br Conflicts of Interest br Author Contributions Z

Conflicts of Interest
Author Contributions Z.H.X. and Z.C. designed experiments and co-wrote the manuscript. Z.H.X. and Y.W. performed most of the experiment, acquired most of the data presented, and performed the statistical analyses. B.C., W.W.H., and Z.X.N. performed in imaging. C.L.X., and Y.W. participated in the animal surgeries and carried out animal care. X.H.W. and Y.W. conducted the learning & memory studies. S.H.Z., S.W., and Y.G. assisted with statistical analyses and in vivo electrophysiological recordings. J.H.L., and S.M.D. assisted with optogenetics and contributed experimental suggestions.
Introduction Depression has a pattern of familial aggregation, which implies the influence of genetic effects, common environmental effects shared by relatives, or both. The genetic component (heritability) has been estimated by a twin study of major depressive disorder (MDD) to be 37% (Sullivan et al., 2000). The SNP heritability (heritability attributed to common genetic variants) of MDD varies across populations and samples (21%–32%) (Lubke et al., 2012; Lee et al., 2013). Subsequently, a ‘children of twins’ study found a significantly greater risk of depression in children of depressed monozygotic (MZ) twins than in the offspring of their non-depressed twin. This implies a potential environmental effect of parental depression on offspring (Singh et al., 2011). Studies have also shown that having a partner with psychiatric disorder may increase an individual\'s risk of MDD (Joutsenniemi et al., 2011; Desai et al., 2012), but meta-analytic studies suggest no effect of the shared sibling environment and other studies have postulated more complex relationships (Olino et al., 2006). Whilst each of these studies separately provided evidence for the genetic and familial environmental components in depression, a precise raf inhibitors of these potential effects should involve estimating them simultaneously in the same model and has yet to be achieved. The accurate separation and estimation of the genetic and environment components on liability to depression provide crucial information, as it reveals the upper limit of the genetic effects, the probability of true positive results from genetic studies and the potential for accurate risk predictions for depression (Makowsky et al., 2011; Tenesa and Haley, 2013). Genetic studies attempting to map causal variants have been performed for various definitions of depression. These include clinically-assessed depression, self-report of clinical diagnosis of depression and self-reported depressive symptoms (consortium, 2015; Major Depressive Disorder Working Group of the Psychiatric et al., 2013; Hyde et al., 2016; Okbay et al., 2016), but the findings are generally inconsistent. Although some of the inconsistent findings were probably due to the limited power of the original studies (Flint and Kendler, 2014), there may also be intrinsic heterogeneity across depression definitions. This is further supported by the fact that studies show very different estimates of the narrow-sense heritability (h2)for several depression definitions or related traits to MDD (h2=37% (Sullivan et al., 2000)): perceived stress: 44% (Bogdan and Pizzagalli, 2009); nine depression definitions in women: 21%–45% (Kendler et al., 1992); depressive symptom scores in childhood: 79% (Thapar and Mcguffin, 1994) and depressive symptoms in an elderly population: 69% in women and 64% in men (McGue and Christensen, 2003). In fact, even for MDD, the genetic correlation of MDD phenotypes between independent datasets was relatively low compared with other psychiatric disorders (Gratten et al., 2014). Because of the heterogeneity across depression definitions, there has been a long debate about the correct phenotype for depression genetic studies. Studies using clinically-assessed depression could provide findings that are directly informative for clinical application. However, the resources required for such data collection are generally very high (consortium, 2015). As an alternative, measuring self-reported depression requires fewer resources and this phenotype is rapidly becoming available for many population-based datasets (Okbay et al., 2016; Hyde et al., 2016). To date, the largest published GWAS of major depression has yielded 15 significant loci (7 loci before meta-analysis) for a self-reported clinical diagnosis (Hyde et al., 2016).