Autism-linked Cullin3 germline haploinsufficiency impacts cytoskeletal dynamics and cortical neurogenesis through RhoA signaling

Abstract

E3-ubiquitin ligase Cullin3 (Cul3) is a high confidence risk gene for autism spectrum disorder (ASD) and developmental delay (DD). To investigate how Cul3 mutations impact brain development, we generated a haploinsufficient Cul3 mouse model using CRISPR/Cas9 genome engineering. Cul3 mutant mice exhibited social and cognitive deficits and hyperactive behavior. Brain MRI found decreased volume of cortical regions and changes in many other brain regions of Cul3 mutant mice starting from early postnatal development. Spatiotemporal transcriptomic and proteomic profiling of embryonic, early postnatal and adult brain implicated neurogenesis and cytoskeletal defects as key drivers of Cul3 functional impact. Specifically, dendritic growth, filamentous actin puncta, and spontaneous network activity were reduced in Cul3 mutant mice. Inhibition of small GTPase RhoA, a molecular substrate of Cul3 ligase, rescued dendrite length and network activity phenotypes. Our study identified defects in neuronal cytoskeleton and Rho signaling as the primary targets of Cul3 mutation during brain development.

Fig. 1. Generation and characterization of Cul3^+/− mouse model.

a Loss-of-function (red font) and missense (black font) mutations in Cul3 gene identified in the patients with neurodevelopmental disorders. ASD autism spectrum disorder, SCZ schizophrenia, DD developmental delay. The location of E246X patient’s mutation from this study is highlighted in the magenta-colored box. b Schematic representation of mouse line generation using CRISPR/Cas9 genome editing. cDNA sequence trace showing 1-bp insertion (purple) in exon 6 is shown. Single guide (Sg) RNA is shown in blue, PAM site is shown in red. Sanger sequencing diagram showing 1 bp insertion is at the bottom. c qRT-PCR showing reduction of mRNA expression of Cul3 in adult cerebral cortex of Cul3^+/− mice (***p < 0.001; two-tailed t-test; n = 5 for both genotypes); Error bars represents mean ± SEM. d Western blot of Cul3 protein, showing its significant reduction in Cul3^+/− mutant mice cortex (p < 0.01; n = 6 for each genotype); hippocampus (p < 0.001; n = 6 for each genotype) and cerebellum (p < 0.001; n = 6 for each genotype); error bars represents mean ± SD. e (Top left panel) Representative image of embryonic day E17.5 WT and Cul3^+/− embryo demonstrating smaller size; (bottom left panel) WT and Cul3^+/− adult brain; (right panel) representative image of adult (8-weeks old) WT and Cul3^+/− males demonstrating smaller body size of Cul3^+/− mutants. f Both adult male and female Cul3^+/− mice have reduced body weight as compared to their WT littermates (***p < 0.001; **p < 0.01; two-tailed t-test) (WT n = 11 male/12 female; Cul3^+/− n = 21 male/14 female); ○-female and ∆-male; Error bars represents mean ± SEM; two-tailed t-test was used for calculating statistical significance for (c–f). Dots represent individual animals.

Fig. 2. Cul3^+/− mice have altered brain morphology.

a Voxel-wise analysis highlighting significant differences in relative volumes throughout the brain between WT and Cul3^+/− mice with 5% false discovery rate (FDR). Left panel is absolute changes and right panel is relative change in brain regions. Scale bar 2.7–10.7 indicates decreasing FDR, where 2.7 = 5% FDR. Red color signifies increased and blue color signifies decreased brain volume compared with WT brain. b Cul3^+/− mice have reduced absolute brain volume (**p < 0.01) compared to WT mice (WT n = 18, Cul3^+/− n = 20). c Reduced relative grey matter volume in Cul3^+/− mouse brain (*p < 0.05). d No significance difference is observed in relative white matter volume. e MRI revealed significant reduction in relative volume, normalized by total brain volume, of primary somatosensory cortex (*p < 0.05); hippocampal DG region (***p < 0.001); cerebellum (**p < 0.01). Dots represent individual animals; ○-female and ∆-male. f Reduced cortical thickness is observed by MRI in Cul3^+/− mice, turquoise color is decreased, red/orange color is increased. g Reduced cortical thickness of somatosensory cortex in adult Cul3^+/− mice, left panel shows somatosensory cortex region stained with NeuN (mature neuron marker) and BRN2 (layer II–IV marker); Scale bar is 100 μm. h Reduction of total cortical thickness, layer thickness and density of NeuN- and BRN2-positive cells/area in layer II–IV are observed (*p < 0.05; two-tailed t-test, n = 6 for each genotype). Dots represent independent samples; two-tailed t-test used for (b–h); error bars represent mean ± SD.

Fig. 3. Cul3^+/− mice display hyperactivity, cognitive, and social impairments.

a Cul3^+/− mice travel longer distances in open field (n = 27 WT; n = 29 Cul3^+/− (*p < 0.05)) compared with WT mice. b The traveling speed (cm/s) is significantly increased in Cul3^+/− mice (*p < 0.05); representative traces of 30 min in open field show Cul3^+/− mice traveling longer distances compared to WT mice. c Time bins showing that Cul3^+/− mice travel significantly longer distance in the last 20 min of the test (*p < 0.05; **p < 0.01). d No difference in time spent self-grooming between WT and Cul3^+/− mice (WT n = 15; Cul3^+/− n = 19). e Cul3^+/− mice demonstrate significantly reduced preference for novel object in novel object recognition test (*p < 0.05; two-tailed t-test, WT n = 18; Cul3^+/− n = 20). f (Upper panel) Schematic diagram of three-chamber social interaction test; reduced sniffing time is observed for Cul3^+/− mice while interacting with a novel mouse (vs novel object) as compared to WT mice (WT n = 17; Cul3^+/− n = 19; *p < 0.05, One-way ANOVA). g Reduced sniffing time is observed for Cul3^+/− mice while interacting with a novel mouse (vs familiar mouse) as compared to WT mice (WT n = 17; Cul3^+/− n = 19); ***p < 0.01; One-way ANOVA. Dots represent individual animals; ○-female and ∆-male; error bars represent mean ± SD.

Fig. 4. Spatiotemporal differential gene expression analyses identifies dysregulation of cytoskeletal processes by Cul3 mutation.

a–c Differential gene expression analyses of cortical samples from embryonic (a), early postnatal (b), and adult (c) developmental periods. (Left panels) Volcano plots of differentially expressed genes in Cul3^+/− vs WT. Genes colored in red are upregulated in Cul3^+/− compared to WT; genes colored in blue are downregulated in Cul3^+/− compared to WT; Cul3 is colored in pink. (Right panels) GO-terms enrichment of differentially expressed genes. Contribution of up- or downregulated genes to specific GO terms are shown in blue and red, respectively. d  Heatmap of enriched common GO terms at different developmental time periods. Fisher P-value combination analysis (Materials and methods) was applied to identify 21 clusters of differentially expressed up- and downregulated genes impacted by Cul3 mutation across developmental periods. Biological processes impacted in two or more clusters are shown (Individual cluster details are present in Supplementary Table S6). e  Enrichment of differentially expressed genes from each period and region with literature-curated gene lists with previous evidence for involvement in autism. These lists include pre- and post-synaptic genes from SynaptomeDB; syndromic and highly ranked (1 and 2) genes from SFARI Gene database (https://gene.sfari.org/database/gene-scoring/); genes with probability of loss-of-function intolerance (pLI) > 0.99 as reported by the Exome Aggregation Consortium; constrained genes; and FMRP target genes. Number of overlapped genes (in parenthesis) and odds ratio are indicated inside each cell, and provided only for FDR < 0.05 and OR > 1.

Fig. 5. Differential protein expression and weighted protein co-expression network analyses of Cul3^+/− mice.

a (Left panel) Volcano plot of differentially expressed proteins between Cul3^+/− and WT embryonic cortex identified from quantitative TMT-MS proteomic profiling. Cul3 is colored in pink, proteins colored in red are upregulated and proteins colored in blue are downregulated in Cul3^+/−embryonic cortex. (Middle panel) Gene ontology enrichment analyses of up- and downregulated proteins are shown as bar plots. Contribution of up- or downregulated proteins to specific GO terms are shown in blue and red, respectively. (Right panel) Expression Heatmap of proteins associated with actin, microtubule, and intermediate filament cytoskeleton in WT and Cul3^+/− embryonic cortex. b Enrichment of differentially expressed proteins (combined by period and region) with literature-curated gene lists with previous evidence for involvement in autism. Number of overlapped proteins (in parenthesis) and odds ratio are shown. c Hierarchical clustering of protein co-expression modules by module eigengene for Cul3^+/− embryonic cortex. Module-genotype associations (* FDR < 0.1) for each module are shown below dendrogram. A total of 9 modules were significantly associated with Cul3^+/− genotype in embryonic cortex. Module enrichment analyses against literature-curated gene lists with previous evidence for involvement in autism are shown at the bottom (* FDR < 0.05) inside each cell, and provided only for FDR < 0.05 and OR > 1. Row annotations are the same as in panel (b) starting from presynaptic proteins in row one and ending with FMRP targets in row six. d Cell type enrichment of co-expression modules from embryonic cortex using P0 mouse cortex scRNA-seq dataset [24]. Modules significantly enriched in at least one cell type are shown for embryonic and early postnatal time periods. e PC1 of representative modules with significant cell type enrichment plotted by genotype. Embryonic module 2blue and early postnatal module 11greenyellow from Cul3^+/− are depleted in excitatory neurons. All comparisons between WT and Cul3^+/− are significant using t-test statistics. f GO terms for 2blue and 11greenyellow modules were obtained using g:Profiler [57]. g Transcriptome vs proteome correlation for cortex. DEGs for cortex from all developmental periods were overlapped with proteomes to extract corresponding proteins.  The Pearson’s correlation coefficient for DEGs and corresponding proteins expression fold changes from cortex is shown, each dot is one gene/protein. h GO term enrichment analysis of correlated genes/proteins for cortex, highlighting cytoskeletal and neurogenesis functions.

Fig. 6. Cul3 haploinsufficiency leads to altered neuron growth, network activity, and to RhoA upregulation in Cul3^+/− mice.

a Representative images of 14DIV primary cortical neurons from WT and Cul3^+/− mice, immunostained with MAP2 (upper panel), and tracings by simple neurite tracer (lower panel). Scale bar is 25 µm. b–d Quantification of total dendrite length, neurite number, and soma size are shown. Symbols represent independent neurons and color represents littermates. Data are shown as mean ± SD (n = 6 per genotype, at least 6–8 neurons per mouse). Significance is calculated using two-tailed t-test; ***p < 0.001; **p < 0.01. e Representative raster plots of spontaneous spike activity from 8DIV primary cortical neurons. f Spontaneous spike activity is significantly reduced in Cul3^+/− cortical neurons; **p < 0.01, two-tailed t-test. g Average burst frequency is significantly reduced in Cul3^+/− neurons; **p < 0.01; two-tailed t-test, n = 6–7 mice per genotype; 1.5 × 10⁶ neurons were seeded from each mouse in each MEA plate well, each containing 64 electrodes. Each dot represents independent mouse and color represents littermates. h Representative images of western blot analysis of Cul3, total RhoA, and GAPDH loading control in embryonic and adult cortices. Densitometry analysis of western blot is shown at the bottom. Data are presented as mean ± SEM (n = 6 per genotype). i Representative images of western blot analysis of active RhoA (RhoA-GTP) pulldown, total RhoA and GAPDH loading control from input lysate of embryonic cortex. Densitometry analysis of western blot is shown at the bottom. Data are presented as mean ± SEM (n = 4 per genotype for total RhoA, n = 4 per genotype for active RhoA). Significance is calculated using two-tailed t-test; **p < 0.01. The significance above bars represents comparison against WT.

Fig. 7. Actin cytoskeleton defects are observed in Cul3^+/− mice.

a, b Module-genotype association and GO functional annotations for black actin module identified by protein co-expression analyses of adult cortex. (n = 4 per genotype). c Representative images of 21DIV primary cortical neurons from WT and Cul3^+/− mice, immunostained with MAP2 (green) and phalloidin-rhodamine (red); Scale bar is 25 μm; yellow triangles point to F-actin puncta on dendritic segments for better visualization. d, e Quantification of F-actin puncta and F-actin intensity on MAP2 positive dendrites, normalized by dendrite length. Symbols represent independent neurons and color represents littermates. Data are presented as mean ± SD (n = 4 per genotype, at least 7–10 neurons per mouse). Significance is calculated using two-tailed t-test; ***p < 0.001.

Fig. 8. RhoA inhibition rescues dendritic growth and network deficits.

a Flow diagram showing treatment timeline of primary cortical neurons with RhoA inhibitor Rhosin (RH). Half of the media containing Rhosin or vehicle was replaced every third day until the day of experiments. b Representative images of 14DIV primary cortical neurons (left panel) and tracings (right panel); Scale bar is 25 μm. The vehicle (Vh) and Rhosin (Rh) treated cells were immunostained with Map2 and dendrite tracing was performed. c Rhosin treatment rescues decreased dendrite length phenotype in Cul3^+/− neurons. Symbols represent independent neurons and color represents littermates. Data are presented as mean ± SD (n = 2 per genotype, at least 8–10 neurons per mouse). Significance is calculated using One-way ANOVA and Tukey test for multiple comparison; *p < 0.05. d Representative raster plots of spontaneous spike activity from 8DIV primary cortical neurons. e Spontaneous spike activity is significantly reduced in Cul3^+/− cortical neurons and rescued by treatment with 10 µM Rhosin; *p < 0.05, One way ANOVA and Tukey test for multiple comparison. f Average burst frequency is significantly reduced in Cul3^+/− neurons and rescued by treatment with 10 µM Rhosin; n = 6–9 mice per genotype, **p < 0.01; *p < 0.05, One-way ANOVA and Tukey test for multiple comparison was used.