The kynurenine pathway is involved in bacterial meningitis

Despite the advances in antimicrobial and intensive care therapies against bacterial meningitis (BM), high mortality and morbidity rates have been observed [1]. Neurological sequelae, such as motor abnormalities, seizures, learning and memory impairment and mental retardation, are frequently reported after the disease. Cellular damage, mainly in the cerebral cortex, hippocampus and inner ear are the histomorphological correlates of these neurofunctional deficits [2]-[4].

Bacterial invasion and proliferation within the cerebrospinal fluid (CSF) induce an intense inflammatory response that leads to the activation of several metabolic pathways. One such pathway is the kynurenine (KYN) pathway (KP), which has been shown to be activated in experimental pneumococcal meningitis [5],[6]. The KP is the major route for tryptophan (TRP) oxidative degradation, and this pathway is involved in several diseases of the nervous system, including cancer, inflammatory disorders and neurodegenerative diseases [7]-[10]. The first step in the KP is the degradation of TRP catalyzed by indoleamine 2,3-dioxygenase (IDO) to generate formylkynurenine, which is rapidly converted into KYN by the action of kynurenine formamidase. KYN can be partially metabolized to kynurenic acid (KYNA) and anthranilic acid (AA) by kynurenine aminotransferase and kynureninase, respectively, and in part converted to 3-hydroxykynurenine (3-HK) by kynurenine 3-monooxygenase. Kynureninase also converts 3-HK to 3-hydroxyanthranilic acid (3-HAA), the substrate for 3-hydroxyanthranilic acid oxidase, which generates quinolinic acid (QUIN), the major substrate for NAD⁺ de novo synthesis [8],[9].

In the inflamed brain, TRP is metabolized through the KP to form neuroactive metabolites such as QUIN, an agonist of excitotoxic NMDA receptors, showing a cytotoxic effect, and KYNA, a neuroprotector antagonist of these NMDA receptors [10],[11]. Because the KP enzymes are differentially expressed in several cell types, quantitative differences in the production of neurotoxic and neuroprotective metabolites are observed [12]-[15], and an unbalance in the KP has been associated with several CNS diseases [7]-[10]. In an animal model of BM, the chemical inhibition of the KP led to decreased cellular NAD levels and increased apoptosis in the hippocampus [6]. This was associated with high activity of PARP-1, a DNA repair protein activated by DNA strand breaks caused by oxidative stress during BM. PARP-1 uses NAD⁺ as a cofactor, and its high activity induces energetic depletion, leading to the cell death and neuronal injury typical of BM [16]. Due to the involvement in NAD⁺ de novo synthesis and KYNA production, the KP is possibly a neuroprotective pathway in BM, despite its neurotoxic metabolites [6].

Important immunomodulatory properties, mainly related to the immunosuppressive effect of IDO, have also been attributed to this pathway [8],[9],[17]. A feedback mechanism in modulating the immune responses has been proposed because proinflammatory stimuli activate the KP and an anti-inflammatory effect mediated by KYNA has been observed [8],[18],[19]. IDO is preferentially induced by interferons, with IFN-γ being the main cytokine involved in its induction. A synergistic effect of IL-1β, TNF-α and IL-6 in IDO induction was also described. However, there is evidence that IDO expression can also be induced by an IFN-γ-independent mechanism that involves NF-κB and stress-activated mitogen-activated protein (MAP) kinases, such as p38 and c-Jun N-terminal kinase (JNK). Conversely, an immune-suppressive effect of the KP has also been described. QUIN and 3-HAA induce the selective apoptosis of TH1 cells through the activation of the caspase pathway; B and NK cells are also susceptible to these compounds. In addition, the expression of IDO in dendritic cells induces the generation of regulatory T-cells. Thus, high levels of IDO will result in a decline of TH1 response, accompanied by an enhanced TH2 response. Moreover, KYNA shows an inhibitory effect on TNF-α at the transcriptional level and as a ligand of GPR35. This inhibition of TNF-α by KYNA may be an important factor in its neuroprotection [8],[9].

In a previous work, we identified the association of the SNP AADAT + 401C/T (kynurenine aminotransferase II -- KAT II) with the host immune response to BM, and our results suggested that this SNP may affect the host’s ability to recruit leukocytes to the infection site [20]. This evidence raises the hypothesis that the KP plays an important role in the pathogenesis of BM in humans. In the present work, we measured the concentrations of metabolites of the KP in CSF samples from patients with meningitis and analyzed their correlation with cytokines, inflammatory modulators previously reported to regulate the pathogenesis of BM. We also investigated the correlation of KAT II genotypes with KYNA levels and the disease.

CSF concentrations of KP metabolites have been demonstrated in a variety of neurological diseases [7]-[10]. In the present study, the concentration of TRP and three KYN metabolites, IDO activity and cytokines were assessed in the CSF of patients with meningitis. The obtained data demonstrated an increase in KYN metabolites in CSF and suggested a role of IDO, kynureninase and anthranilate 3-hydroxylase in accelerating the synthesis of KYN, KYNA and AA in human BM patients, corroborating the findings from animal models [5],[6].

Inflammation in the CNS results in the recruitment of activated T-cells and macrophages from the periphery into the neural parenchyma [21]. Activated T-cells produce cytokine mediators of the inflammatory response, including the macrophage-activating cytokine IFN-γ, which induces IDO activity [22],[23]. In our work, TRP levels are not altered in the CSF of patients with BM or AM compared to NM. However, efficient conversion of TRP to KYN could be documented. IDO activity in the CSF was significantly higher only in the BM group, in agreement with the low IFN-γ level observed in the AM group. Conversely, a strong immune response during BM has been positively correlated with IDO activity. In some non-inflammatory and inflammatory diseases including BM, the CSF QUIN, KYN and KYNA levels were also correlated with immune markers, such as neopterin, white blood cell counts and lgG levels, indicating a close relationship between the KP and the brain’s inflammatory response [24].

Reports linking increased KYN metabolism with neuronal injury, particularly during neuroinflammatory disease, have emerged. This neuronal injury would be mainly attributed to the production of neurotoxic metabolites and oxidative molecules, such as superoxide anion and hydrogen peroxide, during the metabolism of KYN [7]-[10],[25],[26]. In contrast, in vitro studies have attributed antioxidant activity to certain KP metabolites [27]-[29].

Due to these roles, the KP has been investigated as a target for therapeutic intervention. The pharmacological modulation of KYN metabolism can affect de novo NAD synthesis [6],[30]. NAD is involved in many metabolic processes, being an essential co-factor for several enzymes, including PARP-1, which plays a crucial role in the development of meningitis-associated central nervous system complications [16]. In this sense, the KP intervention can affect NAD production and consequently PARP-1 activity, as the inhibition of the KP induced an increase in apoptosis in a BM animal model [6].

During the course of bacterial meningitis, anti-inflammatory cytokines are produced to mitigate the induced inflammation [4],[31]. Some studies have proposed that cells of the immune system use a bidirectional communication in which TRP catabolism through IDO activity drives the generation of IL-10-producing regulatory T-cells [8],[9],[18],[19]. In BM patients, we observed a positive correlation between IDO activity and the inflammatory cytokines IFN-γ and IL-1Ra (Table 2). Although each cytokine displays different roles in the inflammation during BM, IFN-γ seems to have a crucial role, mainly during pneumococcal meningitis [32]-[35], showing to be involved in deleterious proinflammatory effects [34],[35]. Concomitantly, KYN metabolites showed a positive correlation with TNF-α and KYNA metabolites with IL-10. In contrast, AM did not show a strong immune response or a high level of IL-10, despite a significant amount of KYN in the comparison with the NM group. Therefore, we propose that KYN metabolites alone are not able to trigger the cascade of inflammatory mediator production. In animal models, long-term diurnal hypoactivity and nocturnal hyperactivity have been reported in wild-type mice, while these changes were not observed in IDO1(-/-) mice. However, no protection against developing long-term cognitive deficits was observed in IDO deficient mice. These data suggest that KP may be involved in some behavioural sequelae of pneumococcal meningitis and may acts synergistically, or independently of, other metabolic pathways to cause different types of neurological sequelae [36].

Recently, our group demonstrated that a polymorphism in the kynurenine aminotransferase II gene (AADAT + 401C/T) was associated with reduced levels of TNF-α, IL-1β, IL-6, MIP-1 /CCL3 and MIP-1/CCL4 in BM patients; a reduction in cell count with a high correlation with cytokine and chemokine levels was also observed, suggesting a reduced leukocyte recruitment ability in these patients [20]. This SNP is located in a putative exonic splicing silencer (ESS), which may result in a quantitative increase in the production of mRNAs and protein [37]. In agreement with this proposition, analyzing the data from KYNA measurements in the patients homozygous for the TT genotype and comparing with the CC homozygous, we observed an increase in KYNA levels (Figure 4). Interestingly, our results demonstrated a positive correlation between KYNA and IL-10, an anti-inflammatory cytokine that was previously reported to down-regulate the expression of proinflammatory cytokines such as TNF-α. Similar data were obtained in a study carried out by Hsieh et al., showing that KYNA improves the outcomes of heatstroke in rats. In this model, KYNA inhibited the expression of inflammatory molecules such as TNF-α and ICAM-1 and enhanced IL-10 levels [19]. Corroborating the findings regarding neuroprotection, the reduced production of KYNA in astrocytes has been proposed to increase neurological symptoms of cerebral malaria [38]. KYNA was identified as a ligand for the receptor for GPR35, as it was able to attenuate LPS-induced TNF-α secretion in a dose-dependent manner. Because the TRP metabolic pathway is activated by proinflammatory stimuli, the anti-inflammatory effect of KYNA suggests an interesting feedback mechanism in modulating immune responses [18].

In addition to cytokines, other molecules can regulate the KP. There are known interactions between TRP depletion and nitric oxide (NO) production. Several findings have demonstrated that NO is able to inhibit the IDO enzyme by direct interaction or accelerating proteasomal degradation [39],[40]. Furthermore, as shown in the report from Samelson-Jones and Yeh, several cellular factors, including pH, redox environment, NO, and L-Trp abundance can be considered as additional aspects in the regulation of IDO activity [41].

Altogether, these findings suggest that IDO has prominent importance during bacterial meningitis due to the following: its capacity to maintain PARP-1 activity through NAD synthesis, protecting against cell death [6],[16]; the generation of reactive oxygen species (ROS), helping the defense mechanisms in addition to maintaining IDO activity [8]-[10],[26]; its capacity to produce some antioxidant metabolites [27]; and regulating the immune response via the known vessel-relaxing ability which KYN possess during inflammation and other immunosuppressive mechanisms [18],[19],[42]. We also suggest that AADAT + 401C/T patients can present a better outcome during BM disease via a reduced inflammatory response and increased capacity of the neuroprotective role of KYNA, which requires further investigation. Conversely, this reduced response to a pathogen at the beginning of infection may favor invasion of the microorganism, increasing the susceptibility to BM. However, the small number of patients analyzed and the heterogeneity between the groups are limitations of our work, making necessary further studies with a larger cohort for a better understanding of the role of KYNA in the outcome of bacterial meningitis.

In conclusion, our results demonstrate a specific upregulation of the KP in BM patients and suggest that KYN metabolites contribute directly to the disease. Despite the neurotoxic properties reported in the literature to date, we propose a positive and essential role of this metabolic pathway in BM.

LGC, SLL and LFAL proposed the study design. LGC, SC, CLB, FRSS, FLF and DG performed the assays and data analysis. LGC, SC, SLL and LFAG wrote the manuscript. All authors have read and approved the final version of the manuscript.