Introduction

Elevation of brain kynurenic acid (KYNA) is a consistently found biochemical aberration in schizophrenia and bipolar disorder (BD).^{1, 2, 3, 4, 5, 6, 7} Brain KYNA is mainly produced in astrocytes as an end-metabolite of the kynurenine pathway of tryptophan metabolism. This pathway is highly inducible by inflammatory stimuli,⁸ and we have previously reported that cerebrospinal fluid (CSF) levels of the proinflammatory cytokine interleukin (IL)-1β are markedly increased in patients with BD or schizophrenia, although the majority of other cytokines measured in this study were undetectable.^{9, 10}

KYNA is a neuroactive metabolite that antagonizes the glycine co-agonist site of the N-methyl-d-aspartic acid receptor (NMDAR).⁸ Administration of synthetic NMDAR antagonists causes psychotic symptoms in healthy individuals,¹¹ and exacerbates psychotic features in patients with schizophrenia.¹² Psychotic symptoms are core features of schizophrenia, and more than half of patients with BD will experience psychosis in their lifetime.¹³ Supporting that KYNA might be specifically involved in the pathophysiology underlying psychotic symptoms, we have found higher levels of CSF KYNA in BD-I patients with a history of psychosis compared with those who had never experienced psychosis.¹⁴ KYNA also noncompetitively antagonizes the cholinergic α7 nicotinic receptor, and animal studies indicate that increased brain KYNA might cause cognitive deficits.⁸ In rats, increased brain KYNA causes behavioral responses analogous to impaired set-shifting in humans,¹⁵ an index of executive function. Set-shifting dysfunction as measured by the trail making test (TMT) is indeed a feature of schizophrenia and euthymic BD,^{16, 17} especially in BD patients with a history of psychosis.¹⁸

Family history is the strongest risk factor for BD, but an important obstacle for progress in psychiatric genetics is that psychiatric syndromes—based solely on symptom clustering—do not necessarily reflect specific underlying biological dysfunctions and may be insufficient to delineate heritable phenotypes.¹⁹ Indeed, epidemiological and molecular genetic studies have blurred the diagnostic boundary between schizophrenia and BD by demonstrating that these disorders have partly shared genetic causes.^{20, 21} Complementary approaches to unearth causal genetic mutations are therefore needed. One approach is to focus on biomarkers, that is, measurable key components in biological pathways between genotype and disease.²² For this purpose, the use of CSF KYNA may be particularly rewarding given its biological links to distinct subdomains of pathology present in both BD and schizophrenia.

In this study of euthymic BD patients, we found CSF IL-1β and KYNA to be associated with a history of psychosis and set-shifting impairment. CSF levels of KYNA were also strongly associated with the dopamine metabolite homovanillic acid (HVA). We conducted a genome-wide association study (GWAS) against CSF levels of KYNA in BD that revealed a genome-wide significant association with the single-nucleotide polymorphism (SNP) rs10158645 within 1p21.3, a finding that was replicated in an independent cohort of BD patients. Furthermore, we analyzed this SNP in relation to CSF HVA, a history of psychosis (followed by a replication in a large data set of 565 BD patients) and set-shifting ability. As the minor allele in rs10158645 was associated with decreased expression of sorting nexin 7 (SNX7), we attempted to decipher the biochemical chain of events using a multipronged approach including causal inference analyses of clinical data, post-mortem data and cell culture studies. These experiments converged on the proposal that decreased SNX7 expression is linked to increased CSF KYNA concentration and ultimately psychosis and set-shifting difficulties in BD through caspase-8-driven activation of IL-1β.