Abstract
Background Mounting evidence shows genetic overlap between multiple psychiatric disorders. However, the biological underpinnings of shared risk for psychiatric disorders are not yet fully uncovered. The identification of underlying biological mechanisms is crucial for the progress in the treatment of these disorders.
Methods We applied gene-set analysis including 7372 gene sets, and 53 tissue-type specific gene-expression profiles to identify sets of genes that are involved in the etiology of multiple psychiatric disorders. We included genome-wide meta-association data of the five psychiatric disorders schizophrenia, bipolar disorder, major depressive disorder, autism spectrum disorder, and attention-deficit/hyperactivity disorder. The total dataset contained 159 219 cases and 262 481 controls.
Results We identified 19 gene sets that were significantly associated with the five psychiatric disorders combined, of which we excluded five sets because their associations were likely driven by schizophrenia only. Conditional analyses showed independent effects of several gene sets that in particular relate to the synapse. In addition, we found independent effects of gene expression levels in the cerebellum and frontal cortex.
Conclusions We obtained novel evidence for shared biological mechanisms that act across psychiatric disorders and we showed that several gene sets that have been related to individual disorders play a role in a broader range of psychiatric disorders.
Discussion
The current gene-set analyses revealed various new sets of genes – in particular related to the synapse and neuronal functions – and gene-expression profiles of multiple brain tissues that play a role in shared genetic risk across five psychiatric disorders. The most strongly associated gene set was the highly-brain-expressed genes, which has previously been related to ASD (Pinto et al., 2014). However, as this gene set contains over 5000 genes with many different functions, this observation particularly confirms the polygenic nature of psychiatric disorders and its association with brain processes. This finding is in concordance with our tissue-type analysis, which showed the importance of gene expression of brain tissues for psychiatric disorders. Gene expression profiles of the cerebellum showed the strongest association, which confirms studies reporting cerebellar dysfunction in various psychiatric disorders (Phillips et al., 2015). Our finding of an additional effect of expression profiles of the frontal cortex is supported by observations that dysfunction of this region and related networks underlie cognitive and behavioral disturbances in psychiatric disorders (Fornito et al., 2015).
In addition, we identified multiple gene sets related to the synapse, which aligns with synaptic functions of several identified genes for multiple individual psychiatric disorders (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Wray et al., 2018; Demontis et al., 2019; Grove et al., 2019; Stahl et al., 2019). Three of these gene sets point to a specific role of calcium channels, a well-established mechanism related to SCZ (Ripke et al., 2013; Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Pardinas et al., 2018) and suggested for BP and MDD as well (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013b; Wray et al., 2018). A common role across additional disorders is further supported by a cross-disorder genome-wide meta-analysis reporting genes related to the functioning of these channels (Schork et al., 2019). We also replicated a cross-disorder role for the postsynapse, although our findings do not support the previously reported role of histone and immune pathways (The Network and Pathway Analysis Subgroup of the Psychiatric Genomics Consortium, 2015). The target genes of MIR137, a microRNA that is one of the best replicated genetic risk factors for SCZ (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013b; Ripke et al., 2013; Pardinas et al., 2018), have not been implicated yet in other disorders and we now show that alterations in this network of genes are likely also involved in other psychiatric disorders. Multiple studies have reported the involvement of MIR137 in synaptic function, by regulating synaptogenesis, synapse maturation and synaptic transmission (Strazisar et al., 2014; Verma et al., 2015; He et al., 2018). Furthermore, our results suggest a shared role for FMRP targets which have previously been related to SCZ and ASD based on CNVs (Pinto et al., 2014; Szatkiewicz et al., 2014), de novo mutations (Iossifov et al., 2012; Fromer et al., 2014), rare variants (Purcell et al., 2014), and common variants (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Jansen et al., 2017; Pardinas et al., 2018). FMRP is an RNA-binding protein involved in the regulation of translation. The binding transcripts code mainly for postsynaptic proteins (Darnell et al., 2011), and loss of FMRP results in widespread deficits in synaptic plasticity (Darnell and Klann, 2013). Taken together, all identified gene sets converge to an important contribution of communication between neurons, which is supported by the implication of a more common role of altered cortical connectivity in psychiatric disorders (Fornito et al., 2015).
Of note, the biological annotations of gene sets comprise a complex and challenging process, e.g., due to the multiple functions of many genes and incomplete knowledge. The construction of gene sets is in general based on different approaches such as shared cellular mechanism, co-expression patterns, protein-protein interaction, or co-localization. Hence, sets of genes may be based upon different inclusion criteria, creating an overlap between gene sets, as also illustrated by the current study. Clearly, it is important to recognize the impact of particular annotations on gene-set analysis results and their biological interpretation.
To address this issue of confounding and redundancy in gene sets, we applied conditional analyses. This provided insight in how different gene-set associations relate to each other, and whether identified functions may not be biologically relevant to the disorders but rather induced by confounding factors (de Leeuw et al., 2018). Brain-specific gene expression could be such a general confounder for our identified synaptic and neuronal gene sets, but the conditional analyses demonstrated that most of their associated signals were independent of brain expression levels. The conditional analyses between the identified gene sets revealed that part of these gene-set associations is not independent, which might be induced by a more extensive underlying function. Nevertheless, several independent associations suggest that multiple synaptic mechanisms are contributing risk factors for psychiatric disorders. These mechanisms may serve as starting points for future functional studies to disentangle their relation to psychiatric disorders, and potentially provide a first resource for the identification of drug targets and for drug repositioning (Breen et al., 2016).
Our cross-disorder gene-set and gene-property analyses are built on a meta-analysis of the gene-based associations with the individual disorders, therefore possible opposite effects of genetic variants are not taken into account. To explore if genetic variants are related to multiple disorders but with opposite effects, we performed an SNP-based meta-analysis of the five disorders and conducted a gene-set and gene-property analysis based on those results. In this analysis, genetic variants with opposite effects across disorders are cancelled out. Although these results showed strong correlations with the original analysis, we detected differences in association strength that point to partial differences in direction of SNP effects between the disorders for most identified gene sets. Interestingly, the effects on calcium channel activity are unidirectional across disorders. The outcome of different effects across disorders is supported by the recent finding that the highly correlated disorders SCZ and BD are differentiated by several genetic loci with opposite directions of effects (Bipolar Disorder and Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2018). It has indeed been shown that in addition to genetic variants with effects on a general dimension of cross-disorder liability, specific variants uniquely differentiate between psychiatric disorders (Grotzinger et al., 2019). Furthermore, the general cross-disorder liability could reflect biological mechanisms that are related to specific overlapping symptoms, e.g. sleep disturbances, depressive symptoms and cognitive problems. This hypothesis is supported by the finding that polygenic components underlie multiple symptom dimensions of SCZ and BD (Bipolar Disorder and Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2018). Exploring the biological mechanisms that may drive specific symptoms across disorders is therefore required to further advance our understanding of the complexity of the genetic overlap. Moreover, the identification of these mechanisms may help to develop individual-centered therapy driven by symptoms instead of general disorders. Moving the focus from dichotomies to the level of the individual is required to advance precision medicine (Senn, 2018).
We note that the associations of our identified gene sets were to a large extent driven by SCZ. This is in line with previous studies that reported multiple gene sets associated with SCZ (Ripke et al., 2013; Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Pardinas et al., 2018). Yet, hardly any gene sets have been detected for other psychiatric disorders despite the recent successes in identifying many genetic loci for multiple disorders that resulted from the fast increase in sample sizes that approach, or even exceed, the sample sizes of SCZ studies. This suggests that less successful findings for disorders such as MDD are unlikely a result of less statistical power. One possible explanation is that the identified gene sets of the current study have a true stronger effect on SCZ. One could also speculate that that SCZ has a different genetic architecture, or is less genetically heterogeneous compared to other disorders, but future studies are needed to address these issues.
In conclusion, the current study provides novel evidence for shared biological mechanisms that act across psychiatric disorders based on gene-set and gene-property analyses. We showed that several gene sets that previously only had been associated with individual disorders also play a role in a broader range of psychiatric disorders, supporting the view of a common pathogenesis across disorders. This indicates that the genetic overlap between disorders is not randomly distributed, but can be explained by specific biological mechanisms. The strongest evidence in our results was for the involvement of synaptic functions, and gene expression profiles of the cerebellum and frontal cortex. The genetic data collection of additional psychiatric disorders is rapidly increasing and will make it possible to extend our analyses to other disorders in the near future. Understanding the shared biological mechanisms between psychiatric disorders may provide a hint towards a general vulnerability for multiple psychiatric disorders, and could result in potential treatment for a broad spectrum of psychiatric disorders.
Posts
No comments:
Post a Comment