DRD2 co-expression network and a related polygenic index predict imaging, behavioral and clinical phenotypes linked to schizophrenia

Genetic risk for schizophrenia (SCZ) is determined by many genetic loci whose compound biological effects are difficult to determine. We hypothesized that co-expression pathways of SCZ risk genes are associated with system-level brain function and clinical phenotypes of SCZ. We examined genetic variants related to the dopamine D2 receptor gene DRD2 co-expression pathway and associated them with working memory (WM) behavior, the related brain activity and treatment response. Using two independent post-mortem prefrontal messenger RNA (mRNA) data sets (total N = 249), we identified a DRD2 co-expression pathway enriched for SCZ risk genes. Next, we identified non-coding single-nucleotide polymorphisms (SNPs) associated with co-expression of this pathway. These SNPs were associated with regulatory genetic loci in the dorsolateral prefrontal cortex (P < 0.05). We summarized their compound effect on co-expression into a Polygenic Co-expression Index (PCI), which predicted DRD2 pathway co-expression in both mRNA data sets (all P < 0.05). We associated the PCI with brain activity during WM performance in two independent samples of healthy individuals (total N = 368) and 29 patients with SCZ who performed the n-back task. Greater predicted DRD2 pathway prefrontal co-expression was associated with greater prefrontal activity and longer WM reaction times (all corrected P < 0.05), thus indicating inefficient WM processing. Blind prediction of treatment response to antipsychotics in two independent samples of patients with SCZ suggested better clinical course of patientswith greater PCI (total N = 87; P < 0.05). The findings on this DRD2 co-expression pathway are a proof of concept that gene co-expression can parse SCZ risk genes into biological pathways associated with intermediate phenotypes as well as with clinically meaningful information.