Copy number variations (CNVs), which are deletions or duplications of DNA segments, are a significant source of genetic variation linked to various neurodevelopmental and psychiatric conditions. While specific recurrent CNVs have been associated with autism spectrum disorder (ASD), schizophrenia (SZ), and attention-deficit hyperactivity disorder (ADHD), the mechanisms underlying their contribution to these disorders remain unclear. The 16p11.2 and 22q11.2 CNVs are particularly prevalent and impactful, altering the dosage of numerous genes and increasing the risk for ASD, SZ, and ADHD. These CNVs often exhibit a "mirror effect," where deletions and duplications affect neuroimaging measures in opposite directions. This suggests that understanding their impact on brain function could reveal key mechanisms underlying these complex disorders. Functional connectivity (FC), representing the synchronized activity between brain regions, measured through resting-state fMRI (rs-fMRI), offers a valuable tool for investigating these mechanisms. Previous studies have shown that rs-fMRI patterns correlate with gene expression patterns in the adult brain. However, few studies have examined the effects of 'neuropsychiatric' CNVs on FC comprehensively. This study aims to characterize the FC signatures of four high-risk CNVs (16p11.2 and 22q11.2 deletions and duplications), explore whether these signatures reflect dimensions observed in idiopathic ASD, SZ, or ADHD, and investigate the relationship between deletions at the FC and gene expression levels. The hypothesis is that seemingly distinct genetic variants and idiopathic disorders share overlapping patterns of dysconnectivity, which can reveal important FC dimensions involved in psychiatric conditions. This integrated approach, using both a top-down (clinical diagnosis-driven) and bottom-up (genetic variant-driven) strategy, is expected to offer new insights into the complex interplay between genetics, brain connectivity, and neuropsychiatric phenotypes.

Existing literature indicates a strong link between CNVs and neuropsychiatric disorders, particularly ASD, SZ, and ADHD. Studies have demonstrated the association of specific recurrent CNVs with these conditions, but the underlying mechanisms are not well understood. The 16p11.2 and 22q11.2 CNVs, due to their high frequency and impact on numerous genes, serve as significant loci for studying these mechanisms. Prior research has shown that these CNVs often manifest a "mirror effect," influencing neuroimaging and behavioral phenotypes in opposing directions depending on whether a deletion or duplication is present. The study of functional connectivity (FC) using resting-state fMRI (rs-fMRI) has emerged as a powerful tool for investigating brain network alterations in these disorders. Several studies have examined FC alterations in specific regions of interest (ROIs) in CNV carriers, but connectome-wide association studies (CWAS), which analyze all connections without prior hypotheses, have been limited. Additionally, the relationship between CNV-induced FC alterations and idiopathic conditions has not been explored systematically. Existing literature on idiopathic conditions reveals patterns of widespread under-connectivity in ASD and SZ, although some studies suggest over-connectivity in cortico-subcortical regions, particularly involving the thalamus. While these alterations are not completely disorder-specific, they suggest the existence of shared underlying mechanisms and perhaps continuous dimensions of dysconnectivity.

This study employed a connectome-wide association study (CWAS) design using rs-fMRI data from five datasets. The datasets included: (1-2) Two genetic-first cohorts (Simons Variation in Individuals Project (VIP) for 16p11.2 and a UCLA cohort for 22q11.2 CNVs and their controls); (3) Individuals with ASD and their controls from the ABIDE dataset; (4) Individuals with SZ and controls from 10 studies; and (5) Individuals with ADHD and controls from the ADHD-200 dataset. A total of 1928 individuals were included after rigorous preprocessing and quality control. Connectomes were constructed by segmenting the brain into 64 functional seed regions using the multi-resolution MIST brain parcellation. Functional connectivity was computed as the temporal pairwise Pearson's correlation between the average time series of these regions, followed by Fisher transformation. Seven CWAS were performed, comparing FC between cases and controls for four CNVs (16p11.2 and 22q11.2 deletions and duplications) and three idiopathic psychiatric cohorts (ASD, SZ, ADHD). Linear regression was used at the connectome level, adjusting for sex, head motion, and age. Significance was determined by testing whether the β value was different from 0 using a two-tailed t-test, with Benjamini-Hochberg FDR correction (q < 0.05). Global FC shift was calculated as the average β value across all connections. The mirror effect of deletions and duplications was tested by computing the product of β values for each CNV type. Spatial similarity between whole-brain FC signatures across CNVs and idiopathic conditions was assessed using Pearson's correlation, comparing cases and controls using Mann-Whitney U tests with FDR correction. Similar analyses were conducted at the regional level (64 regions). Finally, the relationship between gene expression patterns from the Allen Human Brain Atlas (AHBA) and FC signatures of the deletions was investigated using partial least squares regression (PLSR) and individual gene correlations, with 5000 label-shuffled null FC signatures used to assess significance. Analyses were performed using Python's scikit-learn library and visualized in Python and R.

The 16p11.2 deletion showed a global increase in FC (mean shift = 0.29 z-scores, p = 0.048), with overconnectivity primarily involving frontoparietal, somatomotor, ventral attention, and basal ganglia networks. The 22q11.2 deletion showed a non-significant global decrease in connectivity, with underconnectivity affecting the anterior and lateral default mode network (DMN) and limbic network. Deletions and duplications at both loci showed a mirror effect at the global connectivity level. The effect sizes of connectivity alterations in CNVs were approximately two-fold larger than those observed in idiopathic SZ, ASD, and ADHD. Significant whole-brain FC similarities were observed between idiopathic ASD, SZ, and the four CNVs, but not with ADHD. Regional FC similarity analysis revealed that individuals with SZ and ASD showed higher similarity with the thalamus FC-signatures of both deletions. Higher individual similarity with deletion FC-signatures was associated with more severe symptoms in ASD (ADOS and FSIQ scores). Despite opposing global effects, 16p11.2 and 22q11.2 deletions showed positive correlation in their connectomes, primarily in frontoparietal, ventral attentional, and somatomotor networks. PLSR analysis revealed an association between the FC signatures of both deletions and expression patterns of genes within and outside the studied loci, indicating a lack of high specificity. Genes with correlations exceeding the genome-wide 98th percentile were over-represented in both deletions.

This study provides the first connectome-wide characterization of four CNVs associated with psychiatric disorders. The mirror effects observed in deletions and duplications at both loci highlight the importance of gene dosage in shaping brain connectivity. The findings of shared dysconnectivity patterns across CNVs and idiopathic ASD and SZ, particularly involving the thalamus, somatomotor, and posterior insula regions, support the idea of shared mechanistic building blocks across these conditions. The lack of similar findings in ADHD may reflect differences in the genetic architecture and phenotypic expression of this disorder. The association between FC-signatures and symptom severity further strengthens the link between genetic variants and clinical manifestations. The observed redundancy in the association between gene expression patterns and FC signatures across both CNV loci, and even genome-wide, may explain the similarities in neuropsychiatric symptoms affected by many CNVs.

This study demonstrates that rare neuropsychiatric mutations converge on shared dimensions of brain functional connectivity, particularly involving thalamo-sensory pathways. The mirror effects of deletions and duplications and the association of connectivity alterations with symptom severity underscore the importance of gene dosage and highlight potential therapeutic targets. Future research should focus on investigating the genetic determinants of thalamo-sensory disturbances and integrating top-down and bottom-up approaches across a broader range of genetic variants and diagnostic boundaries to unravel the complex etiology of neuropsychiatric disorders.

Several limitations should be considered. The reproducibility of rs-fMRI in psychiatry is a challenge, although the current study's findings align with many previous studies using similar methodologies. The absence of genetic data for idiopathic cohorts restricts a complete understanding of the genetic contributions. The AHBA expression data, derived from only six adult brains, may not fully capture the developmental trajectory of gene expression relevant to these disorders. The limited power to detect effects in duplications warrants further investigation. Confounding factors such as sex bias, age differences, medication status, and comorbid diagnoses may have influenced the results, although sensitivity analyses were conducted to address these potential confounders.