Phenotypic and functional analysis of SHANK3 stop mutations identified in individuals with ASD and/or ID

Abstract

Background: SHANK proteins are crucial for the formation and plasticity of excitatory synapses. Although mutations in all three SHANK genes are associated with autism spectrum disorder (ASD), SHANK3 appears to be the major ASD gene with a prevalence of approximately 0.5% for SHANK3 mutations in ASD, with higher rates in individuals with ASD and intellectual disability (ID). Interestingly, the most relevant mutations are typically de novo and often are frameshift or nonsense mutations resulting in a premature stop and a truncation of SHANK3 protein.

Methods: We analyzed three different SHANK3 stop mutations that we identified in individuals with ASD and/or ID, one novel (c.5008A > T) and two that we recently described (c.1527G > A, c.2497delG). The mutations were inserted into the human SHANK3a sequence and analyzed for effects on subcellular localization and neuronal morphology when overexpressed in rat primary hippocampal neurons.

Results: Clinically, all three individuals harboring these mutations had global developmental delays and ID. In our in vitro assay, c.1527G > A and c.2497delG both result in proteins that lack most of the SHANK3a C-terminus and accumulate in the nucleus of transfected cells. Cells expressing these mutants exhibit converging morphological phenotypes including reduced complexity of the dendritic tree, less spines, and less excitatory, but not inhibitory synapses. In contrast, the truncated protein based on c.5008A > T, which lacks only a short part of the sterile alpha motif (SAM) domain in the very SHANK3a C-terminus, does not accumulate in the nucleus and has minor effects on neuronal morphology.

Conclusions: In spite of the prevalence of SHANK3 disruptions in ASD and ID, only a few human mutations have been functionally characterized; here we characterize three additional mutations. Considering the transcriptional and functional complexity of SHANK3 in healthy neurons, we propose that any heterozygous stop mutation in SHANK3 will lead to a dysequilibrium of SHANK3 isoform expression and alterations in the stoichiometry of SHANK3 protein complexes, resulting in a distinct perturbation of neuronal morphology. This could explain why the clinical phenotype in all three individuals included in this study remains quite severe - regardless of whether there are disruptions in one or more SHANK3 interaction domains.

Keywords: ASD; Autism; Dendrite; Intellectual disability; Nucleus; SHANK3; Spine; Synapse.

Figure 1

Domain composition and expression analysis of SHANK3 variants. (A) Schematic overview of the SHANK3 constructs/fusion proteins used for overexpression experiments in this study. All constructs harbor an N-terminal GFP tag followed by a Myc-tag (indicated in green and orange, respectively). Domains are marked as it follows: Ank, ankyrin repeat domain; SH3, Src homology 3 domain; PDZ, postsynaptic density 95/discs large/zonula occludens-1 domain; Pro, proline-rich region; H, Homer binding site; C, cortactin binding site; and SAM, sterile alpha motif domain. SHANK3 is based on the full-length human SHANK3a sequence containing all of these domains. The truncated SHANK3 variants are identified as G1527A, 2497delG, and A5008T, and each is based on the SHANK3a sequence harboring either c.1527G > A, c.2497delG, or c.5008A > T, respectively. Predicted premature stop sites are marked by STOP in red. (B) Functional validation of the SHANK3 constructs in HEK293T cells via detection of the overexpressed fusion proteins SHANK3, G1527A, 2497delG, and A5008T by Western blot analysis using either anti-GFP (left panel) or anti-Myc (right panel) antibodies. kDa, kilodalton.

Figure 2

Nuclear accumulation of truncated SHANK3 variants G1527A and 2497delG. (A-E) Subcellular distribution of empty vector-based GFP (Control) (A), full-length GFP-Myc-SHANK3 (SHANK3) (B), or the truncated GFP-Myc-SHANK3 variants G1527A (C), 2497delG (D), and A5008T (E), after transient overexpression (DIV11-14) in rat primary hippocampal neurons. All neurons shown were co-transfected with the DenMark construct [30] containing an mCherry sequence to demarcate dendrites and spines in red. They were further immunostained for VGLUT1 (not shown), and nuclei were visualized by DAPI. In each panel (A-E), one representative co-transfected neuron is depicted. The picture on the left is a merge of the GFP and DenMark signals, while the picture in the middle shows only the GFP signal. The upper insets on the right show a merge of GFP and DAPI signals or the DAPI signal alone, and the lower inset on the right shows a representative secondary dendrite as a merge of the GFP and DenMark signals. Note the strong overlap of both G1527A and 2497delG with the DAPI signal.

Figure 3

Overexpression of truncated SHANK3 variants results in distinct alterations of dendritic tree complexity. (A-D) Dendritic tree complexity of rat primary hippocampal neurons co-transfected with DenMark and either empty vector-based GFP (Control), full-length GFP-Myc-SHANK3 (SHANK3), or the truncated GFP-Myc-SHANK3 variants G1527A, 2497delG, and A5008T (DIV11-14). (A) Representative images of the DenMark signal in rat primary hippocampal neurons overexpressing either Control or SHANK3, as indicated. The panel on the right shows a Sholl analysis of Control (white-filled circles) vs. SHANK3 (black-filled circles). (B-D) Representative images of the DenMark signal in rat primary hippocampal neurons overexpressing either G1527A (B), 2497delG (C), or A5008T (D), as indicated. The panels below each image show Sholl analyses either of Control (white-filled circles, upper panel) or SHANK3 (black-filled circles, lower panel) vs. the corresponding truncated SHANK3 variant (red-filled circles for G1527A in (B), green-filled circles for 2497delG in (C), and blue-filled circles for A5008T in (D)). *P < 0.05, **P < 0.01, ***P < 0.001 compared with either Control or SHANK3.

Figure 4

Overexpression of truncated SHANK3 variants results in distinct alterations of dendritic spines and synaptic contacts. Evaluation of dendritic spines (A-C) and synaptic contacts (D-F) in rat primary hippocampal neurons co-transfected with DenMark and either empty vector-based GFP (Control), full-length GFP-Myc-SHANK3 (SHANK3), or truncated GFP-Myc-SHANK3 variants (G1527A, 2497delG, A5008T) (DIV11-14). (A) Representative images of the DenMark signal in secondary dendrites of rat primary hippocampal neurons overexpressing either Control, SHANK3, G1527A, 2497delG, or A5008T as indicated. (B) Quantitative analysis of spine density. (C) Quantitative analysis of filopodia density. (D) Representative images of VGLUT1-positive (upper two rows) and VGAT-positive (lower two rows) presynaptic specializations among secondary dendrites of rat primary hippocampal neurons overexpressing either Control, SHANK3, G1527A, 2497delG, or A5008T, as indicated. (E) Quantitative analysis of VGLUT1-positive puncta density. (F) Quantitative analysis of VGAT-positive puncta density. White bars, Control; black bars, SHANK3; red bars, G1527A; green bars, 2497delG; and blue bars, A5008T. *P < 0.05 and ***P < 0.001 compared with Control; ^oo P < 0.01 and ^ooo P < 0.001 compared with SHANK3.