Abstract
Purpose Synaptotagmin-1 (SYT1) is a critical mediator of neurotransmitter release in the central nervous system. Previously reported missense SYT1 variants in the C2B domain are associated with severe intellectual disability, movement disorders, behavioural disturbance and EEG abnormalities. Here, we expand the genotypes and phenotypes and identify discriminating features of this disorder.
Methods We describe 22 individuals with 15 de novo missense SYT1 variants. Evidence for pathogenicity is discussed, including ACMG criteria, known structure-function relationships, and molecular dynamics simulations. Quantitative behavioural data for 14 cases are compared to other monogenic neurodevelopmental disorders.
Results Four variants lie in the C2A domain with the remainder in the C2B. We classify 6 variants as pathogenic, 4 as likely pathogenic, and 5 as variants of uncertain significance. Prevalent clinical phenotypes include delayed developmental milestones, ophthalmic problems, abnormal EEG, movement disorders and sleep disturbance. Discriminating behavioural characteristics were severity of motor and communication impairment, presence of motor stereotypies and mood instability.
Conclusion SYT1 variants associated with neurodevelopmental disorder extend beyond previously reported regions, and the phenotypic spectrum encompasses a broader range of severity than initially reported. This work guides diagnosis and molecular understanding of this rare neurodevelopmental disorder, and highlights a key role for SYT1 function in emotional regulation, motor control and emergent cognitive function.
Introduction
The tightly regulated synaptic vesicle cycle involves the trafficking, docking, fusion, and recycling of neurotransmitter-filled vesicles at the presynaptic terminal. Precision and efficiency of these processes is critical for synchronous neurotransmission, neural network development and emergent cognitive functions1,2. Inherited and de novo variants in synaptic vesicle cycling genes have been associated with a broad spectrum of neurodevelopmental phenotypes including epilepsies, movement disorders, delayed acquisition of motor milestones, intellectual disability (ID), visual impairment and emotional-behavioural disturbances3. Improving the delineation, diagnosis, and management of these disorders requires comprehensive phenotyping in parallel with detailed genetic, molecular and cellular analysis of variants.
Variants in the SYT1 gene, coding for synaptotagmin-1 (SYT1) give rise to SYT1-associated neurodevelopmental disorder, also known as Baker-Gordon Syndrome (MIM 618218). SYT1 is a synaptic vesicle protein that couples action potentials to the synchronous exocytosis of neurotransmitters through its calcium sensing activity4. SYT1, the dominant synaptotagmin family member in the forebrain5,6, is a transmembrane protein with two cytoplasmic calcium-binding (C2A and C2B) domains7. Membrane depolarisation triggers an influx of Ca2+ ions into the nerve terminal, which bind to negatively charged aspartate residues that reside in loops at the “top” of each C2 domain. This neutralises the charge of the loops, acting as an electrostatic switch and allowing the hydrophobic tips of these loops to penetrate the negatively-charged plasma membrane, thereby facilitating fusion of synaptic vesicle and plasma membranes4.
We previously reported 11 cases of de novo missense variants in SYT18,9. All affected individuals presented with hypotonia, developmental delay, and ID varying from moderate to profound severity, and one-third exhibited symptoms of an involuntary movement disorder. Behavioural characteristics included unpredictable switches from placidity to agitation, and pronounced motor stereotypies such as hand-biting. EEG was abnormal in all cases, characterised by intermittent low frequency high amplitude oscillations. Five different SYT1 variants were identified (Met303Lys, Asp304Gly, Asp366Glu, Ile368Thr, and Asn371Lys), all located in highly conserved residues of the C2B domain of SYT1, clustering around the Ca2+-binding pocket. These missense variants were found to inhibit evoked exocytosis in a dominant-negative and mutation-specific manner9,10.
The present paper expands the genotypic and phenotypic spectrum of SYT1-associated neurodevelopmental disorder. We evaluate evidence for pathogenicity of novel variants through in silico analysis and molecular dynamics simulations. We provide quantitative evaluation of behavioural characteristics within the cohort and comparison to individuals with other monogenic neurodevelopmental disorders.
Materials and Methods
Recruitment and sample description
SYT1 variants were identified via whole exome sequencing (trio or solo) within clinical laboratories or ethically approved research studies. The identification, validation, confirmation of de novo status, and clinical reporting of SYT1 variants was carried out by each participant’s clinical centre. Notification of diagnosed variants was communicated to senior authors of the study by personal communication, or identified through database searching (ClinVar, Decipher) or GeneMatcher. An additional 51 participants with ID of known monogenic origin (excluding synaptic vesicle cycling disorders as listed by John et al. (2021)3) were recruited as a comparison group for behavioural data. Genetic diagnoses within the comparison group are listed in Supplementary Materials (Table S1). The SYT1 and ID comparison groups did not significantly differ in range and distributions of age, sex, and global adaptive function (Table S2).
Evaluation of variants
Evaluation of pathogenicity for the 15 de novo sequence variants followed the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG) classification guidelines11 supplemented by Association for Clinical Genomic Science (ACGS) UK best practice guidelines12 (for details see Supplementary Material).
Molecular Dynamics Simulations
Homology models of mutant versions of SYT1 C2 domains were generated from Ca2+-bound C2A and C2B solution structures (PDB: 1BYN13 and 1K5W14; note that amino acid numbering used throughout this paper follows human sequence for simplicity). The WT structure and each mutated homology model (with Ca2+ ions both present and removed), were subjected to four individual atomistic molecular dynamics simulations with trajectory lengths of ∼400ns each (for full details see Supplementary Material). Root-mean-square deviation (RMSD) of the backbone and root-mean-square fluctuations (RMSF) of the backbone C-alpha atoms of each domain variant were measured over the course of the simulations as readouts of overall and local mobility of the domains. To assess Ca2+ mobility and binding pocket occupancy, the distance between each Ca2+ atom and a reference amino acid was measured over time.
Phenotyping methods and analysis
Clinical information for all individuals with de novo SYT1 variants was collated from clinical documentation and parent-report questionnaire (online or by post) using a standard template (for individual clinical histories see Table S3). The study-specific Medical History Interview gathered information about perinatal history, infant and child health, neurological symptoms and developmental milestones. The Vineland Adaptive Behaviour Scales is a standardised assessment of everyday adaptive functioning commonly used to support evaluation of neurodevelopmental disorders. Within the sample, either second edition15 (ID control n=34) or third edition16 (SYT1 n=14; ID control=17) of the Vineland were administered. The Developmental Behaviour Checklist 217 (DBC-P) assesses emotional and behavioural problems in individuals with ID and comprises five subscales (disruptive/antisocial, self-absorbed, communication disturbance, anxiety and social relating). The Social Responsiveness Scale18 (SRS) is a standardised questionnaire enquiring about the presence and severity of social impairments (social motivation, social awareness, social cognition, social communication, and restricted interests and repetitive behaviour). The distributions of all outcome measures were examined for normality prior to parametric or non-parametric analyses as appropriate.
Results
1. SYT1 Variants
Overall, 15 de novo variants in SYT1 were identified in 22 individuals (Figure 1a; 5 missense variants in 11 individuals have previously been described8,9). No alternative candidate variant potentially explaining neurodevelopmental presentation was identified in any case. ACMG criteria classified the variants as pathogenic (n=6), likely pathogenic (n=4) or uncertain (n=5; Table 1). The 11 newly-reported cases comprise one in-frame insertion (Lys367dup) and 10 missense variants (8 at novel loci, one at the previously reported Met303 locus, and one recurrent Ile368Thr variant). All missense variants are at highly evolutionarily conserved residues from human to invertebrates (COBALT; Table S4). While missense variation is not constrained across SYT1 overall (observed to expected ratio = 0.47; gnomAD v2.1.1), the C2A and C2B domains lie within a region demonstrating significant missense constraint, with an observed to expected ratio of 0.24 (chi-square 48.87; ExAC).
2. Molecular Impacts of SYT1 Variants
To further investigate the pathogenic potential of SYT1 variants, we carried out molecular dynamics simulations of the mutated protein domains and searched for known roles of affected residues. For all newly-identified variants, the distances of Ca2+ ions from a reference amino acid in the Ca2+-binding pocket were comparable to the WT simulations (Figure S1), indicating that disturbed Ca2+ retention is unlikely to be a major pathogenic mechanism. In contrast, simulations of the previously reported Asp304Gly variant recapitulated defects in the retention of Ca2+ (Figure S1)9. Although no variant caused major structural changes to the C2 domain (Figure S2), many mutants altered the mobility of discrete regions of the domains (Figures 2, S3 and S4), or would be expected to impact intra- or intermolecular interactions (Figure 2, Table S5). These results suggest that disease-associated variants are likely to impact the structure or function of SYT1 through diverse molecular mechanisms. Literature-informed predictions of molecular impacts and molecular dynamics simulation results are described below for each newly-reported variant.
All four C2A substitutions occur in regions of undetermined function. Leu159, lying in β-2, faces the hydrophobic interior of the domain. The side chain of the introduced charged arginine (Leu159Arg) inserts between the β-sheets and disrupts the stability of the region, with elevated RMSF in multiple regions, particularly across the bottom (non-Ca2+-binding) loops of the C2A domain (Figures 2a and S3). Two substitutions, Thr196Lys and Glu209Lys, are in close proximity to each other in opposing β-strands (β-4 and β-5, respectively) on the edge of the C2A β-sandwich (Figure 1b). Thr196 is structurally important with its side chain buried between β-3 and β-4, inducing a distortion in the β-sheet structure19. Thr196Lys has substantially increased RMSF over residues Lys190 to Lys201 (Figure 2b and S3), indicating localised instability possibly arising from impaired anchoring of β-4 to the rest of the β-sheet. Glu209 forms hydrogen bonds with Lys197 in β-4 of the opposing β-sheet that are removed by lysine substitution (Figure 2f, Table S5). However, there is no observable alteration to RMSF of the Glu209Lys mutant (Figure S3). Glu219 is situated in a short alpha-helix distal to the Ca2+- binding loops; its side chain points away from the protein and could possibly participate in intermolecular interactions. Glu219Gln neutralises the negative charge at that site and removes a salt bridge between Glu219-Lys223 (Figures 2g and 2j, Table S5), but does not result in an appreciable effect on regional RMSF (Figure S3).
Similar to previously reported cases9, all newly-identified C2B missense variants map to the region surrounding the Ca2+-binding pocket (Met303Val and Ser309Pro in loop 1; Tyr365Cys, Lys367dup, and Gly369Asp in loop 3) with one notable exception, Asn341Ser, located in a β-strand (Figure 1b).
Met303 anchors Ca2+-binding loop 1 and replacement with valine is predicted to render the loop more flexible, with increased RMSF across residues 301-306 in Ca2+-free Met303Val simulations (Figures 2c and S4). Ser309Pro would abolish transient hydrogen bonds between Ser309-Met303 and Ser309-Asp304 that are normally present in the Ca2+-bound WT C2B domain (Figure 2h, Table S5), but no change to the RMSF of Ca2+-binding loop 1 was observed in Ser309Pro simulations (Figure S4). In the Ca2+-bound simulations, both Met303Val and Ser309Pro increased the mobility of the distal arginine apex (Arg399, Arg400) at the opposite end of the C2B domain (Figure S4).
No structural alterations were detected in simulations of the Asn341Ser mutant (Figure S4), but importantly Asn341Ser may perturb the interaction between SYT1 and the SNARE complex. Asn341 faces outwards on β-5 and is proximate to the primary binding interface between SYT1 C2B and SNAP-2520. Crystal structures of the SYT1-SNARE complex (PDB: 5CCH, 5CCG and 5KJ7)20 show that Asn341 may interact directly with Asp166 of SNAP-25 or form a hydrogen bond with the neighboring Tyr339 in SYT1, which binds Asp166 of SNAP-25 (Figure 2i). Interestingly, two variants at Asp166 have been identified in individuals with SNAP25 developmental and epileptic encephalopathy21.
Tyr365 stabilises Ca2+-binding loop 3 and the mobility of this loop is increased in Tyr365Cys mutant simulations (Figures 2d and S4). Furthermore, Asp366 flips out of the Ca2+-binding pocket, and Lys367 twists to impinge on the Ca2+-binding pocket in Ca2+-free Tyr365Cys simulations. The positive charge at Lys367 is important for phospholipid binding22-24. The Lys367dup variant introduces an additional positively charged lysine to Ca2+-binding loop 3 (Figure 2k), and thus could potentially increase attraction between the tip of loop 3 and anionic phospholipids. Additionally, simulations showed increased flexibility in loop 3 (Figures 2e and S4) and transient interactions between the inserted lysine and Ca2+-coordinating residues Asp366, Asp373, Asp310 and Asp304. Gly369Asp introduces an additional negative charge to the Ca2+-binding pocket (Figure 2k), more specifically the membrane-penetrating tip of loop 3, which could be expected to repel the anionic plasma membrane.
3. Clinical histories
In keeping with previous findings, the most common features within the SYT1 group were developmental delay, ophthalmic problems and abnormal EEG (cohort summary in Table 2; individual data in Table S3). There is a wider range of severity of neurodevelopmental impairments than previously reported – approximately one third of cases presented with mild or moderate delay to motor and communication milestones, whereas almost all previously reported cases were severely delayed8,9. Other common features were sleep disorder, feeding difficulties, gastrointestinal reflux, and finger-chewing or other self-injury, each affecting around two thirds of cases.
Movement disorder is a feature in two thirds of the cohort, a higher frequency than previously reported9. The types and severities of involuntary movements are variable and include ataxia, tremor and myoclonus, as well as dystonia, chorea and complex hyperkinetic movement disorders in more severely affected cases.
In contrast to our previous case series with no cases of epilepsy9, four individuals in the current group have received an epilepsy diagnosis. Reported seizure phenotypes include absences (n=3), tonic-clonic seizures (n=1) and infantile spasms (n=1). Additional cases are required to clarify the association between SYT1 and epilepsy risk.
EEG abnormalities remain very common across the cohort, encompassing low frequency background oscillations as previously reported, but also ictal features in individuals with and without overt seizures.
Although the majority of individuals within the group are young children, it is possible to make some preliminary comments on the longer-term trajectory of the condition based on 9 individuals currently over the age of 10 years. We note the potential for long-term positive progress in motor development, with two individuals learning to walk after the age of 5 years. However, two individuals have developed movement disorder symptoms during late childhood, accompanied by relative decline in adaptive skills. We also observe that social and emotional difficulties may emerge with time, with some older children and adolescents experiencing obsessions, anxieties and mood disturbance.
4. Behavioural phenotyping and comparison to ID control group
Vineland Adaptive Behaviour Composite (ABC) scores within the SYT1 group ranged from 20 to 74. Four participants scored in the borderline or mild ID range, three in the moderate ID range, and seven in the severe or profound ID range (see Figure 3a for breakdown of subscale scores for each individual). Inspecting Vineland subscale scores within the SYT1 group, we observed that communication ability is more severely impaired on average than motor abilities, socialisation and daily living skills. In contrast, motor ability is a relative strength within the ID comparison group. To establish whether these differences reflect a consistent and discriminating profile of adaptive functions within the SYT1 group we carried out General Linear Model analysis (within-subjects factor: Vineland Subscale; between subjects factors: Group). This highlighted significant interaction between Group and Subscale (F=4.54, df 2.33 Greenhouse Geisser corrected, p=0.01; reducing to p=0.09 after co-varying for Vineland edition). Post-hoc non-parametric analyses indicated significantly lower scores for the SYT1 group in Motor ability (p=0.036) and Communication (p=0.019) but not Socialisation (p=0.07) or Daily Living skills (p=0.28) (Figure 3b).
We next explored DBC-P and SRS-2 scores, assessing emotional or behavioural problems and autism-related social impairment respectively. No significant differences emerged between SYT1 and ID comparison groups for total scores on either measure, and GLM analyses identified no significant Group x Subscale interaction for either measure (Figures S5 and S6).
In our previous description of SYT1-associated phenotypes4, based on clinical reports only, we noted that a high proportion of diagnosed individuals displayed motor stereotypies, unpredictable mood switches, and episodes of agitation. To explore whether these features are increased in the expanded SYT1-diagnosed population, beyond expectation for ID, five relevant items reflecting these symptoms were selected from the DBC-P (6. Bangs head; 10. Chews or mouths body parts; 33. Hits or bites self;
47. Mood changes rapidly for no reason; 60. Has repeated movements of hands, body or head). These items appear across two DBC-P subscales (disruptive/antisocial, self-absorbed). Raw scores (0-2) for each item were summed for each participant. Univariate analysis, co-varying for age, identified significantly higher scores for these selected DBC items within the SYT1 group (SYT1 group M= 5.57, SD= 2.95, range 0-9; ID comparison group M=3.67, SD=2.98, range 0-10; F=5.95, df 1, p=0.02). Age-adjusted summed scores showed a strong negative correlation with Vineland ABC within the SYT1 group (Spearman’s rho -0.73, p=0.003), but not the comparison group (Spearman’s rho -0.18, p=0.26; Fisher’s test25 z=-2.19, p=0.01) (Figure S7). In summary, self-injury, mood instability and repetitive movements are elevated within the SYT1 group, with a strong relationship between these features and global adaptive impairment.
Discussion
This work builds on the previous identification and characterisation of SYT1-associated neurodevelopmental disorder to broaden the range of potentially pathogenic variants for clinical laboratory reporting. We detail the expansion of the genotypic and phenotypic spectrum of this syndrome with the objective of increasing the diagnostic efficiency of this ultra-rare disorder as well as improving prognosis estimation, patient management, and information available for families and clinicians after diagnosis.
1. Broadening the genetic landscape
We describe four novel variants in the SYT1 C2A domain as well as additional C2B loci, revealing that neurodevelopmental disorder-associated variants are found in the two functional C2 domains of SYT1. Whilst C2A variants remain of uncertain significance according to ACMG criteria, and functional evidence is required to substantiate pathogenicity, molecular dynamics simulations and existing literature are supportive of likely functional impact. Novel variants may exert dominant-negative effects, as seen for previously identified mutants8-10, but we cannot preclude haploinsufficiency as a possible pathogenic mechanism. Ca2+-binding loops 1 and 3 of the C2B domain appear to be highly sensitive to mutation as disease-associated missense variants cluster in these regions in both SYT1 and the homologous SYT226,27, and no missense SYT1 variants in these loops are recorded in gnomAD v2.1.1. Therefore, any de novo missense variant in the C2B Ca2+-binding loops should be investigated for possible pathogenicity and variants in other highly conserved residues of the SYT1 C2 domains should also be considered.
2. Broadening the phenotypic spectrum
The major clinical features associated with SYT1 variants in this larger cohort are broadly in keeping with those reported in the previous case series9. SYT1-associated neurodevelopmental disorder presents with individually non-specific features, but may be suspected when neonatal hypotonia, developmental delay, ophthalmic problems, movement disorders and EEG abnormalities are present in any combination. While some of the newly-identified cases present with profound developmental delay and involuntary movement disorders, in line with the presentation of initial cases, others show milder neurodevelopmental difficulties thereby widening the range of clinical severity. For these less severe cases, it is more difficult to distinguish a discriminating phenotypic signature to aid in variant interpretation and diagnostic confirmation. Given the young age of some individuals within the cohort and uncertainty of the developmental trajectory of this disorder, a cautious prognosis is warranted. We note that amongst some older diagnosed cases, later-onset movement disorder and decline in adaptive function and emotional wellbeing have been observed. Longitudinal sampling of individuals with SYT1 variants will facilitate mapping of the developmental progression of this disorder and improve prognosis estimations.
Questionnaire data revealed that all domains of adaptive function are impaired within the SYT1 cohort, and highlighted disproportionate impact on motor and communication function compared to a comparison ID group (although note that data could not be collected for 7 SYT1 cases due to language barriers or young age). Moreover, features of rapid mood change and motor stereotypies (specifically hand biting) are prominent, in line with previous observations9. Improved understanding of the cell-type specific dependence on SYT1 for efficient neurotransmitter release will aid symptom explanation at the circuit and large-scale network levels. Future investigation will involve additional parent/carer-report questionnaires to further probe neurological and behavioural domains impacted by SYT1 mutations, such as visual behaviour, movement disorder symptoms, repetitive behaviours and hyperactivity.
With expansion of the cohort, SYT1-associated neurodevelopmental disorder continues to demonstrate intersecting clinical features that are common amongst synaptic vesicle cycling disorders3. However, seizures are a significant feature of disorders associated with mutation of other core synaptic vesicle fusion machinery including VAMP2, STX1B, SNAP25, STXBP1, CPLX1 and other accessory proteins28,29. In notable contrast, while four SYT1 cases have reported seizures, epilepsy is evidently not a prevalent or prominent feature in this disorder. While EEG abnormalities were frequent within the SYT1 group, the electrophysiological features observed were variable and there were inconsistencies in reporting methods, provision of recordings for review, patient age and conditions of recordings. Standardised and systematic EEG data collection and analysis will be needed to confirm common electrophysiological characteristics and inform understanding of the neurophysiological origins of symptoms. Furthermore, systematic behavioural characterisation of other disorders of fusion, and synaptic vesicle cycling disorders more broadly, will allow more detailed comparison of these mechanistically-related syndromes and interrogation of distinctive characteristics of this group of disorders.
3. Diversity of molecular mechanisms
Molecular dynamics simulations and SYT1 structure-function relationships sourced from the literature were used to carefully consider variants in the context of the 3D protein structure. Newly-identified variants are predicted to have nuanced effects on the local stability of discrete protein regions (including Ca2+-binding loops or β-sheets), penetration of the phospholipid membrane, or interaction with SNARE proteins, rather than grossly destabilise the C2 domain or substantially impair Ca2+ binding. Such perturbations could be expected to disrupt SYT1 function and synaptic transmission. Some disorder-associated variants, particularly those within the C2A domain, provide the first indication of the importance of previously unrecognised residues and regions of SYT1. It should be noted that while molecular dynamics simulations provide atomic-level predictions of variant impact on protein structure and Ca2+ retention, these simulations are limited in length and unable to reveal impacts on protein-protein or protein-lipid interactions or consequences on neurotransmission. Functional studies at the molecular, cellular and circuit levels will provide further evidence for pathogenicity and insight into the specific mechanisms underlying neurodevelopmental impairments.
4. Genotype-phenotype links
We inspected the current datasets for evidence that the specific SYT1 variant may contribute to clinical features and severity. Questionnaire data was available for five cases with the recurrent Ile368Thr variant, which revealed that the neurological and behavioural phenotype was highly consistent between these individuals (Figure 3, Table S3). Although questionnaire data was not available for other recurrent variants, clinical reports of recurrent locus Met303 and variants Asp366Glu and Asn371Lys show similar consistency in phenotype severity (Table S3). This raises the possibility of a relationship between diversity of molecular mechanisms and phenotypic variation. No obvious patterns emerged between the clinical phenotype and either the nature of the amino acid substitution or the conservation of the affected residue across synaptotagmin isoforms.
We subsequently asked whether there are consistent differences between C2A and C2B domain variants in global severity of impairments. While acknowledging that cases of C2A variants are limited, it is notable that all four of these individuals present with mild or moderate constraints to adaptive function and absence of early-onset movement disorder. Correspondingly, the variants linked to the most severe clinical phenotypes are all situated in the C2B domain. The prospect of differential clinical severity between the C2A and C2B domains is congruent with the current understanding that each C2 domain plays a distinct role. Specifically, the C2B domain is essential for Ca2+-dependent membrane binding and fast synchronous neurotransmitter release, while the C2A domain may be facilitatory and mediate other modes of release30-34. However, there is not an absolute divide in clinical severity between C2A and C2B domains, as not all C2B variants are severe with several cases of intermediate severity and two notably milder novel variants (Asn341Ser and Gly369Asp). It is also important to note that no identified C2A variant is located around the Ca2+-binding loops (in contrast to most C2B variants), which may also account for this differential presentation. Most C2A variants fall within β-strands and could have diffuse structural impacts that differ from specific disruption of the Ca2+- binding loops. Indeed, the only C2B variant outside the Ca2+-binding loops (Asn341Ser) is associated with comparatively milder features. However, mildly presenting Gly369Asp is situated in a C2B Ca2+- binding loop, arguing that location within the domain does not completely explain clinical severity. Identification of additional variants and detailed phenotyping of recurrent variants will assist in clarifying genotype-phenotype relationships. Further, the impacts of variants on all functions of SYT1, not only evoked exocytosis but also suppression of spontaneous and asynchronous release and modulation of endocytosis35,36, need to be investigated to fully appreciate similarities and differences in molecular and cellular impacts. Integration of genetic, cellular, neural systems and cognitive investigations will enable a thorough understanding of SYT1-associated neurodevelopmental disorder, with the prospect of precision medicine targeting each individual’s symptoms and underlying mechanisms.
Data Availability
Reporting of SYT1 variants in open access repositories is listed in Table S3. Molecular dynamics simulation data not included in the Supplementary Material are available upon request to the corresponding author.
Data Availability
Reporting of SYT1 variants in open access repositories is listed in Table S3. Molecular dynamics simulation data not included in the Supplementary Material are available upon request to the corresponding author.
Author Information
Author Contributions
Conceptualization: K.B., H.M., S.L.G.; Formal Analysis: F.B. K.B., A.K.-T., A.J. Funding acquisition: K.B., S.L.G.; Investigation: F.B., H.M., A.K.-T., K.B., E.N.-C., A.J.; Resources: J.J.Z., P.C., S.J., M.L., C.F., J.T., F.E., M.W., T.K., R.P., R.B., R.C., L.L., M.A., J.v.d.S., M.N., M.K.; Visualization: H.M., A.K.-T., F.B.; Writing –original draft: H.M., K.B., F.B.; Writing – review & editing: H.M., K.B., S.L.G., D.J.S, all other authors reviewed and approved this manuscript.
Ethics Declarations
This study was conducted within the “Phenotypes in Intellectual Disability” project, which received approval from Cambridge Central Research Ethics Committee (REC ref: IRAS 83633). Written informed consent was provided by each diagnosed individual’s parent or consultee prior to information sharing by referring clinicians, and questionnaire completion by parents or carers.
Competing interests
The authors declare no conflict of interest.
Supplementary Material
Supplementary Methods
Evaluation of Variants
gnomAD v2.1.1 (The Genome Aggregation Database) (https://gnomad.broadinstitute.org/) was searched to identify allele frequency (AF) of reported variants in control populations. To identify the evolutionary conservation of the amino acids in the protein sequence across 14 species from Human to Caenorhabditis elegans, we employed COBALT (Constraint-based Multiple Alignment Tool) (https://www.ncbi.nlm.nih.gov/tools/cobalt/cobalt.cgi). We deployed three in-silico predictive programmes, namely SIFT version 5.2.2 (Sorting Intolerant From Tolerant) (http://sift.jcvi.org) to predict the effect of amino acid substitution on protein function, PolyPhen-2 (Protein Analysis Through Evolutionary Relationship) (http://genetics.bwh.harvard.edu/pph2/) to predict deleteriousness of single nucleotide variants and insertions/deletions variants, M-CAP (Mendelian Clinically Applicable Pathogenicity) (http://bejerano.stanford.edu/mcap/) to combine previous pathogenicity scores (SIFT, PolyPhen) to increase the overall sensitivity. Evidence for pathogenicity and classification were carried out independently by two authors. Phenotypic similarity (supporting criterion PP4 in Richards et al. 20151) was not applied, to avoid bias of our analysis of phenotypic spectrum.
Molecular Dynamics Simulations
MD simulations were carried out based on NMR structures of the synaptotagmin-1 (SYT1) C2A (PDB: 1BYN2) and C2B (PDB: 1K5W3) domains, or mutant variants thereof, either in the presence or absence of bound Ca2+ atoms. Single mutant homology models of SYT1 C2A comprising mutations Leu158Arg, Thr195Lys, Glu208Lys and Glu218Gln, and SYT1 C2B comprising mutations Met302Val, Asp303Gly, Asn340Ser, Tyr364Cys and Gly368Asp were generated using Swiss-PdbViewer v4.1.0, while the homology model containing the Lys366 duplication (K366dup) was generated using SWISS-MODEL4 (note that numbering used throughout this paper follows human sequence for simplicity (i.e. Leu159Arg, Thr196Lys, Glu209Lys and Glu219Gln, Met303Val, Asp304Gly, Asn341Ser, Tyr365Cys and Gly369Asp). The SYT1 C2A models were C- and N- terminally capped with Ace and NMe groups, respectively, to eliminate aberrant interactions with Ca3. Ca2+ atoms were removed from homology models prior to equilibration for simulations without Ca2+. All MD simulations were performed with the Gromacs molecular dynamics simulation package running version 2019.4 or 2020.35 using the Amber ff99SB-ILDN forcefield6 under periodic boundary conditions and in a rhombic dodecahedron unit cell. Each system was solvated with simple point charge (SPC) water molecules7 and Na+ and Cl− ions were added to neutralize the total charge of the systems at concentrations of 150 mM. The neighbour list, Coulomb and van der Waals interaction cut-offs were set to 1 nm and the particle mesh Ewald (PME) algorithm8 was utilised for long-range electrostatic interactions. All systems were first subjected to a 2000 step steepest descent energy minimization, alternatively completing when the maximum force on any atom has reached 1000 kJ mol-1 nm-1. A 1 ns NVT (constant Number of particles, Volume and Temperature) was then performed, heating the systems to 310 K using the V-rescale thermostat9 with position restraints on. This was followed by a 1 ns NPT (constant Number of particles, Pressure and Temperature) equilibration run performed with position restraints on using the V-rescale thermostat9 and the Berendsen barostat10. A third equilibration phase using randomised initial velocities, the V-rescale thermostat11 (310 K) and the Parrinello-Rahman barostat12 was performed for 10 ns using a 2 fs time step. Four individual production runs of ∼400ns were carried out for each system under the same conditions as the final equilibration phase but with position restraints switched off. We expected to observe nuances across four simulations that may not be observable in a single trajectory and, importantly, results for WT and Asp304Gly structures obtained with these four shorter simulations are in good agreement with previous results for single ∼1200ns trajectories based on the same starting models13. The MD trajectories were simplified by extracting every nanosecond and by removing all water molecules using the GROMACS trjconv utility. All simulations were carried out on the ‘Big Red 3’ system provided by the Indiana University Pervasive Technology Institute, Bloomington IN. Data analysis was carried out in VMD (v1.9.3) using the RMSF, RMSD and hydrogen bond (cut-off set to 3.5Å and 20°) plugins and Graphpad Prism (v9.0.0), and images of molecular snapshots were generated with PyMOL (v2.4.0). To measure Ca2+ retention, the distance between each Ca2+ atom and the Cγ of Asp231 of C2A or Asp364 of C2B was measured for all trajectories and plotted as a function of simulation time. Asp231 and Asp364 were used as reference amino acids as they are directly involved in electrostatic interactions with Ca1 and Ca2 in each C2 domain and are the most stable among Ca2+-interacting residues.
While Ca1 and Ca2 of both C2 domains remained stably bound in WT simulations (Figure S1), Ca3 occupancy of the C2A Ca2+-binding pocket was remarkably unstable with Ca3 dissociating from the WT domain at early timepoints in all four trajectories (Figure S8). Ca3 also dissociated from each mutated C2A domain (Leu159Arg, Thr196Lys, Glu209Lys, Glu219Gln) in at least one trajectory (Figure S8). Early Ca3 dissociation is reasonable given the low affinity of Ca3 (Kd of >>1mM)14,15, and it is unlikely that any of the mutations make a significant contribution towards stabilising Ca3 as a correlation with RMSF values of residues involved in Ca3 binding could not be established.
Supplementary Data
Supplementary Table 4. Evolutionary conservation at sites of novel SYT1 variants. Sequence alignment performed with COBALT (https://www.ncbi.nlm.nih.gov/tools/cobalt/cobalt.cgi). Patient variants are shown in red, while blue highlights amino acids that are not identical to the human sequence.
Acknowledgements
We appreciate the important and generous contributions of each individual with a SYT1 variant, their families and carers. We appreciate the roles of each clinician and laboratory scientist involved in the diagnostic pathway of each case. Further, we appreciate the involvement of individuals in the comparison ID group and their families and carers. We acknowledge the NIHR UK Rare Genetic Disease Research Consortium for assistance with recruitment to the study. HM is supported by an Australian Government Research Training Program Scholarship. SLG and HM are supported by Grant 2003710 from The National Health and Medical Research Council (NHMRC) of Australia and a Florey Fellowship. The Florey Institute of Neuroscience and Mental Health acknowledges the strong support from the Victorian Government and in particular the funding from the Operational Infrastructure Support Grant. This project was supported by funding to KB from the UK Medical Research Council (G101400), the Baily Thomas Charitable Trust, and Great Ormond Street Hospital Children’s Charity. FB and JJZ were supported by the Indiana University Precision Health Initiate. Molecular dynamics simulations were supported in part by Lilly Endowment, Inc., through its support for the Indiana University Pervasive Technology Institute. We thank Dr. Trayder Thomas (University of Chicago) and Dr. Billy Williams-Noonan (RMIT University) for assistance with molecular dynamics simulations. The study makes use of data generated by the DECIPHER community. A full list of centres who contributed to the generation of the data is available from https://decipher.sanger.ac.uk and via email from decipher{at}sanger.ac.uk. Funding for the DECIPHER project was provided by Wellcome Trust. Individuals who contributed to DECIPHER bear no responsibility for the interpretation of the data by the authors. We acknowledge the DDD study for identification of variants reported in this paper. The DDD study presents independent research commissioned by the Health Innovation Challenge Fund [grant number HICF-1009-003] a parallel funding partnership between the Wellcome Trust and the Department of Health, and the Wellcome Trust Sanger Institute [grant no. WT098051]. See Nature 2015; 519:223-8 or www.ddduk.org/access.html for full acknowledgement.
References
Supplementary References
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.