Very excited to share the 1st (!?) preprint from @GandalLabUCLA & super talented MSTP student Minsoo Kim & excellent team: "Network signature of complement component 4 variation in the human brain identifies convergent molecular risk for schizophrenia" biorxiv.org/content/10.110… 
Previous work from @s_mccarroll & others showed that increased C4A copy number --> incr expression --> incr SCZ risk, (partially) explaining the largest EUR @PGCgenetics GWAS hit in MHC region ... implicating complement signaling as ?new mechanism for SCZ nature.com/articles/natur…
But it is unknown how complement signaling is altered in SCZ brain & whether/how incr C4A relates to other established risk/protective factors. Here, we used RNAseq in frontal cortex from 531 cases & 895 controls to interrogate complement expression & networks in SCZ brain
We find a striking bimodal pattern with early components of complement signaling strongly up in SCZ and late components including C3 and microglial complement receptors mostly downregulated. These changes were largely absent in bipolar disorder samples (n=217).
Upregulation of C4A, an early component, is clearly associated with copy number variation as previously shown. We hypothesize that late component downregulation is a compensatory response to a pathologically active C4A.
Despite this strong differential expression, however, we do NOT find that the classical complement pathway (or its protein interaction partners) to be significantly enriched for common/rare genetic SCZ risk
But gene set annotations can be incomplete particularly for the complex biology of the human brain. Here, we turned to network biology to annotate brain complement/C4A function using "guilt-by-assocation". We generated C4A-seeded coexpression networks & stratified by C4A copy #
The network shows a non-linear expansion with # of C4A copies. At low risk, co-expr is mostly positive with astrocyte/microglial/NFkB genes. At higher risk (CN>2), the network grows and pulls in negative edges which are synaptic genes - a transcriptomic signature of pruning?
These negatively C4A co-expressed synaptic genes (but NOT positively co-expression astrocyte/microglial/NFkB genes) show strong, convergent enrichment for other genetic risk factors for SCZ.
We find that these C4A co-expression networks are largest (most vulnerable?) in frontal ctx & in 30-60yr age window, despite little difference in C4A expression across those regions/timepoints
Finally, we see much less C4A co-expression in M vs F, which may reflect F "protective" effect, again despite no difference in absolute expression levels. We find largest differences between M vs F C4A co-expr in mTOR signaling, oligodendrocytes, and supragranular exc neurons
We conclude that while complement signaling plays an important role in SCZ, it is less likely to be a 'core' disease pathway' but rather a peripheral, negative regulator of an underlying core synaptic pathology.
