Atomistic simulations reveal impacts of missense mutations on the structure and function of SynGAP1

Abstract

De novo mutations in the synaptic GTPase activating protein (SynGAP) are associated with neurological disorders like intellectual disability, epilepsy, and autism. SynGAP is also implicated in Alzheimer's disease and cancer. Although pathogenic variants are highly penetrant in neurodevelopmental conditions, a substantial number of them are caused by missense mutations that are difficult to diagnose. Hence, in silico mutagenesis was performed for probing the missense effects within the N-terminal region of SynGAP structure. Through extensive molecular dynamics simulations, encompassing three 150-ns replicates for 211 variants, the impact of missense mutations on the protein fold was assessed. The effect of the mutations on the folding stability was also quantitatively assessed using free energy calculations. The mutations were categorized as potentially pathogenic or benign based on their structural impacts. Finally, the study introduces wild-type-SynGAP in complex with RasGTPase at the inner membrane, while considering the potential effects of mutations on these key interactions. This study provides structural perspective to the clinical assessment of SynGAP missense variants and lays the foundation for future structure-based drug discovery.

Keywords: SynGAP1; in silico mutagenesis; intellectual disability; missense mutation; molecular dynamics (MD) simulation; structural bioinformatics.

Figure 1

SynGAP domains and structure. (A) SynGAP is composed of PH (Pleckstrin homology), C2 (the second conserved PKC domain), Gly-rich Ω-loop motif (G), and GAP (GTPase activating protein) domains at the N-terminal region and the CC (coiled-coil) domain at the C-terminus. (B) Based on the pLDDT (predicted local distance difference test), the ≥50 reliable parts of the N-terminal region and less reliable Ω-loop from the human AlphaFold2 model are shown in 3D (cartoon model). (C) The X-ray crystal structure of rat C2-GAP fragment (orange cartoon; PDB: 3BXJ [26]) aligns well against the model (grey cartoon). (D) the X-ray crystal structure of the mouse CC trimer (PDB: 5JXC [27]) was not used in the modelling.

Figure 2

The C2 domain of SynGAP at the membrane surface. (A) At 300 ns, the C2 domain of WT-SynGAP (cartoon model) is shown stably standing via its ‘loop legs’ at the surface of the inner leaflet membrane model (stick representations). (B) The C2 domain of SynGAP is shown to have positive residues at the interface of the inner leaflet membrane. (C) A zoom-in highlights the hydrogen bond (or H-bonds) and salt bridges formed by Arg335 (orange stick model), explaining the pathogenic potential of the R335H variant when the positive charge is lost and/or bonding network is altered. (D) On the left, those missense mutation residue positions that are at the membrane interface and directly involved in H-bonding are highlighted with green- and orange-coloured spheres, respectively. On the right, a zoom-in shows Val306 packing against Trp308 (CPK models) for the WT system at 50 ns. The V306D mutation disrupts this interaction to allow H-bonding with Ser300 at the membrane-facing loop.

Figure 3

The SynGAP-RasGTPase complex at the inner postsynaptic membrane surface. (A) At 130 ns, the WT-SynGAP-Ras complex (cartoon model) is shown at the inner leaflet membrane model (stick representations) with the bound GTP molecule (CPK model). Covalently linked Cys-palmitoylated lipids 181 and 184 anchor Ras to the membrane (CPK). (B) A zoom-in shows the Ras-bound GTP-Mg²⁺ complex (ball-and-stick models), interacting with the catalytic arginine finger or Arg485 of SynGAP and Gln61 of Ras (stick models). (C) For reference, the human GAP-Ras complex (pink ribbon model; PDB: 1WQ1 [37]) is shown aligned with the SynGAP model (grey ribbon model). (D) The missense mutations directly at the GAP-Ras interface or close-by are shown as pink spheres (C^α atoms) and the ones H-bonding with Ras as blue spheres. The close-ups show SynGAP residues Ser604 and Lys642 H-bonding with Ras, accordingly, their mutagenesis to leucine or threonine, respectively, could weaken the GAP-Ras association.

Figure 4

SynGAP missense mutations disrupting secondary structure, tertiary structure bonding and causing inside-out disruptions. (A) The positions of potentially pathogenic mutations L327P on the β hairpin (res. 322–349), E436K on the α helix (res. 414–436), A469D on the α helix (res. 461–476), and W572R on the α helix (res. 563–578) are shown as spheres on the SynGAP model structure (cartoon model). (B) The W572R variant at 150 ns of the second replicate simulation shows the C2 and GAP domains moving away from each other (arrow). (C) A close up of the W572R variant shows how the positively charged arginine is unable to remain in the hydrophobic inter-helix space as was the case with the original tryptophan. (D) In the L327P variant, the introduced proline cannot maintain the same hairpin backbone H-bond with Gly344 as the leucine does in the WT protein. (E) In the E436K variant, when the negatively charged glutamate is substituted with the positively charged lysine, the salt bridge with Lys444 is exchanged for a charge repulsion that breaks the parent α helix (res. 440–458). (F) the Ala469 resides in a hydrophobic inter-helix space in the WT protein, but, in the A469D variant, the negatively charged aspartate escapes to the outer surface of the α-helix and forms a salt bridge with Arg575. The zoom-in views (D, E, F) were generated using snapshots captured at the 50 ns timepoints in the MD simulations. The mutated residues are shown as ball-and-stick models.

Figure 5

Protein folding stability calculations. (A) At 150 ns, Asp201 and Glu201 (stick models) at the antiparallel β sheet strand (res. 205–208) in both the WT and benign D201E variant simulations, respectively, H-bond with Ser225. The lack of adverse simulation effects is matched by the neutral ΔΔG value (0.42 kcal/Mol) in the FoldX-MD calculations. Conversely, the W308R variant introduces a positive arginine into an anti-parallel β sheet strand (res. 305–315) and within the C2 domain’s hydrophobic core. Although impossible from the folding point of view, the mutation causes only structural distortion without large-scale unfolding in the MD simulations. In contrast, FoldX-MD calculations give the W308R variant a strongly destabilizing ΔΔG value (5.40 kcal/Mol), highlighting its pathogenic status. (B) The confusion matrix compares the neutral and destabilizing predictions of the FoldX-MD approach against the benign (or likely benign) and pathogenic (or likely pathogenic) variants of ClinVar. The percentage of the true positives and negatives are in blue, while the percentages of the false positives and negatives are in red. (C) The number of neutral predictions is only slightly higher than the number of known benign variants, nonetheless, more variants are deemed as destabilizing than have been assigned as pathogenic. The conflicting variants E310K, P562L, and R573W were included as likely pathogenic and R575C as likely benign based on their most prevalent submitted status.