Metal health: a shared genetic basis for schizophrenia, inflammation, lipid metabolism and cortical structure

Background: In addition to core positive and negative symptoms, psychotic disorders are frequently characterised by a pattern of multimorbidity with chronic medical conditions such as cardiovascular disease, diabetes, and chronic inflammation. 40% of antipsychotic-naïve patients with first-episode psychosis exhibit dysglycaemia [1], and low-grade chronic inflammation in adolescence has been linked to increased risk of both psychosis and insulin resistance in adulthood [2]. Emerging evidence suggests that IL-6 signalling, particularly non-classical trans-signalling, may contribute to the pathogenesis of psychosis [3]. However, the extent to which this multimorbidity is driven by shared genetic mechanisms remains unclear. We therefore aimed to understand whether the genetic interplay between peripheral IL-6 signalling, metabolic biomarkers, and neuroimaging traits could provide insights into schizophrenia (SZ) aetiology and identify novel target mechanisms.

Methods: We performed hypothesis prioritisation multi-trait colocalization (HyPrColoc) analysis [4] to examine genetic associations shared across SZ and 14 related traits: two cardiometabolic risk factors (triglycerides:HDL (TG:HDL) ratio and fasting insulin), three structural MRI traits (left- and right-hemispheric cortical thickness and peripheral grey matter volume), six fMRI independent component analysis (ICA) traits, and plasma levels of three IL-6 signalling proteins (IL-6, IL-6R, and IL-6ST). Bayesian HyPrColoc was applied to 232 SNPs associated (p<5×10^-8) with at least one trait. Prior probabilities (π) were optimised to detect shared causal variants (π₁=0.05; π₂=0.5–0.9999), quantifying the likelihood of shared genetic architecture within a 1 Mbp window.

Results: We identified one region on chromosome 4 where SZ risk colocalised with IL-6ST, TG:HDL ratio, and bilateral cortical thickness (Table 1). The likely causal variant was rs13107325, a missense SNP in SLC39A8 (p.Ala391Thr), encoding ZIP8, a multi-metal trans-membrane transporter mediating Zn²⁺, Fe²⁺, and other ions. rs13107325 was significantly associated with each colocalising trait (TG:HDL p=5.30×10^-32, β=0.07±0.004; LH cortical thickness p=8.30×10^-14, β=-0.12±0.02; RH cortical thickness p=1.00×10^-16, β=-0.12±0.02; IL-6ST p=6.79×10^-6, β=0.06±0.01; SZ p=2.90×10^-21, β=0.15±0.02). Previous studies link rs13107325 to reduced cortical dendritic spine density in vivo and altered glutamate signalling in vitro [5]. Using publicly available scRNAseq data from developing human neocortex, we showed SLC39A8 expression to be enriched in vascular cells, peaking during infancy. Expression quantitative trait loci (eQTLs) for SLC39A8 did not correlate with associated SNPs for any co-localising traits (all p>0.4), suggesting functional changes to ZIP8 rather than reduced ZIP8 expression are associated with the co-localising traits.

Conclusion: Our findings suggest that loss of ZIP8 function may increase risk for SZ, elevate IL-6ST levels, increase TG:HDL, and reduce cortical thickness. Hence, disrupted transmembrane metal ion transport may represent a shared mechanism underlying multimorbidity across SZ, metabolic disorders, and inflammation. Further Mendelian randomisation and functional assays are warranted to clarify causality.

Table 1 Results from co-localization Bayesian analysis for π₂=0.9999 and largest regional/alignment thresholds for co-localisation of >1 trait with SZ. π₁ = 0.05 in all results presented.[Table presented]