Genomic findings in schizophrenia and their implications

Abstract

There has been substantial progress in understanding the genetics of schizophrenia over the past 15 years. This has revealed a highly polygenic condition with the majority of the currently explained heritability coming from common alleles of small effect but with additional contributions from rare copy number and coding variants. Many specific genes and loci have been implicated that provide a firm basis upon which mechanistic research can proceed. These point to disturbances in neuronal, and particularly synaptic, functions that are not confined to a small number of brain regions and circuits. Genetic findings have also revealed the nature of schizophrenia's close relationship to other conditions, particularly bipolar disorder and childhood neurodevelopmental disorders, and provided an explanation for how common risk alleles persist in the population in the face of reduced fecundity. Current genomic approaches only potentially explain around 40% of heritability, but only a small proportion of this is attributable to robustly identified loci. The extreme polygenicity poses challenges for understanding biological mechanisms. The high degree of pleiotropy points to the need for more transdiagnostic research and the shortcomings of current diagnostic criteria as means of delineating biologically distinct strata. It also poses challenges for inferring causality in observational and experimental studies in both humans and model systems. Finally, the Eurocentric bias of genomic studies needs to be rectified to maximise benefits and ensure these are felt across diverse communities. Further advances are likely to come through the application of new and emerging technologies, such as whole-genome and long-read sequencing, to large and diverse samples. Substantive progress in biological understanding will require parallel advances in functional genomics and proteomics applied to the brain across developmental stages. For these efforts to succeed in identifying disease mechanisms and defining novel strata they will need to be combined with sufficiently granular phenotypic data.

Fig. 1. Effect size and frequency of known risk variants for schizophrenia.

Effect sizes expressed as odds ratios (OR) versus general population frequencies for copy number variants (CNVs), damaging rare coding variants (protein-truncating variants (PTVs) or missense variants), common single nucleotide polymorphisms (SNPs), and polygenic risk scores (PRS). The OR less than 1 of 22q11dup CNV denotes a protective effect. CNVs effect sizes are from [20]. To constrain the ORs for 22q11.2 del and PWS/AS dup below infinity, a single carrier for each CNV was added to the controls. Dup and del refer to duplication and deletions. PTV and missense associated genes are from references [29, 30]. Population frequencies and ORs for RCV associated genes are from [29]. RCV effect sizes refer to the excess burden of RCVs in the named gene. CNV and RCV effect sizes are imprecise due to the small number of observations. SNP and PRS data are from [12]. PRS ORs are given for individuals in the top centile relative to all other individuals, and top centile versus the bottom centile. Data for European, East Asian and African American genetic ancestry are given separately. The effect size in the Latino population is not plotted as it has not yet been estimated in sufficient samples. It should be noted that most rare alleles are not expected to confer large effects. The shape of the curve provides an indication of the maximum frequency that selection pressures permit alleles of a given effect size to attain, not the expected effect size for an allele of a particular frequency.

Fig. 2. Components of variance in liability to schizophrenia.

Inner ring: heritability is estimated from twin studies at around 80%. The remaining 20% attributed to non-heritable risk factors include environmental risk factors, stochastic effects, and de novo mutations. Middle ring: estimates of the contribution to variance in liability from currently known classes of heritable risk alleles. Outer ring: variance in liability assigned to specific risk alleles, or in the case of RCVs, a burden test of RCVs in associated genes. Percentages refer to variance in total liability and are based on studies of people largely of European biogeographic ancestry. Values are approximations (see text). SNP single nucleotide polymorphisms and small insertion/deletion polymorphisms with minor allele frequencies greater than 0.01. CNV large copy number variants with population frequencies less than 0.01. RCV rare coding variants with frequencies typically less than 0.0001. GWS significance surpassing the relevant thresholds allowing for multiple testing for SNPs, CNVs, and burden tests of RCVs.

Fig. 3. Synaptic location and enrichment of schizophrenia risk genes.

A Synaptic location of prioritized protein coding genes from Schizophrenia Working group of the Psychiatric Genomics Consortium [12], genes attaining FDR < 0.05 for enrichment for rare coding variants in the study of the Schizophrenia Exome Meta-analysis Consortium [29], and neurexin 1, the only schizophrenia associated CNV to implicate a single gene [20]. Plot and locations were generated and defined according to the Synaptic Gene Ontologies (SYNGO) Consortium Website (https://syngoportal.org). The data required to generate the plot and obtain granular detail of, and the evidence for, the locations and synaptic functions for each gene are available as Supplementary Table 1. Colours denote the number of genes in each cellular component. Numbers are cumulative from the periphery to the centre of the plot respectively depicting the lowest and highest levels of the hierarchy of the ontology. B Components of the SYNGO ontology hierarchy are denoted as significantly enriched for genes as in A. Colour denotes the significance of enrichment as determined by SYNGO. Enrichment is calculated relative to a background of all brain expressed genes.