Synaptic-dependent developmental dysconnectivity in 22q11.2 deletion syndrome

Abstract

Chromosome 22q11.2 deletion increases the risk of neuropsychiatric disorders like autism and schizophrenia. Disruption of large-scale functional connectivity in 22q11 deletion syndrome (22q11DS) has been widely reported, but the biological factors driving these changes remain unclear. We used a cross-species design to uncover the developmental trajectory and neural underpinnings of brain dysconnectivity in 22q11DS. In LgDel mice, a model for 22q11DS, we found age-specific patterns of brain dysconnectivity, with widespread fMRI hyperconnectivity in juvenile mice reconfiguring to hippocampal hypoconnectivity over puberty. These changes correlated with developmental alterations in dendritic spine density, and both were transiently normalized by GSK3β inhibition, suggesting a synaptic origin for this phenomenon. Notably, analogous pubertal hyperconnectivity-to-hypoconnectivity reconfiguration occurs in human 22q11DS, affecting cortical regions enriched for GSK3β-associated synaptic genes and autism-relevant transcripts. This dysconnectivity also predicts age-dependent social alterations in 22q11DS individuals. These results suggest that synaptic mechanisms underlie developmental brain dysconnectivity in 22q11DS.

Fig. 1.. Developmental fMRI dysconnectivity in LgDel mice.

(A) Experimental timeline of fMRI mapping in wild-type (WT) and LgDel mice. (B) Whole-brain voxel-wise mapping of global fMRI connectivity revealed widespread hyperconnectivity in LgDel juvenile mice, reverting to focal hippocampal hypoconnectivity in the same animals after puberty. Maps are thresholded at |t| > 2.0, followed by family-wise error rate (FWER) correction at P < 0.05. Semitransparent right hippocampal blob indicates a cluster not surviving FWER correction (see fig. S1 for a comparison between uncorrected and corrected maps). Corresponding age × genotype maps and quantifications of fMRI global connectivity in dorsal hippocampal areas are reported for reference [two-way repeated-measures (RM) analysis of variance (ANOVA), age × genotype interaction, F = 10.78, P = 0.0021; WT, n = 22; and LgDel, n = 21]. (C and D) Seed-based mapping of the default-mode network (DMN) (C) and hippocampus (HPC) (D), respectively. Maps on the left show extension of reference DMN (C) and hippocampal (D) networks in juvenile WT mice. Corresponding between-group difference maps are reported for each of the probed developmental ages in the center panels. Red indicates increased fMRI connectivity, and blue indicates reduced fMRI connectivity compared to control WT littermates (|t| > 2.0, FWER corrected, P < 0.05). (E and F) Regional quantification of pairwise fMRI connectivity between regions of the DMN (E) and hippocampal networks (F) exhibiting significant age × genotype interaction (two-way RM ANOVA, DMN: F = 15.48, P = 0.0003, and F = 14.72, P = 0.0004; HPC: F = 8.67, P = 0.0053, and F = 7.36, P = 0.0097) as per maps in (C) and (D). Each point represents a mouse. **P < 0.01, and ***P < 0.001. BF, basal forebrain; CG, cingulate cortex; HY, hypothalamus; LS, lateral septum; MO, motor cortex; PFC, prefrontal cortex.

Fig. 2.. Developmental fMRI dysconnectivity in human 22q11DS.

(A) Schematic representation of human cohort divided into the childhood cohort (HC, n = 31; and 22q11DS, n = 21) and the postpubertal cohort (HC, n = 86; and 22q11DS, n = 118). Distribution of age for each diagnosis (HC and 22q11DS) across sites and scanners. (B) Voxel-wise (left) mapping of global fMRI connectivity revealed increased functional connectivity in 22q11DS carriers during childhood and reduced fMRI connectivity in the postpubertal cohort. Semitransparent maps in the background represent unthresholded t values. Clusters surviving FWER correction are outlined in black. Areas exhibiting a significant age × genotype interaction were identified using a linear model (two-way ANOVA, age × genotype interaction, F = 7.87, P = 0.0054). Whole-brain distribution of t values resulting from group differences at each age revealed a robust shift (arrow) from prevalent hyperconnectivity in childhood to prevalent hypoconnectivity after puberty (bottom right). (C and D) Seed-based analysis using clusters exhibiting significant age × genotype interaction in our global fMRI connectivity analysis as seeds: OPC (two-way ANOVA, F = 16.72, P < 0.0001) (C) and HPC (two-way ANOVA, F = 12.36, P = 0.0005) (D). Distribution of t values resulting from between-group connectivity differences over development are also reported for reference. Note shift from hyper- to hypoconnectivity (arrow) occurring over puberty. **P < 0.01, and ***P < 0.001. SFG, superior frontal gyrus.

Fig. 3.. Developmental fMRI dysconnectivity is paralleled by GSK3β-dependent alterations in dendritic spine density.

(A) Dendritic spine density measurements across development in the PFC and the HPC revealed a significant age × genotype interaction (two-way ANOVA, age × genotype interaction; PFC: F = 19.86, P = 0.0005; HPC: F = 58.28, P < 0.0001). (B) Voxel-wise global fMRI connectivity in these two areas exhibited a similar developmental trajectory, with a significant age × genotype interaction both in PFC and HPC (two-way RM ANOVA, age × genotype interaction; PFC: F = 8.8, P = 0.005; and HPC: F = 10.42, P = 0.0025). Each point represents a different mouse. (C) Intergroup differences in synaptic density and fMRI global connectivity show a linear relationship [coefficient of determination (R²) = 0.68; P = 0.01]. Color of the dots represents group [red, juvenile vehicle (VEH); orange, juvenile SB treated; dark blue, adult VEH; green, adult SB treated], while shape represents region (circle, HPC; diamond, PFC). This plot includes data from (C) and fig. S10. (D) Experimental timeline of the GSK3β inhibition treatment protocol (from PND7 to PND27), followed by fMRI and spine density measurements. (E) Developmental GSK3β inhibition rescued spine density increase in PFC of juvenile mice (two-way ANOVA, Sidak’s multiple comparisons test, LgDel SB versus LgDel VEH, P = 0.01). Similarly, the same treatment restored global fMRI connectivity alterations in juvenile LgDel mice (two-way ANOVA, Sidak’s multiple comparisons test, LgDel SB versus LgDel VEH, P = 0.0009). Maps are thresholded at |t| > 2.0, followed by FWER correction at P < 0.05. Errors bars represents SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 4.. Gene decoding supports involvement of synaptic mechanisms in 22q11DS developmental dysconnectivity.

(A) Illustration of gene decoding and gene enrichment analyses used to investigate molecular mechanisms underlying developmental reconfiguration. The term “decoded genes” refers to genes that were spatially enriched (i.e., genes that displayed significantly higher expression) in areas that undergo hyperconnectivity-to-hypoconnectivity reconfiguration in 22q11DS. (B) Decoded genes are specifically and significantly enriched for synaptic-related gene ontology (GO) gene lists, corroborating the involvement of synaptic mechanisms in 22q11DS dysconnectivity. Color scale indicates OR, while size of the dots represents −log₁₀(q value). Only visible dots were statistically significant at q < 0.05. Black outlines around dots represent observed OR above 95th percentile of null distribution. cAMP, cyclic adenosine 3′,5′-monophosphate; GPCR, G protein–coupled receptor. (C) Schematic representation of gene enrichment analyses (left) and results (right) showing significant enrichment for synaptic interactors of GSK3β (hypergeometric test, OR = 2.91, q = 0.0002, overlapping genes, N = 45). Bar color indicates OR, length represents –log₁₀(q value). Horizontal dashed line (in gray) represents significance at q < 0.05. ***P < 0.001. n.s., not significant.

Fig. 5.. Developmental fMRI dysconnectivity within somatomotor areas correlates with reciprocal social behavior in 22q11DS.

(A) Gene enrichment analyses showing highly significant enrichment between decoded genes and schizophrenia- and autism spectrum disorder (ASD)–related genes. Bar color indicates OR, and length represents –log₁₀(q value). Vertical dashed line (in gray) represents significance at q < 0.05. (B) Voxel-wise (left) and region of interest–based quantifications of relationship between opercular cortex connectivity and SRS score in 22q11 deletion carriers. Map represents areas of significant correlation between the age × connectivity interaction and SRS score (|t| > 2.0, FWER cluster corrected). Visualization of relationship between opercular cortex to postcentral gyrus (PCG) connectivity and SRS within the two age groups separately (right).