BackgroundN-methyl-D-aspartate receptor (NMDAR) hypofunction is an important pathophysiological mechanism in schizophrenia. At the postsynapse the NMDAR interacts with the post-synaptic density (PSD). Neuronal nitric oxide synthase 1 (NOS1) binds to the PSD scaffolding proteins PSD-93 and PSD-95, enabling NMDAR-mediated release of nitric oxide via NOS1. NOS1AP (adaptor of NOS1) is capable of disrupting the interactions between NOS1, PSD-93, and PSD95. Therefore, NOS1AP is closely involved in both glutamatergic and nitrinergic neurotransmission. NOS1AP has been implicated as a risk gene for schizophrenia and cognitive dysfunction. Its increased expression has been observed in dorsolateral prefrontal post-mortem brain tissue of patients with schizophrenia, and NOS1AP SNPs have been associated with established schizophrenia endophenotypes. These findings suggest that the influence of NOS1AP variants should be observable in neural systems implicated in schizophrenia. In the present study, we investigate the impact of NOS1AP and its interaction partners at the glutamatergic synapse on the cortical working memory (WM) networks using fMRI and a gene set analysis approach.Methods97 right-handed individuals with no personal or family history of psychiatric disorders underwent fMRI in a 3T Siemens Trio scanner during the performance of a visuospatial change detection WM task. Data analysis in Brain Voyager QX 2.8 included standard data preprocessing. Additionally, a multiscale curvature driven cortex based alignment procedure was used to minimize macro-anatomical variability between subjects. Subsequently, data were analyzed using a random-effects multi-subject general linear model. We investigated 19 regions of interest (ROIs) within the core fronto-parietal WM network. We studied all phases of our WM paradigm (encoding, maintenance, retrieval), which were modeled by a total of 5 regressors (encoding, delays 1–3, retrieval). Genetic data was quality controlled and imputed using the RICOPILI pipeline. Gene-set analyses of the 19 ROIs were performed using MAGMA. Two gene sets were selected: 1) NOS1AP/NOS1; 2) NOS1AP/glutamatergic synapse. We applied a Bonferroni correction for the total of 19 ROIs and 5 regressors (95 tests) to both analyses.ResultsBoth gene set analyses revealed multiple associations between brain activation in core fronto-parietal WM areas. For the NOS1/NOS1AP set, most associations were observed during the late maintenance phase (Delay 3) of our WM paradigm. One association was significant Bonferroni correction: a cluster in the left intraparietal sulcus during the late maintenance phase (Delay 3; β=2.2459, SD=0.0239, SE=0.6451, p=0,00025). For NOS1AP / glutamatergic synapse interaction partners, two associations were significant after Bonferroni correction: a cluster in the right IPS during the early maintenance phase (Delay 1; β=0.8525, SD=0.0257, SE=0.2127, p=0.0000308) and a cluster in a different part of the right IPS during the late maintenance phase (Delay 3; β=0.7186, SD=0.0216, SE=0.2119, p=0,000348).DiscussionIn our gene set analyses we observed multiple associations between brain activation during WM and NOS1AP and its interaction partners, which were most pronounced during the late maintenance phase of our WM task in bilateral areas within the IPS. Both the more constrained NOS1AP / NOS1 gene set and the NOS1AP / glutamatergic synapse gene set showed similar association patterns. Our results implicate the NOS1AP interactome and the glutamatergic system in information processing and brain function in a cognitive domain strongly impaired in schizophrenia. They also indicate that altered activation of parietal WM areas during the maintenance phase is most strongly affected.