Rare tandem repeat expansions associate with genes involved in synaptic and neuronal signaling functions in schizophrenia

Abstract

Tandem repeat expansions (TREs) are associated with over 60 monogenic disorders and have recently been implicated in complex disorders such as cancer and autism spectrum disorder. The role of TREs in schizophrenia is now emerging. In this study, we have performed a genome-wide investigation of TREs in schizophrenia. Using genome sequence data from 1154 Swedish schizophrenia cases and 934 ancestry-matched population controls, we have detected genome-wide rare (<0.1% population frequency) TREs that have motifs with a length of 2-20 base pairs. We find that the proportion of individuals carrying rare TREs is significantly higher in the schizophrenia group. There is a significantly higher burden of rare TREs in schizophrenia cases than in controls in genic regions, particularly in postsynaptic genes, in genes overlapping brain expression quantitative trait loci, and in brain-expressed genes that are differentially expressed between schizophrenia cases and controls. We demonstrate that TRE-associated genes are more constrained and primarily impact synaptic and neuronal signaling functions. These results have been replicated in an independent Canadian sample that consisted of 252 schizophrenia cases of European ancestry and 222 ancestry-matched controls. Our results support the involvement of rare TREs in schizophrenia etiology.

Fig. 1. Study overview.

This flowchart summarizes the study design and analytic workflow.

Fig. 2. Genome-wide burden of rare TREs in schizophrenia.

The error bars denote 95% confidence interval.

Fig. 3. Network of genes with rare TREs that are differentially expressed in schizophrenia.

Interactions between genes were extracted from GeneMANIA. Top 50 scoring connected genes were pulled from pathways or physical interactions from the network data with automatic weighting. Node color: gray for known schizophrenia genes (defined in Methods), red for eQTL genes with rare TREs that are differentially expressed in schizophrenia (CMC genes with eQTLs and rare TREs); and blue for additional connected genes (top 50 genes) pulled from GeneMANIA. Node size is proportional to the strength of predictions for a given gene function.

Fig. 4. Constraint scores for genes with and without rare TREs.

Constraint scores (extracted from gnomAD’s observed/expected (o/e) upper bound LOEUF values) of genes with rare TREs are compared against those of the other genes without rare TREs in schizophrenia. A Only genes differentially expressed between schizophrenia and controls as determined by the Common Mind Consortium are compared, B only genes with eQTL are compared and (C) comparison is done for all protein coding genes. Box plots show Q1-1.5×IQR, Q1, median, Q3 and Q3 + 1.5×IQR. P-values reported were calculated from one-sided Wilcoxon rank-sum test assuming lower gnomAD o/e upperbound in genes with rare TREs. In all three categories assessed, the genes with rare TREs found in schizophrenia were more constrained (i.e., had on average lower LOEUF values) than the genes without rare TREs found in schizophrenia.

Fig. 5. Confirmation of EHdn detected tandem repeats.

A, B, Images of the gel electrophoresis showing bands that correspond to the expanded alleles in schizophrenia (SCZ) cases and the unexpanded allele in the reference sample NA12878. A 100 bp ladder is shown for size reference.