Neuroinflammation and the kynurenine pathway: microglia, KYNA, and QUIN
How immune signals direct tryptophan into the kynurenine pathway, how astrocytic and microglial branches differ, and where clinical conclusions from biochemistry end.
How immune signals direct tryptophan into the kynurenine pathway, how astrocytic and microglial branches differ, and where clinical conclusions from biochemistry end.
Depression cannot be explained by a single serotonin deficiency, but it would also be a mistake to declare inflammation its sole cause. Signs of immune activation are found in some people with depressive symptoms, and experimental inflammatory signals alter tryptophan metabolism and glial-cell activity. The kynurenine pathway is interesting precisely because it is where these processes meet: it begins in peripheral tissues, continues in the brain, and produces compounds with different effects on neurotransmission. Yet a convincing mechanism in a cellular diagram is not the same as a diagnostic test or a ready-made treatment plan for a particular person.
Tryptophan is an essential amino acid obtained from food. A small share of the available substrate enters the serotonin branch: tryptophan hydroxylase TPH1 works mainly in the periphery, while TPH2 works in serotonergic neurons; aromatic amino-acid decarboxylase then converts 5-hydroxytryptophan (5-HTP) to serotonin. In the pineal gland, AANAT and ASMT convert serotonin to melatonin. A substantial share of tryptophan catabolism is associated with the kynurenine pathway, which both supports immune functions and contributes to NAD⁺ synthesis. The proportions allocated to each pathway depend on tissue, diet, and the body’s state, so the “95% versus 5%” rule should not be mechanically applied to an individual patient’s brain.
The first step in the kynurenine branch opens the indole ring: L-tryptophan + O₂ → N-formyl-L-kynurenine. Heme-containing TDO or IDO1 catalyzes the reaction, after which formamidase produces L-kynurenine. TDO is especially important in the liver and is regulated, among other things, by glucocorticoids and substrate availability. IDO1 is activated by immune signals in many cell types; IDO2 is a distinct enzyme with different catalytic activity and should not be treated as a simple copy of IDO1. Thus, a rise in the kynurenine/tryptophan ratio may reflect a shift in metabolism, but by itself it does not show which enzyme caused it.
| Checkpoint | Main context | What matters for interpretation |
|---|---|---|
| TDO | Primarily the liver; regulated by substrate and glucocorticoids | Peripheral kynurenine does not always indicate inflammatory induction of IDO1. |
| IDO1 | Immune cells and other tissues during cytokine responses | IFN-γ is a potent inducer; combinations of cytokines modify the response. |
| IDO2 | Separate immune regulation | Its contribution to total pathway flux depends on tissue and conditions; a single test cannot distinguish it. |
Interferon-γ activates JAK1/JAK2 through the IFNGR receptor, leading to STAT1 phosphorylation and IRF1 formation; these factors increase IDO1 transcription. NF-κB and other inflammatory signals can modify the response to interferon. This cascade explains how immune activity can redirect metabolic flux, but the phrase “tryptophan depletion” should not create the false impression that serotonin in neurons necessarily falls to a clinically significant level in everyone with inflammation. Antidepressants also act in more complex ways than simply blocking one transporter, and a single tryptophan measure is not enough to predict response to an SSRI.
TDO is encoded by the TDO2 gene, while IDO1 and IDO2 are encoded by the separate genes IDO1 and IDO2. All three enzymes catalyze tryptophan dioxygenation, but they differ in distribution, substrate affinity, and regulation; the chemically identical first product therefore does not reveal which tissue made the main contribution. Hepatic TDO helps maintain systemic tryptophan metabolism and responds to glucocorticoid signals. IDO1 is especially important in immune regulation, including restricting tryptophan availability in an inflamed microenvironment and establishing tolerance. The properties of IDO2 do not allow it to be assigned the same catalytic role as IDO1 automatically, even when both genes are detected in immune cells.
The first product, N-formyl-L-kynurenine, is hydrolyzed to L-kynurenine by formamidase. This is the shared node from which KAT-dependent transamination and KMO-dependent hydroxylation diverge. The serotonin and kynurenine pathways should not be pictured as two communicating vessels with an unchanging total: intracellular tryptophan availability, transporter activity, protein synthesis rates, and local enzymes alter flux independently. The Kyn/Trp ratio measured in blood is the net result of many processes, not a direct measurement of competition between TPH2 and IDO1 inside one neuron.
When IFN-γ binds its receptor complex, IFNGR1/IFNGR2 activates JAK1/JAK2. Phosphorylated STAT1 forms a dimer, enters the nucleus, and binds GAS elements; the IRF1 it induces acts through interferon-stimulated response elements (ISRE) in the IDO1 regulatory region. TNF-α and IL-1β signals can modify the strength of this response through IKK, IκB degradation, and NF-κB release; stress-activated MAP kinases also participate in different cell types. This is an example of transcriptional cooperation, not proof of fixed synergy in every tissue. IL-6 is also a marker and mediator of inflammatory state, but its contribution to specific IDO1 expression depends on the model.
Kynurenine is produced outside the brain as well. Tryptophan and kynurenine can cross the blood–brain barrier through the large neutral amino acid transporter LAT1 in complex with 4F2hc; other amino acids also compete for transport. 3-Hydroxykynurenine can also reach the central nervous system. By contrast, kynurenic acid (KYNA) and quinolinic acid (QUIN) cross an intact barrier only to a limited extent, so their blood concentrations are not a direct snapshot of the brain’s microenvironment. Barrier permeability and the contribution of peripheral cells may change during illness; the absolute claim that “everything is synthesized only in the brain” is just as inaccurate.
Kynurenine that enters brain tissue does not have one predetermined fate. Astrocytes are especially associated with KYNA synthesis through kynurenine aminotransferases (KAT), particularly KAT II. This transamination requires pyridoxal phosphate (PLP, the active form of vitamin B₆): L-kynurenine + 2-oxoglutarate → kynurenic acid + L-glutamate. Microglia and infiltrating macrophages show a more prominent KMO branch, which leads to 3-hydroxykynurenine (3-HK), then through kynureninase (KYNU) and 3-hydroxyanthranilate 3,4-dioxygenase (3-HAO) to the intermediate ACMS. This intermediate can spontaneously cyclize to QUIN; the alternative decarboxylating reaction catalyzed by ACMSD diverts some of the flux away from it. Cells exchange metabolites in reality, so the “astrocyte–KYNA, microglia–QUIN” distinction is a useful map of predominant pathways, not a strict ban on all other reactions.
Quinolinic acid itself is not merely waste. The enzyme QPRT converts it into a precursor for NAD⁺ synthesis; the concern is an imbalance between production, clearance, and local concentration. Likewise, KYNA cannot automatically be called “good” at every level: its antagonism at NMDA receptors may limit excessive excitation, but excessive suppression of transmission can impair cognitive functions. In model systems, KYNA competes with glycine at the NMDA receptor co-agonist site, reducing the likelihood that the channel will open at a given glutamate level. Interactions with α7 nicotinic acetylcholine receptors and GPR35 have also been described. These observations depend on concentration, cell system, and method; they do not establish either “universal NMDA blockade” or an inevitable improvement in memory. Excessive KYNA may have the opposite effect by weakening normal excitatory transmission needed for learning and working memory. The biological meaning of the KYNA/QUIN ratio therefore depends not on the slogan “the more KYNA, the better,” but on receptor and cellular activity in context.
KAT I–IV are a family of aminotransferases, not one “neuroprotective enzyme.” The best-studied in the human brain, KAT II, is associated with the AADAT gene, but isoform contributions differ across cells and conditions. PLP temporarily accepts the amino group from L-kynurenine; the carbon skeleton then cyclizes to KYNA, while the amino group is transferred to 2-oxoglutarate to form glutamate. Thus, the equation L-Kyn + 2-oxoglutarate → KYNA + L-glutamate describes the net reaction, not a one-step direct conversion. The glutamate formed in this reaction should not automatically be equated with pathological glutamate release into the synaptic cleft.
On the other branch, KMO is located on the outer mitochondrial membrane and uses FAD, NADPH, and oxygen to hydroxylate kynurenine to 3-HK. Then PLP-dependent KYNU removes the side chain to form 3-hydroxyanthranilate; iron-containing 3-HAO opens its aromatic ring to produce unstable ACMS. If ACMS cyclizes without an enzyme, QUIN is formed. When ACMS is decarboxylated by ACMSD, an alternative intermediate is produced from which other metabolites, including picolinic acid, may form. Finally, QPRT attaches a phosphoribosyl residue to QUIN and introduces its carbon into the NAD⁺ synthesis pathway. These reactions form a network of competing rates: the concentration of a single metabolite does not determine the flux through every step.
3-hydroxykynurenine can participate in redox reactions and produce superoxide, peroxide, and other reactive oxygen species under suitable conditions. But 3-HK is not “toxic by definition”: its effect depends on concentration, the accompanying antioxidant systems, and the cell type. When discussing vitamins, it matters that PLP serves both KAT and KYNU, while FAD is required by KMO; this diagram alone cannot predict the direction of a shift caused by taking B₆ or B₂.
KMO is an FAD-dependent enzyme on the outer mitochondrial membrane: L-kynurenine + NADPH + H⁺ + O₂ → 3-HK + NADP⁺ + H₂O. In model systems, 3-HK may support the formation of reactive oxygen species. KYNU uses PLP to convert 3-HK to 3-hydroxyanthranilic acid, and 3-HAO opens the ring to form ACMS. These reactions also matter because the simple division of vitamins into “protective B₆” and “harmful B₂” is wrong: PLP is required by both KAT and KYNU, while FAD supports KMO’s ordinary enzymatic function. These facts do not imply a therapeutic vitamin dose for “switching” pathway branches. ACMS is another genuine branch point: without enzymatic intervention it cyclizes to QUIN, while ACMSD diverts some substrate along an alternative route. QUIN can enter QPRT-dependent NAD⁺ synthesis, so its accumulation cannot be explained by production alone; the tissue’s capacity to use the product also matters. Simplified pathway diagrams sometimes present high QUIN production and limited utilization as an inevitable result of any inflammation. In practice, the balance depends on location, timing, and the activity of several enzymes.
In experimental conditions, QUIN activates NMDA receptors and may increase Ca²⁺ influx into neurons. With excessive stimulation, calcium overloads mitochondria, disrupts their membrane potential, and triggers oxidative and proteolytic reactions. Additional effects described in models include reduced astrocytic glutamate uptake through EAAT2/GLT-1 and iron-dependent radical formation. A feedback loop may result: more extracellular glutamate, more receptor stimulation, and greater stress on neurons and glia. But concentrations, cell types, and exposure durations differ across experiments; finding QUIN in plasma alone does not mean that synapses are already dying or that demyelination is occurring in a person. In neuronal cultures, if Ca²⁺ influx remains above the buffering capacity for a prolonged period, mitochondria take up calcium, lose some of their membrane potential, and may open the mitochondrial permeability transition pore (mPTP). Reactive oxygen species then form and cell-death pathways, including the caspase cascade, may be activated. In neighboring astrocytes, reduced EAAT2 function leaves more glutamate outside the cell. Iron–QUIN complexes are considered an additional source of oxidative stress on membranes; oligodendrocyte and myelin vulnerability has been described in experimental systems. This is a sequence of possible cellular effects under a suitable exposure, not a proven chain from a serum test to brain injury in a particular patient.
Synaptic plasticity also depends on BDNF, CREB, mTOR, and the state of dendritic spines. Inflammation and excess glutamatergic load may affect these systems, helping generate hypotheses about memory, motivation, and mood. Reduced BDNF-dependent signaling, changes in CREB and mTOR, and changes in postsynaptic-density protein PSD-95 and synaptophysin are discussed as ways in which prolonged cellular stress could alter dendritic spines. At the behavioral level, this cascade is sometimes linked to anhedonia and cognitive difficulties, but one mechanism does not explain the full range of depression. Associations between metabolites and hippocampal volume in clinical studies remain associations: other biological and life factors also change the size of this structure. Similarly, the rapid effect of ketamine does not prove that QUIN caused the original symptoms; NMDA signaling participates in several different processes.
An NMDA receptor opens only when several conditions coincide: glutamate occupies its site, glycine or D-serine occupies the co-agonist site, and depolarization relieves the voltage-dependent Mg²⁺ block of the pore. In model systems, KYNA interferes with occupancy of the co-agonist site; QUIN can activate the glutamate site of certain NMDA receptors. The word “balance” therefore refers to a functional network, not simple arithmetic between two substances. The precise contribution of GluN2A and GluN2B subunits varies with brain region and cellular context, and the effects of high experimental QUIN concentrations cannot be treated as a map of every receptor in the living brain.
If activation persists, Ca²⁺ triggers enzymes that remodel the cytoskeleton and postsynaptic machinery; mitochondria simultaneously try to buffer the incoming calcium. When adaptive capacity is exceeded, CREB-dependent transcription, BDNF/TrkB signaling, and proteins that support spines and the vesicular machinery change. Further reduction in astrocytic glutamate clearance through EAAT2 intensifies the extracellular signal. In the original model, these stages are joined in a direct line to apoptosis, but in a living person feedback, recovery, and other causes of network change intervene. That is why cellular excitotoxicity and clinical depression represent different levels of evidence.
Depression studies have found changes in inflammatory cytokines and kynurenine metabolites in some groups, but results are heterogeneous. A postmortem study by Steiner found more QUIN-positive microglia in a specific region of the cingulate cortex in severe depression. Savitz and colleagues reported a lower KYNA/QUIN ratio in treatment-resistant depression, while cerebrospinal-fluid studies examined the association between QUIN and suicidal behavior. None of these observations turns a specific QUIN value or metabolite ratio into a stand-alone diagnosis, a suicide-risk prediction, or a reason to forgo standard care. Anhedonia is associated with reward circuitry and dopaminergic signaling, but the presence of inflammatory metabolites does not prove that QUIN disrupted that circuitry in a particular person. Suicidal behavior is equally multifactorial: cerebrospinal-fluid research results cannot be used as an individual test of intent or risk.
The infliximab clinical trial in treatment-resistant depression is a useful lesson in patient heterogeneity: it did not show convincing overall benefit for the full sample, while a baseline inflammatory profile was discussed as a possible response modifier. Omega-3 studies found changes in depressive symptoms in some groups and regimens, but a meta-analytic signal does not mean that EPA has been proven to “switch off IDO1” in a particular patient. Anti-inflammatory treatment and kynurenine-metabolism modification remain subjects of research alongside psychotherapy and standard pharmacotherapy.
The Kyn/Trp ratio is usually calculated from kynurenine and tryptophan concentrations in the same sample. It is sometimes used as an indirect indicator of kynurenine-pathway activity, but it is affected by TDO, diet, kidney function, the measurement method, and inflammation outside the brain. The inverse ratios KYNA/QUIN and QUIN/KYNA must be distinguished: if the same analytes are measured in the same medium, an increase in one means a decrease in the other. Plasma and cerebrospinal fluid should not be mixed in one formula, and research thresholds should not be presented as universal clinical cutoffs.
| Measure | What it may show | What it does not prove |
|---|---|---|
| Tryptophan, kynurenine, and Kyn/Trp in one medium | Substrate availability and an overall shift in the pathway at the time of sampling | Isolated IDO1 activity or serotonin at the synapse. |
| KYNA, QUIN, and their ratio | A research description of metabolite distribution | Neurotoxicity in an individual patient based on serum alone. |
| hs-CRP, IL-6 | Context of systemic inflammation | The cause of depression or cytokine concentration in the brain. |
| NfL | Under appropriate indications, damage to neuronal structures | Specific injury caused by QUIN; the measure depends on age and disease. |
These tests may be used in research or ordered for other clinical indications, but there is no standardized diagnostic “kynurenine algorithm” for depression. A clinician assesses symptoms, duration, risk of self-harm, coexisting conditions, and medications. Thoughts of suicide require immediate professional help, not independent interpretation of biomarkers.
Agudelo and colleagues showed in mice that muscle PGC-1α1 expression increases KAT, converts some circulating kynurenine to KYNA, and is associated with resilience to stress. This is a compelling example of a muscle–brain connection: peripheral conversion may reduce the substrate available to central branches. In the proposed model, muscle contraction activates energy and calcium signals associated with AMPK and CaMK, then increases expression of the coactivator PGC-1α1. Regulation of muscle genes increases KAT activity: some kynurenine available in blood is converted to KYNA, which crosses an intact blood–brain barrier much less readily. This reduces one potential source of substrate for central branches. But the experiment does not establish a specific exercise “zone,” frequency, or duration for treating depression in humans. Physical activity may be part of a coordinated care plan for general clinical reasons, and its effect cannot be reduced to kynurenine alone.
KMO inhibitors, including the research compounds Ro 61-8048 and JM6, changed metabolites and outcomes in preclinical models. They are not ready-to-use drugs for self-treatment of depression. Curcumin, resveratrol, EGCG, sulforaphane, and N-acetylcysteine have each shown different anti-inflammatory or antioxidant effects in laboratory systems; this does not mean that taking any form of these substances lowers brain QUIN in humans or replaces psychiatric care. The proposed targets differ: for EPA, specialized inflammation-resolving mediators and cytokine signals are discussed; for curcumin and EGCG, NF-κB or JAK/STAT in experimental systems; for resveratrol, SIRT1 and the inflammatory response; and for sulforaphane, Nrf2-dependent antioxidant enzymes. None of these signaling pathways makes an individual substance a proven IDO1 inhibitor in humans. Translation is further limited by product composition, bioavailability, duration of use, and tissue exposure. For N-acetylcysteine, glutathione-precursor supply and effects on glutamate metabolism are discussed; the cystine–glutamate antiporter and EAAT2 pathway does not establish that the compound clinically protects synapses from QUIN. Concentrated extracts have their own limitations and interactions; in particular, a safe dose cannot be inferred from a cell experiment.
The physiological block of the NMDA channel by Mg²⁺ ions also does not mean that oral magnesium “shields” the receptor from QUIN in depression. PLP participates in KAT and KYNU, and FAD in KMO, but therapeutic switching of these enzymes with high vitamin doses has not been shown. Nutrient deficiencies can be considered as part of ordinary clinical assessment, without promising a specific correction of the kynurenine pathway.
Direct KMO inhibition is a logical research target because the enzyme is upstream of 3-HK formation. In inhibitor experiments, peripheral and central metabolites change, but drug passage across the barrier, the direction of Kyn transport, and compensatory KAT activity require separate analysis. The Zwilling study showed protective effects in experimental neurodegeneration models; it cannot be relabeled a clinical trial in treatment-resistant depression. Even a reduction in tissue QUIN would be an intermediate biomarker until safety and benefit for meaningful symptoms in humans are demonstrated.
For omega-3, the clinical question differs from the mechanistic one. Meta-analyses study depressive symptoms across heterogeneous products and populations; they do not measure mandatory IDO1 suppression by resolvins in every participant. Plant polyphenols add another chain of uncertainty: the parent molecule, its metabolites, blood concentration, and access to the brain may differ substantially. An “integrative” combination of EPA, curcumin, resveratrol, EGCG, sulforaphane, NAC, magnesium, and vitamins has not been tested as a unified kynurenine protocol. Adding up cellular mechanisms does not create evidence that such a mixture is effective or safe.
The kynurenine pathway links immune response, tryptophan metabolism, and glutamatergic transmission. The enzyme reactions themselves and some cellular effects of metabolites are described most reliably; links with depression in humans are more complex and depend on subgroup, sample, and study method. Therefore, Kyn/Trp or QUIN/KYNA should not be used as a personal “neuroinflammation index,” and a mechanistic diagram should not be treated as a supplement prescription.
For someone with depressive symptoms, a comprehensive clinical assessment and treatment with demonstrated benefit matter more than independently selecting a panel of rare metabolites. Research on the kynurenine system may eventually improve patient selection and reveal new therapeutic targets. For now, its main practical value is to help us understand more precisely why immune, metabolic, and neural processes cannot be considered in isolation.
This is a narrative review of the kynurenine pathway based on a previously assembled selection of 18 publications on enzyme reactions, glia, preclinical interventions, and clinical observations. Editorial work was completed on October 4, 2026. No systematic search using a prespecified protocol, meta-analysis, or formal risk-of-bias assessment of each study was conducted; the source list is not exhaustive.
Cellular and animal mechanisms are considered separately from observations in people. An association between an inflammatory marker and symptoms does not establish causality or the effectiveness of an intervention. Interpretation depends on the cellular compartment, measurement method, population, and endpoint. Unvalidated diagnostic thresholds and treatment promises are not inferred from a single biochemical pathway diagram; likewise, a study’s absence from this selection does not mean that no evidence on the question exists.
No. The ratio reflects an overall shift in metabolism in the sampled material; other enzymes and factors affect it, and blood is not a direct measurement of brain tissue.
No. QUIN participates in NAD⁺ synthesis, while excess KYNA can disrupt normal neurotransmission. Local balance and context matter, and a single serum value cannot reveal them.
There is no standardized clinical algorithm with universal thresholds. Decisions about care are based on assessment of symptoms, risk, and coexisting conditions.
No. The mechanism was shown primarily in an experimental model; exercise intensity and volume for an individual are selected according to ordinary clinical indications.
Cell and animal data do not confirm that taking supplements independently redirects these branches in the human brain or replaces depression treatment.