The degradation of tryptophan (TRP) along the KP plays a crucial role in the regulation of the immune response, notably as a counter-regulatory mechanism in the context of inflammation (1–3). An overview of the KP is presented in Figure 1. Three rate-limiting enzymes of KP, tryptophan 2,3-dioxygenase (TDO) and indolamine 2,3-dioxygenase (IDO) 1 and 2, have been described in the literature thus far. TDO is positively regulated by TRP in order to maintain the homeostasis of TRP (4, 5). Furthermore, the expression and activity of TDO is regulated by hormones such as cortisol, insulin, glucagon, or epinephrine (6–8). IDO1 and 2 are upregulated by inflammatory stimuli such as interferon-γ (IFN-γ) (9–14). The significance of KP activation depends on the production of biologically active metabolites such as kynurenine (KYN), kynurenic acid (KYNA), quinolinic acid (QUIN), or anthranilic acid mediating various immuno- and neuromodulative functions. Within the central nervous system, it has been well documented that metabolites such as KYNA and QUIN modulate neurological functions. Thus, KYNA acts as an antagonist affecting all ionotropic glutamate receptors including NMDA, AMPA, and kainate receptors as well as the α7 nicotinic acetylcholine receptor (α7nAChR) assuming it as a neuroprotective metabolite (15–18). However, the inhibition of α7nAChR by KYNA is extensively debated because some later studies addressing this mechanism (3, 19–21) could not recapitulate the original results from Hilmas et al. (18). Dysregulation of KP, resulting in alterations of the balance between KYNA and QUIN, has been described in many neurological disorders (22). However, alterations of KYNA are also described in several inflammatory-related states, such as sepsis or inflammatory bowel disease (IBD), and are discussed as a potential marker in cancer patients (23–25). It is generally accepted that KYNA mediates immunosuppressive effects (22), notably by targeting the G-protein-coupled receptor 35 (GPR35)- or aryl hydrocarbon receptor (AhR)-associated signaling pathways (2, 26, 27).

Generation of KYNA was described in endothelial cells (28), epithelial cells (29, 30), fibroblasts (31), pancreatic islet cells (32), human peripheral blood mononuclear cells (33), skeletal muscle cells (34), and red blood cells (35). Under physiological conditions, KYNA is produced by kynurenine aminotransferases (KATs), which catalyze the irreversible transamination reaction between l-KYN and 2-oxoacid, as a co-substrate, to form KYNA (36, 37), or in the presence of reactive oxygen species (ROS) as illustrated in Figure 1. Currently four proteins named KAT I–IV are described in mammals (38–43). Mediating overlapping biological functions, the mammalian enzymes KAT I and KAT III share high homologies in sequence and genomic structure (42). KAT activity was described in various tissues, such as liver, kidney, small intestine, dermal fibroblasts, and brain (31, 38, 44, 45). In human cardiac muscle, there is evidence that the KAT system differs from brain KAT regulation regarding optimum pH, co-substrate specificity, and sensitivity to inhibition by amino acids such as l-TRP (36). Regarding the inflammatory regulation of KATs, there is no consensus, assuming the existence of cell type-dependent regulatory differences. Studies in human dermal fibroblasts reveal that tumor necrosis factor α (TNF) alone does not affect the number of transcripts, whereas IFN-γ alone (or in combination with TNF) decreases the transcript abundance of KAT I, III, and IV after 48 h. Due to the fact that in this study the metabolite concentration of KYNA was increased in supernatants the decrease of KATs 48 h after cytokine stimulation may reflect a negative feedback mechanism (31). In fetal astrocytes, IFN-γ increases the transcript levels of KAT I and II after 24 h (46), whereas no effect of IFN-γ was observed in neuronal cells (47). In mice, an intraperitoneal (i.p.) lipopolysaccharide (LPS) challenge increased KAT-I mRNA expression in the hippocampus 6 h postchallenge, followed by a decrease after 24 h, whereas KAT-II mRNA expression was decreased 24 h after LPS treatment (48).

In addition to KAT-catalyzed KYNA synthesis, alternative routes for KYNA synthesis in the presence of ROS have been described (49) (Figure 1). Thus, it is assumed that indole-3-pyruvic acid, a reaction product of tryptophan-2-oxoglutarate-induced transamination of TRP, undergoes pyrrole ring cleavage followed by a spontaneous cyclization generating KYNA (50). Furthermore, l-KYN can be converted to KYNA in the presence of hydrogen peroxide (51) or KYNA formation can result from reactions of KYN or indole-3-pyruvic acid under conditions generating free radicals (52, 53).

Elevations of KYNA blood concentrations were experimentally induced under different inflammatory conditions. In pigs, the i.p. application of LPS increased plasma levels of KYNA. However, the ex vivo LPS stimulation of whole blood culture failed to elevate KYNA in supernatants (54), assuming sources of KYNA production other than blood cells or the requirement of additional inflammatory mediators, which are not produced in blood cells. Increased KYNA plasma concentrations were also detected in mice that were repeatedly stressed. In this context, the elevated KYNA levels were proposed to be induced by a systemic low-grade inflammation due to an altered intestinal barrier function (55). Interestingly, in this study the application of the IDO inhibitor l-1-methyltryptophan (1-MT), a TRP analog, specifically increased plasma concentrations of KYNA by a yet unknown pathway (55).

Kynurenic acid is described as one of the end products of KP in animals, assuming no uptake or further metabolism of KYNA. In a study including different rodent species, 90% of radioactively labeled KYNA was excreted in urine within 24 h of i.p. application (56). Thus, 80–100% of labeled KYNA was excreted unchanged and only small amounts of quinaldic acid and quinaldylglycine were detected (0.3 and 5%, respectively). This is supported by studies in rats, finding that radioactively labeled KYNA was eliminated rapidly after intracerebroventricular microinjection and substantial amounts of radioactivity were recovered in urine 30 min after injection (57). Studies in rabbits described differences in KYNA metabolism depending on the type of administration (58). After oral administration of KYNA, the majority of the dose was detected in the form of quinaldic or 8-hydroxyquinaldic acid, indicating a dehydroxylation of the molecule. In contrast, after subcutaneous administration, 99% of KYNA was recovered unchanged, indicating that the dehydroxylation occurs in the gastrointestinal tract (58) most probably by the gut microbiota. This is supported by the finding that approximately 30% of ingested KYNA was excreted in urine as quinaldic acid in humans (59). The assumption that KYNA is metabolized by microorganisms is supported by the finding that extracts of Pseudomonas spp. and Aerococcus spp. were able to enzymatically partially degrade KYNA (60, 61).

In recent years, numerous in vivo and in vitro studies have been directed toward the immunomodulatory functions of KYNA. There are strong indications that the action of KYNA varies depending on whether inflammatory or homeostatic conditions are considered. Under homeostatic conditions, KYNA induced interleukin 6 (IL6) mRNA expression 2 h after treatment in the breast cancer cell line MCF-7 (27) and cytokine secretion (TNF, IL6, IL1β, and IL10) in primary murine splenocytes after 72 h (62). A further in vitro study indicated that KYNA may be an early mediator of leukocyte recruitment, acting by triggering the activation of neutrophils as well as the adhesion of monocytes to fibronectin and intercellular adhesion molecule 1 via β1-/β2 integrin (63). In contrast, KYNA treatment decreased the mRNA expression of IL6 after 6 h in the rat mast cell line RBL-2H3 followed by a return to baseline level after 24 h (64). In addition, 24-h KYNA treatment under homeostatic conditions did not influence IL6 or TNF secretion in the murine microglial cell line BV-2 (65). At first glance, all these findings seem to be very inhomogeneous. However, they clearly demonstrate that knowing that mRNA expression of pro-inflammatory cytokines is normally tightly controlled by mRNA decay and cytokine secretion measurement needs sufficient accumulation time, it is very important for interpretation of such results to know at which time what (mRNA expression or cytokine secretion) was analyzed. In conclusion, these studies, analyzing the effect of KYNA under non-inflammatory conditions, suggest a time- and/or cell type-dependent influence of the treatment.

The KYNA effect under inflammatory conditions appears to be more uniformly. Several in vitro studies, using various primary or immortalized leukocyte cell types, have revealed that KYNA can attenuate inflammation elucidated by different stimuli (e.g., LPS). For instance, KYNA reduces TNF expression and secretion (26, 55, 65, 66) and diminishes high-mobility group box 1 (HMGB1) protein secretion in monocytes (66, 67). Likewise, KYNA has been shown to inhibit the secretion of α-defensin HNP1–3 in granulocyte cultures (66) and reduce interleukin 4 release in T-cell receptor stimulated invariant natural killer-like T cells (iNKT) (68). Recently, Elizei et al. (69) demonstrated that KYNA reduced LPS-induced IL23 expression of dendritic cells and inhibited Th17 cell differentiation in vitro. The downregulation of the IL23/IL17 axis is known to be beneficial for anti-inflammatory treatment of many immune-mediated diseases (70).

These anti-inflammatory effects of KYNA, frequently observed in many cell models, were confirmed by in vivo studies in mice and dogs. For example, KYNA treatment inhibited the LPS induced increase of TNF and nitric oxide (NO) in mice serum and also drastically reduced LPS-induced death in those animals (67). Leukocytes of KYNA-treated mice also exhibited a reduced release of TNF in response to an ex vivo LPS challenge (55). Moreover, the increased mucosal leukocyte accumulation and the xanthine oxidoreductase activity, a predominant marker of mucosal superoxide radical production, in the gastrointestinal tract of dogs with experimental colon obstruction were reduced by KYNA treatment (71).

Kynurenic acid is a ligand of the GPR35 (26) and the AhR (27, 72) (Figures 1 and 2). The affinity of KYNA for both receptors is in the low micromolar range. However, in various inflammatory and tumor diseases high levels of this metabolite are produced, so it is not surprising that under these conditions KYNA levels are sufficient to activate these receptors (72). In addition to GRP35- and AhR-mediated signals, KYNA has a relevant role as an antioxidant and ROS scavenger (73, 74) (Figures 1 and 2). This indicates that KYNA also actively prevents tissue damage triggered by overshooting inflammation. In addition, induction of the KYNA-synthesizing branch of TRP metabolism may also be relevant for the synthesis of other TRP metabolites, such as serotonin or melatonin. Serotonin and melatonin are known immune regulators whose decrease may influence immune response (75, 76). KYNA synthesis may decrease their abundance either simply by the reduction of the necessary substrate TRP, or by direct inhibition of their synthesis, or induction of their degradation. In this regard, it was found that furafylline-mediated inhibition of CYP1A2, a “classic” AhR-inducible gene, increased 6-hydroxymethylation of melatonin in rat liver slides. This result indicates that AhR signaling, perhaps triggered by KYNA, may be relevant in melatonin catabolism (77).

G-protein-coupled receptor 35 is expressed in various subpopulations of immune cells, including peripheral monocytes (26), mast cells, basophils, eosinophils (78), and iNKT cells (68). A high level of GPR35 expression was detected throughout the digestive tract (26, 79), as well as in lung, skeletal muscle, uterus, and dorsal root ganglion (79). Moderate expression was found in heart, liver, bladder, spinal cord, whole brain, and cerebrum (79).

Recently, it was found that GPR35 is a high-affinity receptor for the mucosal chemokine CXCL17 (80). Nevertheless, KYNA was the first reported agonist ligand for GPR35. This was identified by high-throughput screening using changes of intracellular calcium (Ca²⁺) in the Chinese hamster ovary cell line, CHO, co-expressing GPR35 and a G-protein mixture as a readout (15). Further in-depth studies revealed that KYNA–GPR35 interaction inhibited N-type Ca²⁺ channels in sympathetic neurons (81) and reduced the plateau phase of ATP-induced calcium transients in astrocytes (82). The later study also demonstrated that KYNA-mediated GPR35 activation decreased forskolin-induced cAMP elevation. Furthermore, the recruitment of β-arrestin 2 mediated GPR35 internalization upon KYNA activation, which led to receptor desensitization (82, 83).

Kynurenic acid may also have an inhibitory effect on the phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt) and mitogen-activated protein kinase (MAPK) pathways. Walczak et al. (84) demonstrated that KYNA decreased phosphorylation of extracellular signal-regulated kinases (ERK) 1/2, p38 MAPK, and Akt in colon epithelial cells. They also found indications that KYNA induced accumulation of β-catenin. MAPK, PI3K/Akt and β-catenin pathways are well-known targets of GPR signaling (85, 86).Therefore, it is possible that the observed inhibition of ERK and p38, as well as the induction of β-catenin accumulation after KYNA treatment, are a consequence of GRP35 activation. Interestingly, all of these described effects of KYNA–GPR35 signaling might lead to the suppression or limitation of inflammation. Increased intracellular calcium is associated with inflammatory signal secretion (87, 88) and triggers the activation of NF-κB (89), which is an essential transcription factor in inflammation (90). The cAMP pathway is known to regulate innate and adaptive immune cell activities (91) [e.g., T-cell functions (92)]. In this respect, there is strong evidence that KYNA–GPR35-mediated inhibition of adenylate cyclase is causal for the downregulation of the IL23/IL17 immune axis observed after KYNA treatment (69). Furthermore, the PI3K/Akt pathway and MAPK’s play crucial roles in generating an inflammatory response (93, 94). Conversely, the β-catenin signaling pathway is known to inhibit inflammation through limiting NF-κB activation by stabilizing the NF-κB inhibitory IκB-factors (95).

Aryl hydrocarbon receptor is a ubiquitously expressed promiscuous ligand-operated receptor of KYNA (27), mediating crucial effects on the regulation of the immune response (96). When binding to a ligand, AhR dimerizes with the AhR nuclear translocator (ARNT) and acts as a transcription factor.

Several studies using AhR knockout mice indicated that this receptor has an important immune regulatory role in inflammation. For instance, AhR-deficient mice have a high susceptibility to LPS-induced septic shock (97) and developed a stronger response after local inflammatory challenge (or insult) in the lung (98, 99). AhR activation is involved in preventing an overshooting pro-inflammatory cytokine induction in response to an inflammatory stimulus in various cells including fibroblasts, endothelial cells, and macrophages (97, 100, 101). Therefore, it has been proposed that a clinical treatment with an appropriated AhR ligand like KYNA may offer a promising therapeutic intervention in inflammatory disorders. Interestingly, brains of AhR knockout mice exhibited an increased KAT-II expression and a higher KYNA level in the cerebral cortex and striatum that is associated with protection against oxidative stress induced by an excitotoxic insult via intrastriatal application of QUIN (102). This may be either a result of KYNA-mediated counteraction of NMDAR activation by QUIN, which is known to mediate exitotoxic properties, or of the receptor-independent antioxidative properties of KYNA discussed above. Furthermore, the data of this study indicated a negative feedback-loop between AhR, KAT II, and the AhR ligand KYNA in the brain. Unfortunately, there are no data regarding KYNA levels and KAT expression outside the brain of AhR-deficient mice. Therefore, further analyses are necessary to prove whether this is only a tissue-specific observation or a general finding. However, studies using AhR-deficient mice need to be carefully evaluated because, although these animals appeared relatively normal, this knockout influences several physiologic processes in the animals including structure of the central nervous system and blood cell differentiation (103).

In addition to KYNA, other TRP metabolites, such as the KYNA precursor KYN, have been shown to be ligands of the AhR (104). In humans, pigs, and mice KYNA normally have 3- to 10-fold lower plasma concentrations than KYN (54, 55, 105, 106). However, KYNA is described as a more potent AhR ligand (27) and has a higher stability than KYN (56). It was speculated that AhR interaction with TRP metabolites contributed to immune homeostasis during endotoxin tolerance by activating immunomodulatory signaling (104). The data of Bessede et al. indicated that AhR-associated Src activity, triggered by TRP metabolites bound to AhR, was responsible for IDO1 phosphorylation. These studies revealed that the TRP metabolite KYN, without the IDO1 protein, was insufficient to induce TGFβ expression. The authors speculated that IDO1 phosphorylation represented an independent signaling pathway necessary for TGFβ-mediated immune tolerance. Whether the TRP metabolite KYN or KYNA could be the relevant AhR ligand mediating those immunosuppressive effects remains unclear. It has been demonstrated that KYNA ligated to AhR induced IL6 mRNA expression in breast cancer cells. These authors also showed that a combination of KYNA and pro-inflammatory IL1 induced IL6 much more strongly than either of these factors alone. This synergistic activation of IL6 could be mediated by direct interaction of AhR–ARNT and the NF-κB factor RELB. NF-κB factors are known to be activated by pro-inflammatory cytokines, such as IL1. The role of AhR-NF-κB cross talk was previously described for IL8 gene expression (107). IL6 features pleiotropic activities (108). Although IL6 plays essential roles in promoting inflammation, it also has many anti-inflammatory and regenerative activities (109) reviewed in Ref. (110, 111). Therefore, it is difficult to estimate if KYNA-mediated IL6 expression contributes only to the immunosuppressive function of KYNA. However, studies have shown that IL6 is involved in the development of many chronic inflammatory and cancer diseases (111). For example, the study that demonstrated a KYNA-mediated induction of IL6 in breast cancer cells discussed this observation as part of the mechanisms allowing tumor cells to escape immune surveillance (27). A further interesting point is that IL6 can induce IDO1 via STAT3 activation. This signaling is known as the AhR–IL6–STAT3 loop, which is associated with poor prognosis in lung cancer (112). There is increasing evidence that the interaction of AhR with metabolites of the KP, such as KYNA, is relevant for maintaining the immunosuppressive microenvironment in many cancer types (72, 113). The probable mode of action is a TGFβ- and IL6-mediated suppression of T-cell response by interfering with differentiation and activation of regulatory T cells. This is still supported by the fact that various cancer cells secrete KYNA (Table 1).

Irrespective of the presumed KYNA-mediated tumor-immune escape, another research group found that high KYNA concentrations inhibit proliferation and migration of cancer cell lines in vitro (84, 140, 141). This seems to be mediated by interference with the cyclin-dependent kinase inhibitor p21(WAF1/CIP1) pathway (142). All this indicates that KYNA can act as both tumor-promoting as well as tumor-inhibiting factor. However, all mentioned studies used cancer cell lines and characterized the effect of KYNA under in vitro conditions. Therefore, it must be kept in mind that further studies are needed to validate the KYNA effects on tumor development.

Alterations of KYNA metabolism in both periphery and brain are described for several pathological states in humans (Table 1).

High GPR35 expression in the gastrointestinal tract (26, 79) indicates that this receptor, and probably its ligand KYNA, could have a function in gut homeostasis (152). The potential significance of KYNA for gut health emerges from its association with various bowel diseases and colon cancer [Table 1 (30, 138, 139)], as well as the potential anti-inflammatory effects of KYNA treatment in dogs with experimental colon obstruction (71). Studies in rats and pigs have shown a high concentration of KYNA in the intestinal lumen (153, 154). The intestinal KYNA concentration increased from the proximal to the distal part of the gut, reaching ~16 μM in the distal ileum of the rat (153) and ~1.6 μM in the colon of the pig (154). The studies in rat suggest that relevant amounts of KYNA in the gut originated from the intestinal microflora, due to the relatively low concentrations in the wall of the ileum (~0.2–0.3 μM) and the food (~0.6 μM). However, certain foods and herbs may contain relatively high amounts of KYNA like broccoli (~2 μM), honey (~1 μM), basil (~74 μM), and thyme (~9 μM) (155, 156). Furthermore, the intestinal commensal Escherichia coli can produce and liberate KYNA through aspartate aminotransferase (AspAT) (153, 157, 158). KYNA is readily absorbed from the gut into the bloodstream (155). Interestingly, rats with the probiotic Bifidobacteria infantis in the gut have significantly higher KYNA levels in the blood then un-colonized control animals (159). Furthermore, blood from B. infantis-colonized animals exhibits a lower TNF induction after ex vivo challenge with LPS, which is a typical indication of an endotoxin tolerance. There are also indications that KYNA selectively regulates the growth, and thereby the composition, of the intestinal microbiota (160). In this context, the microbial-mediated KYNA catabolism, known so far from Pseudomonas and Aerococcus (60, 61), might be relevant. Interestingly, feed supplementation with very high amounts of KYNA might have a toxic/stress-inducing effect in rainbow trout (161). Hence, further studies are necessary to evaluate if a supplementation of KYNA is beneficial or detrimental to human health.

Due to the proven relevance of KYNA for various diseases, it is often mooted as both a target and agent for therapeutic interventions. However, the interference of KYNA with diverse immune-related signaling pathways requires further in-depth analysis to avoid unexpected adverse consequences.

EW and JG have contributed equally to this work; they designed the manuscript and approved it for publication. AH and AR contributed to critically revising the paper.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.