854
)
[PubMed]
74
Braun
D.
Longman
R.S.
Albert
M.L.
A two-step induction of indoleamine 2,3 dioxygenase (IDO) activity during dendritic-cell maturation
Blood
2005
, vol. 
106
 (pg. 
2375
-
2381
)
[PubMed]
75
Babcock
T.A.
Carlin
J.M.
Transcriptional activation of indoleamine dioxygenase by interleukin 1 and tumor necrosis factor α in interferon-treated epithelial cells
Cytokine
2000
, vol. 
12
 (pg. 
588
-
594
)
[PubMed]
76
Krakauer
T.
Oppenheim
J.J.
IL-1 and tumor necrosis factor-α each up-regulate both the expression of IFN-γ receptors and enhance IFN-γ-induced HLA-DR expression on human monocytes and a human monocytic cell line (THP-1)
J. Immunol.
1993
, vol. 
150
 (pg. 
1205
-
1211
)
[PubMed]
77
Robinson
C.M.
Hale
P.T.
Carlin
J.M.
NF-κB activation contributes to indoleamine dioxygenase transcriptional synergy induced by IFN-γ and tumor necrosis factor-α
Cytokine
2006
, vol. 
35
 (pg. 
53
-
61
)
[PubMed]
78
Adams
O.
Besken
K.
Oberdorfer
C.
MacKenzie
C.R.
Russing
D.
Daubener
W.
Inhibition of human herpes simplex virus type 2 by interferon γ and tumor necrosis factor α is mediated by indoleamine 2,3-dioxygenase
Microbes Infect.
2004
, vol. 
6
 (pg. 
806
-
812
)
[PubMed]
79
Belladonna
M.L.
Orabona
C.
Grohmann
U.
Puccetti
P.
TGF-β and kynurenines as the key to infectious tolerance
Trends Mol. Med.
2009
, vol. 
15
 (pg. 
41
-
49
)
[PubMed]
80
Fallarino
F.
Puccetti
P.
Toll-like receptor 9-mediated induction of the immunosuppressive pathway of tryptophan catabolism
Eur. J. Immunol.
2006
, vol. 
36
 (pg. 
8
-
11
)
[PubMed]
81
Fox
J.M.
Crabtree
J.M.
Sage
L.K.
Tompkins
S.M.
Tripp
R.A.
Interferon Lambda upregulates IDO1 expression in respiratory epithelial cells after influenza virus infection
J. Interferon Cytokine Res.
2015
 
doi:10.1089/jir.2014.0052
82
Furset
G.
Floisand
Y.
Sioud
M.
Impaired expression of indoleamine 2, 3-dioxygenase in monocyte-derived dendritic cells in response to Toll-like receptor-7/8 ligands
Immunology
2008
, vol. 
123
 (pg. 
263
-
271
)
[PubMed]
83
Hissong
B.D.
Byrne
G.I.
Padilla
M.L.
Carlin
J.M.
Upregulation of interferon-induced indoleamine 2,3-dioxygenase in human macrophage cultures by lipopolysaccharide, muramyl tripeptide, and interleukin-1
Cell. Immunol.
1995
, vol. 
160
 (pg. 
264
-
269
)
[PubMed]
84
Hissong
B.D.
Carlin
J.M.
Potentiation of interferon-induced indoleamine 2,3-dioxygenase mRNA in human mononuclear phagocytes by lipopolysaccharide and interleukin-1
J. Interferon Cytokine Res.
1997
, vol. 
17
 (pg. 
387
-
393
)
[PubMed]
85
Jung
I.D.
Jeong
Y.I.
Lee
C.M.
Noh
K.T.
Jeong
S.K.
Chun
S.H.
Choi
O.H.
Park
W.S.
Han
J.
Shin
Y.K.
, et al. 
COX-2 and PGE2 signaling is essential for the regulation of IDO expression by curcumin in murine bone marrow-derived dendritic cells
Int. Immunopharmacol.
2010
, vol. 
10
 (pg. 
760
-
768
)
[PubMed]
86
Jung
I.D.
Lee
C.M.
Jeong
Y.I.
Lee
J.S.
Park
W.S.
Han
J.
Park
Y.M.
Differential regulation of indoleamine 2,3-dioxygenase by lipopolysaccharide and interferon γ in murine bone marrow derived dendritic cells
FEBS Lett.
2007
, vol. 
581
 (pg. 
1449
-
1456
)
[PubMed]
87
Roshick
C.
Wood
H.
Caldwell
H.D.
McClarty
G.
Comparison of γ interferon-mediated antichlamydial defense mechanisms in human and mouse cells
Infect. Immun.
2006
, vol. 
74
 (pg. 
225
-
238
)
[PubMed]
88
Robinson
C.M.
Hale
P.T.
Carlin
J.M.
The role of IFN-γ and TNF-α-responsive regulatory elements in the synergistic induction of indoleamine dioxygenase
J. Interferon Cytokine Res.
2005
, vol. 
25
 (pg. 
20
-
30
)
[PubMed]
89
Suh
H.S.
Zhao
M.L.
Rivieccio
M.
Choi
S.
Connolly
E.
Zhao
Y.
Takikawa
O.
Brosnan
C.F.
Lee
S.C.
Astrocyte indoleamine 2,3-dioxygenase is induced by the TLR3 ligand poly(I:C): mechanism of induction and role in antiviral response
J. Virol.
2007
, vol. 
81
 (pg. 
9838
-
9850
)
[PubMed]
90
Zhu
W.H.
Lu
C.Z.
Huang
Y.M.
Link
H.
Xiao
B.G.
A putative mechanism on remission of multiple sclerosis during pregnancy: estrogen-induced indoleamine 2,3-dioxygenase by dendritic cells
Mult. Scler.
2007
, vol. 
13
 (pg. 
33
-
40
)
[PubMed]
91
Romani
L.
Bistoni