Complement mediated synapse elimination in schizophrenia

Schizophrenia (SCZ) is a devastating psychiatric disorder with a typically age of onset in late adolescence. The heritability is estimated to be in between 60-80% and large-scale genome-wide studies have revealed a prominent polygenic component to SCZ risk and identified more than three-hundred common risk variants. Despite a better understanding of which genetic risk variants that increases SCZ risk, it has been challenging to map out the pathophysiology of the disorder. This has stalled the development of target drugs and current treatment options display moderate efficacy and are prone to produce side-effects. SCZ is generally considered a neurodevelopmental disorder and it was proposed more than forty years ago that physiological removal of less active synapses in adolescence, i.e., synaptic pruning, is increased in SCZ and hereby causes the core symptoms of the disorder. This theory has then been supported by post-mortem brain tissue and imaging studies displaying decreased synapse density in SCZ. More recently, it was then shown that the most strongly associated risk loci can largely be explained by copy numbers of a gene coding for the complement factor 4A (C4A). As microglia prune synapses with the help of complement signalling, we therefore decided to use a recently developed human 2D in vitro assay to assess microglial uptake of synaptic structures in models based on cells from individuals with SCZ and healthy controls (study I). We observed excessive uptake of synaptic structures in SCZ models and by mixing synapses from healthy controls with microglia from SCZ patients, and vice versa, we showed the contribution of microglial and neuronal factors contributing to this excessive uptake of synaptic structures. We then developed an in vitro assay to study neuronal complement deposition dependent on copy numbers of C4A in the neuronal lines. Complement 3 (C3) deposition increased by C4A copy numbers but was independent of C4B copy numbers (also unrelated to SCZ risk). Similar C4A copy numbers correlated with the extent of microglial uptake of synapses. Microglial uptake of synaptic structures could also be inhibited by the tetracycline minocycline that also decreased risk of developing SCZ in an electronic health record cohort. In study II, we cerebrospinal fluid (CSF) from first-episode psychosis patients to measure protein levels of C4A. In two independent cohorts, we observed elevated C4A levels (although not C4B levels) in first-episode patients that later were to develop SCZ and could show correlations with markers of synapse density. However, elevated C4A levels could not fully be explained by more copy numbers of C4A in individuals with SCZ and in vitro experiments revealed that SCZ-associated cytokines can induce C4A mRNA expression while also correlating with C4A in patient-derived CSF. In study III, we set-up a 3D brain organoid models to more fully comprehensively capture processes in the developing human brain and then also included innately developing microglia. We display synaptic pruning within these models and use single cell RNA sequencing to validate them. In conclusion, this thesis uses patient-derived cellular modelling to uncover a disease mechanism in SCZ that link genetic risk variants with bona fide protein changes in living patients