Background
Autism spectrum disorder (ASD) is a neuropsychiatric condition manifesting in early development and is characterized by two core features: A) persistent deficits in social interaction and communication and B) the presence of restricted interests and/or repetitive behaviors [
1]. The strong involvement of genetics in the development of ASD is supported by the identification of causative genetic abnormalities in more than 20% of cases, with a significant number of the identified genes encoding proteins required for the correct formation, maturation, and maintenance of synaptic connections in the brain [
2-
9]. Among these, the
SHANK gene family plays a decisive role because diverse genetic variation in
SHANK1,
SHANK2, and
SHANK3 - all encoding large postsynaptic scaffold proteins - has been identified in individuals with ASD [
10-
16]. A crucial role of
SHANK3 mutations in this context is supported by the following three facts: 1)
SHANK3 haploinsufficiency is the critical factor for the development of neuropsychiatric symptoms in 22q13 deletion syndrome, also known as Phelan-McDermid syndrome, 2) the current prevalence for
SHANK3 mutations in individuals with ASD in general is between 0.5% and 0.7%, and 3) data indicate that a
SHANK3 mutation is present in approximately 2% of individuals with both ASD and intellectual disability (ID) [
16-
18].
Some individuals diagnosed with either ASD, ID, or both harbor frameshift or nonsense mutations in
SHANK3 resulting in a premature stop codon and causing a truncation of SHANK3 protein [
15-
17,
19-
22]. However, only a few studies have thus far addressed the impact of such mutations and their corresponding truncated proteins on neuronal function and morphology [
15,
19,
22-
25]. In this context, the
de novo exon 21 frameshift mutation c.3679_3680insG - identified in two brothers diagnosed with both ASD and ID [
15] - and the
de novo exon 21 nonsense mutation c.3349C > T - identified in three brothers, all of them diagnosed with ID, with two having an additional diagnosis of schizophrenia (SCZ) [
19] - have been most intensely studied up to date [
15,
19,
23-
25]. Insertion of either stop mutation into the rat
Shank3a sequence at the corresponding sites results in the expression of truncated Shank3a variants lacking distinct parts of the C-terminus, a region crucial for appropriate synaptic targeting and assembly [
26-
29]. In contrast to wild-type Shank3a, Shank3a harboring either c.3679_3680insG or c.3349C > T mutations no more cluster at synapses, but rather distribute in the somatodendritic compartment and localize to the nucleus when overexpressed in primary hippocampal neurons [
15,
19,
23-
25]. Overexpression of Shank3a harboring the c.3679_3680insG mutation affects growth cone mobility and negatively interferes with synaptic transmission and transsynaptic signaling; the same mutation leads to reductions in the number of excitatory synapses and dendritic spines [
15,
23,
24]. Shank3a harboring the c.3349C > T mutation impairs the ability of Shank3a to promote the outgrowth of primary neurites, results in a less complex dendritic arbor, and leads to a specific reduction of excitatory, but not inhibitory synapses [
19,
25].
In vitro examination of a
de novo exon 21 nonsense mutation c.2997C > G identified in a boy with ID demonstrated a reduction in neurite nodes, tips, and length, at early stages of neuronal differentiation [
22]. Taken together, these
in vitro studies show that truncations of the distal C-terminus of Shank3a, caused by the c.3679_3680insG, c.3349C > T or c.2997C > G mutations are sufficient to disrupt neuronal morphology when the truncated variant is overexpressed in primary neuronal cultures.
However, the three stop mutations studied to date all affect exon 21 of
SHANK3, but less is known about mutations identified in ASD and/or ID that affect other parts of the gene [
15-
17,
19-
22]. It is therefore of high interest to evaluate additional mutations, including those that disrupt other exons of
SHANK3 and to identify converging and/or distinct neuronal pathologies.
We inserted three mutations identified in subjects with ASD and/or ID into the human
SHANK3a sequence: a nonsense mutation affecting exon 12 (c.1527G > A), a frameshift mutation affecting exon 21 (c.2497delG) - both recently described [
17] - and one novel nonsense mutation affecting exon 22 (c.5008A > T). Clinical assessment of the corresponding subjects was followed by characterization of the impact of the three truncated SHANK3a variants with respect to subcellular localization, dendritic branching, and spine and synapse formation, when overexpressed in rat primary hippocampal neurons with a wild-type Shank3 background.
Methods
The Institutional Review Board (IRB) of Icahn School of Medicine at Mount Sinai approved all studies involving humans, and all subjects were recruited under an IRB approved protocol as part of ongoing studies in Phelan-McDermid syndrome at the Seaver Autism Center for Research and Treatment at the Icahn School of Medicine at Mount Sinai with parents providing informed consent. An inter-disciplinary evaluation team conducted comprehensive assessments using the following clinical evaluation tools: (1) Clinical genetic evaluations and dysmorphology exams; (2) Neurological examination to evaluate gross motor skills and gait, fine motor coordination, cranial nerves, and deep tendon reflexes; (3) ASD focused diagnostic evaluation using the Diagnostic and Statistical Manual for Mental Disorders-IV (DSM-IV), the Autism Diagnostic Observation Schedule-2 (ADOS-2), and the Autism Diagnostic Interview Revised (ADI-R); (4) Cognitive testing using the Mullen Scales of Early Learning; (5) The Vineland Adaptive Behavior Scales-II, Survey Edition, to evaluate independence in daily life skills, including communication, socialization, and motor skills; (6) Medical record review including analyzing any results from electroencephalographic and brain imaging studies; and (7) Genetic testing for confirming mutation and de novo origin, using Sanger sequencing.
Vector constructs
Human
SHANK3a complementary DNA (cDNA) based on NP_277052.1 was re-designed in collaboration with GeneArt® (Life Technologies, Carlsbad, CA, USA) for optimized GC content, and a Myc-tag was added immediately after the initiation codon. Subcloning was performed using In-Fusion HD (Clontech Laboratories, Mountain View, CA, USA). The entire
SHANK3a cDNA was amplified using the following set of primers: 5′-GTCCGGACTCAGATCTATGGAGCAGAAGCTGATCAG-3′ and 5′-GTCGACTGCAGAATTCTCAGCTGCCGTCCAGCTGT-3′, and further inserted into the pAcGFP1-C1 (Clontech Laboratories, Mountain View, CA, USA) vector using Bgl2 and EcoR1 sites. The sequence was confirmed by Sanger sequencing. The c.1527G > A variant was generated by using the primer set 5′-GCTTCTGaGAGGGCACCGTGAAG-3′ and 5′-TGCCCTCtCAGAAGCCGCCctcg-3′. The c.2497delG variant was generated by inserting a fragment containing the deletion followed by authentic human
SHANK3 cDNA corresponding to the sequence from immediately after the variation to the predicated premature termination codon. The c.5008A > T variant was generated by using the primer set 5′-GTGGTCCtAGTTCGACGTGGGCGACTGG-3′ and 5′-CGAACTaGGACCACAGCTGCAGGGGTTT-3′. The eGFP-Shank3a and DenMark constructs have been described previously [
29,
30].
Antibodies
A novel polyclonal antibody directed against the rat Shank3a N-terminus (aa 333-470) was generated for this study according to the antibody production and purification protocol described in [
31]. The anti-Shank3 PRC antibody has been described previously [
31]. The following primary antibodies were purchased from commercial suppliers: anti-histone H3 (Cell Signaling Technology, Danvers, MA, USA), anti-green fluorescent protein (GFP) (Clontech, Laboratories, Mountain View, CA, USA), anti-c-Myc (Roche Applied Science, Mannheim, Germany), as well as anti-GAPDH, anti-VGLUT1, and anti-VGAT (all from Synaptic Systems, Goettingen, Germany).
Biochemistry
For whole culture extracts, transfected HEK293T cells were lysed in Triton X-100 Lysis Buffer (150 mM NaCl, 50 mM Tris HCl, 1% Triton X-100, pH 8,0, protease inhibitor mix, Roche Applied Science, Mannheim, Germany). The NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, Bonn, Germany) were further used to obtain nuclear and cytoplasmic fractions from transfected HEK293T cells. Protein concentrations were determined by Bradford protein assay, and the same amount of protein was loaded per lane for SDS-PAGE. Western blot analysis was conducted following standard protocols. HRP-conjugated secondary antibodies (Dako, Glostrup, Denmark) and the SuperSignal detection system (Thermo Scientific, Bonn, Germany) were used to visualize protein bands on X-ray films (GE Healthcare, Freiburg, Germany).
Animal experiments
All animal experiments in this study were performed based on the guidelines for the welfare of experimental animals issued by the Federal Government of Germany and by the local ethics committee (Ulm University), ID Number: 0.103.
Cell culture
HEK293T cells were maintained in DMEM at 37°C in 5% CO
2. The preparation of hippocampal cultures from rat was performed at embryonic stage 18 (E18) as described previously [
32]. In brief, hippocampal neurons were seeded on poly-l-lysine (0.1 mg/ml, Sigma-Aldrich, Steinheim, Germany)-coated glass coverslips. Cells were grown in neurobasal medium, complemented with B27 supplement, 0.5 mM
L-glutamine and penicillin/streptomycin at 100 U/ml (all reagents from Life Technologies, Darmstadt, Germany), and maintained at 37°C in 5% CO
2.
Immunocytochemistry
Immunocytochemistry was performed as described previously with minor modifications [
33]. Cultured cells were fixed with 4% paraformaldehyde (PFA)/1.5% sucrose in phosphate-buffered saline (PBS) at RT for 20 min and processed for immunocytochemistry. After permeabilization of the cells with 0.1% Triton X-100 in PBS for 5 min, blocking was performed using 5% FCS in PBS, followed by the primary antibody at 4°C overnight. Washing with PBS was followed by incubation with the secondary antibody coupled to Alexa Fluor® 488, 568, or 647 (all from Life Technologies, Darmstadt, Germany) for 1 h at room temperature. The actin cytoskeleton was visualized by Alexa Fluor® 647 Phalloidin in some experiments. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and after further washing steps, cells were mounted in mowiol. Images were captured using an upright fluorescence microscope (Axioskop 2, Zeiss, Oberkochen, Germany) and Axiovision software (Zeiss, Oberkochen, Germany).
Transfections
Vector constructs were transfected into HEK293T cells using PolyFect reagent (Qiagen, Hilden, Germany) as described previously [
34] or into hippocampal neurons using Lipofectamine 2000 reagent (Life Technologies, Darmstadt, Germany).
Analysis of neuronal morphology
All analyses were done in a blinded fashion. Sholl analysis was performed as described previously [
35]. Concentric circles (15, 30, 45, 60, 75, 90, 105, 120, 135, and 150 μm in diameter) were drawn around the soma of each neuron included in the analysis. The number of all dendrites crossing each circle was counted manually. For analysis of spines and filopodia, two secondary dendrites were randomly chosen per neuron and dendritic protrusions were counted manually among approximately 35-μm-long segments per dendrite. Dendritic protrusions shorter than 1 μm with clearly visible head and neck were counted as spines, and dendritic protrusions longer than 1 μm and devoid of head and neck were counted as filopodia. For analysis of synaptic contacts, four secondary dendrites were randomly chosen per neuron and signals positive for either VGLUT1 (excitatory contacts) or VGAT (inhibitory contacts) were manually counted among approximately 50-μm-long segments per dendrite.
Statistical analysis
For all analyses in primary culture, five to eight neurons from three independent experiments were analyzed per condition; ‘n’ therefore ranged between 15 and 22. GraphPad Prism 5.01 (GraphPad Software, La Jolla, CA, USA) was used for all statistical analyses. Depending on the datasets, analysis was performed with unpaired Student’s t-test or one-way ANOVA with Bonferroni post hoc test if data were normally distributed or with the Mann-Whitney U test or the Kruskal-Wallis test with Dunn’s multiple comparison post hoc test if data were not normally distributed.
Discussion
In this study, we have described, for the first time, a novel truncating stop mutation in an individual with the diagnosis of both ASD and ID (c.5008A > T) (Table
2). In addition,
in vitro analyses with this and two other recently identified truncating stop mutations (c.1527G > A and c.2497delG, [
17]) addressed the question if distinct mutations in
SHANK3 result in mutation-specific or converging morphological phenotypes.
Table 2
Genetic information on the three participants included into this study
P1 | M | ASD/ID | c.1527G > A | p.W509X |
De novo
| |
P2 | M | ID | c.2497delG | p.P834RfsX58 |
De novo
| |
P3 | M | ASD/ID | c.5008A > T | p.K1670X | Not from mother | This study |
Interestingly, we found that SHANK3a variants harboring either c.1527G > A or c.2497delG, thus lacking large but distinct parts of the protein’s C-terminus (Figure
1A), exhibit overlapping phenotypes in all morphological parameters investigated. When overexpressed, both truncated proteins exclusively accumulate in the nuclear compartment of transfected cells (Figures
2C,D, Additional file
5: Figure S3A,B) and this is accompanied by a severe reduction of dendritic tree complexity (Figure
3A,B,C). Moreover, we found a specific reduction of dendritic spine and excitatory, but not inhibitory synapse density (Figure
4A,B,C,D,E,F). These findings are in line with previous
in vitro studies showing similar phenotypes for rodent homologue Shank3a variants harboring either c.3679_3680insG or c.3349C > T [
23,
25].
In contrast, none of these phenotypes was observed in neurons overexpressing SHANK3a haboring the c.5008A > T mutation. However, some minor alterations in neuronal morphology were detected, and interestingly, they were opposite from the ones related to c.1527G > A and c.2497delG as c.5008A > T produced slightly enhanced complexity of the distal dendritic tree (Figure
3D).
With respect to subcellular localization of truncated SHANK3, an important aspect to consider is domain composition. The SHANK3a c.5008A > T variant is only lacking a small part of the protein’s C-terminus, thereby exclusively disrupting the SAM domain (Figure
1A), a domain needed for the correct assembly of Shanks in the PSD [
27]. In line with this, it still localizes to dendrites and excitatory synapses of transfected neurons in cluster-like structures, although with a much lower efficiency as compared to full-length SHANK3a (Figure
2E, Additional file
4: Figure S2A,B). The two other truncated SHANK3a variants analyzed here are lacking major parts of the C-terminus including distinct stretches of the proline-rich domain and - in all cases - the Homer binding site (a dendritic localization signal), a synaptic targeting element in between the Homer and cortactin binding sites and the SAM domain (Figure
1A). In line with this, each of these variants never forms cluster-like structures within dendrites or at spine synapses, but rather localizes to the nucleus of transfected neurons (Figure
2C,D).
A recent study on the transcriptional and functional complexity of the rodent
Shank3 gene convincingly delineates all rodent Shank3 isoforms (Shank3a-f) and reports distinct and/or overlapping phenotypes for each isoform with respect to subcellular distribution and neuronal morphology after overexpression in primary hippocampal neurons [
37]. Summarizing what was observed, selected Shank3 isoforms either increase the number of spines and excitatory synapses or show the opposite effect, likely depending on their domain composition and subcellular localization. Intriguingly, the Shank3 isoform Shank3b, which only contains the ankyrin repeats, the SH3 domain, and the PDZ domain [
37], resembles two of the shorter truncated SHANK3a variants described here. In primary hippocampal neurons, overexpressed Shank3b localizes to the nucleus accompanied by a reduced number of mature spines and excitatory synapses [
37] - just as we report in this study for overexpression of SHANK3a harboring either c.1527G > A or c.2497delG mutations (Figures
2,
4) and as others have reported in previous studies for the rodent Shank3a homologue harboring either c.3679_3680insG or c.3349C > T [
23,
25].
Although the exact function of each isoform is yet to be determined, it can already be hypothesized at this point that in a healthy neuron, the expression of SHANK3 isoforms is fluid, adapting to the ever-changing needs of the cell. Hence, we can speculate that in neurons from individuals with heterozygous stop mutation in SHANK3, there will be disrupted functionality under specific conditions, due to expression of variant proteins at key stages.
Genotype-phenotype correlations can be complicated because of the potential for additional genetic variation contributing to more or less severe phenotypes. In fact, P1 has both an early truncating mutation in SHANK3, as well as a 17q duplication, and has one of the more severe presentations observed at the Seaver Autism Center. However, it is still of interest to note that the three individuals described here are all significantly affected, irrespective of the extent of the truncation. In fact, it is of particular interest that the A5008T variation is associated with a very mild cellular phenotype but with a severe behavioral phenotype in the participant. A better understanding of this association will require additional studies as noted below, although it is also possible that A5008T causes a different cellular phenotype, which has not been assessed in this study.
Transfection experiments as we have carried out here involve the modification of a single SHANK3 isoform (SHANK3a). In addition, we cannot assess other genetic variation that may alter severity and we cannot exclude the possibility that in patient cells, both c.1527G > A and c.2497delG - as being located in coding exons upstream the natural stop codon - lead to SHANK3 haploinsufficiency due to degradation of the truncated proteins by the nonsense-mediated mRNA decay machinery [
39]. It will therefore be essential to differentiate neurons from induced pluripotent stem cells (iPSCs) of participants affected by any of the aforementioned stop mutations and analyze neuronal morphology as well as domain composition, subcellular localization, and mRNA and protein expression of endogenous SHANK3. However, neuronal culture will never completely reflect the impact of a given mutation on the network level
in vivo. Hence, another essential approach to study mutations in
SHANK3 would be to generate animal models that carry the human mutation. With respect to stop mutations disrupting exon 21 of
SHANK3, such as c.2497delG from this study, the recently published
Shank3
ΔC/ΔC
mouse, which is an exon 21 deletion model [
40], may already provide some insights. Importantly, increases in C-terminally truncated Shank3 variants are detectable in hippocampal lysates from these animals. Compared to other
Shank mutant mice [
41,
42], only minimal social deficits were observed in these animals, and increased repetitive behavior was only evident at a certain age (10 to 13 months). CA1 pyramidal neurons did not exhibit a morphological phenotype. The
Shank3
ΔC/ΔC
mutants did clearly show impaired spatial learning and corresponding abnormalities in hippocampal CA3-CA1 physiology though, including decreased long-term potentiation (LTP) and decreased NDMA receptor-mediated synaptic transmission. However, it has to be noted that only analyses of homozygous mutants have been reported so far. Studying heterozygous mutants - as the more ‘human-like’ model - would possibly reveal different phenotypes not only in this but also in other models mimicking human mutations in
SHANK3.
Conclusions
Our in vitro overexpression data show that the location of a stop mutation within the SHANK3a sequence determines both subcellular localization of the truncated protein and the morphological phenotype of the transfected neuron.
Considering domain composition of SHANK3, our data support previous studies [
15,
23-
29,
37] and strengthen the fact that only SHANK3 variants with an intact C-terminus including the Homer binding site, a synaptic targeting element in between the Homer and cortactin binding sites and the SAM domain, correctly and efficiently localize to synapses. Any disruption of these domains results in a distinct phenotype of dendritic and synaptic morphology. Interestingly, our data imply that loss of the Homer and cortactin binding sites is sufficient to induce nuclear accumulation of the corresponding SHANK3 variants. This again supports previous studies proposing that SHANK3 might not only serve as a synaptic but also serve as a nuclear protein [
25,
37] - although its exact nuclear function still remains an enigma.
Considering recent findings on the complexity of both subcellular localization and distinct morphological impact of different rodent Shank3 isoforms in healthy neurons [
37], we propose that any heterozygous deleterious mutation in
SHANK3 will lead to altered SHANK3 isoform expression and thereby result in distinct spatial and temporal perturbations of SHANK3-dependent cellular processes. Although we cannot further investigate such a proposed patho-mechanistic model in the
in vitro assay used in this study, its effect on the network level within the complex and dynamic architecture of the brain would be predicted to be substantial. Moreover, it would also be consistent with our observation that the clinical phenotype remains quite severe in all three individuals included in this study, with clear overlap in the symptom presentation, regardless of whether most or only part of
SHANK3 is missing.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The Icahn School of Medicine at Mount Sinai and Dr Joseph Buxbaum hold a shared patent for Insulin-Like Growth Factor-1 in the treatment of Phelan-McDermid syndrome. The authors declare that they have no competing interests.
Authors’ contributions
MJS conceived the outline of this study together with AK, JDB, and TMB. AK performed and supervised all of the clinical assessments, and SL coordinated the clinical assessments and descriptive data collection. YK subcloned the SHANK3 variants. DMC performed all biochemistry and cell culture experiments together with MS and MPL. DMC and MJS analyzed all in vitro data. MJS drafted the figures and wrote the paper. All authors reviewed and approved the final version of the manuscript.