Variants in BSN, encoding the presynaptic protein Bassoon, result in a distinct neurodevelopmental disorder with a broad phenotypic range

Summary

Disease-causing variants in synaptic function genes are a common cause of neurodevelopmental disorders (NDDs) and epilepsy. Here, we describe 14 individuals with de novo disruptive variants in BSN, which encodes the presynaptic protein Bassoon. To expand the phenotypic spectrum, we identified 15 additional individuals with protein-truncating variants (PTVs) from large biobanks. Clinical features were standardized using the Human Phenotype Ontology (HPO) across all 29 individuals, which revealed common clinical characteristics including epilepsy (13/29, 45%), febrile seizures (7/29, 25%), generalized tonic-clonic seizures (5/29, 17%), and focal-onset seizures (3/29, 10%). Behavioral phenotypes were present in almost half of all individuals (14/29, 48%), which included ADHD (7/29, 25%) and autistic behavior (5/29, 17%). Additional common features included developmental delay (11/29, 38%), obesity (10/29, 34%), and delayed speech (8/29, 28%). In adults with BSN PTVs, milder features were common, suggesting phenotypic variability, including a range of individuals without obvious neurodevelopmental features (7/29, 24%). To detect gene-specific signatures, we performed association analysis in a cohort of 14,895 individuals with NDDs. A total of 66 clinical features were associated with BSN, including febrile seizures (p = 1.26e−06) and behavioral disinhibition (p = 3.39e−17). Furthermore, individuals carrying BSN variants were phenotypically more similar than expected by chance (p = 0.00014), exceeding phenotypic relatedness in 179/256 NDD-related conditions. In summary, integrating information derived from community-based gene matching and large data repositories through computational phenotyping approaches, we identify BSN variants as the cause of a synaptic disorder with a broad phenotypic range across the age spectrum.

We identify 14 individuals with de novo variants in BSN, a gene encoding the presynaptic protein Bassoon, linked to NDD and epilepsy. Detailed phenotypic analysis revealed a broad range of clinical features in individuals with BSN variants. Phenotypic expression varies significantly between pediatric and adult cohorts, suggesting an age-specific signature.

Introduction

Variants in genes linked to synaptic function have emerged as common contributors to neurodevelopmental disorders (NDDs) and epilepsy.^1,2, At the presynaptic active zone, the protein encoded by the BSN (MIM: 604020) functions as a scaffolding protein that coordinates the positioning of synaptic vesicles and organizes molecular components critical for rapid neurotransmitter release, supporting precise synaptic signaling and plasticity.^3,4,5 Disruption of such genes is a known causal mechanism for NDDs, including those caused by variants in SHANK3 (MIM: 606230), SYNGAP1 (MIM: 603384), and DLG4 (MIM: 602887).^6,7,8 Disorders of synaptic function are increasingly associated with clinical phenotypes spanning epilepsy, autism spectrum disorder (ASD [MIM: 209850]), and intellectual disability.^6,7,9,10

BSN is highly expressed in the brain, and several Bsn-deficient mouse models suggest its potential link to seizures.^11,12,13,14 In Bsn mutant mice, the loss of functional BSN protein disrupts synaptic ribbon architecture in the retina and impairs presynaptic function, leading to sensory deficits and epileptic seizures.^12,14,15,16 Although BSN has been linked to brain disorders, few clinical cases with BSN variants have been reported, leaving the associated phenotypic spectrum unclear. Prior studies suggested variation in the BSN as a contributor to epilepsy with febrile seizures and a largely favorable outcome.¹⁷ However, the full spectrum of BSN-related phenotypes in larger cohorts has not been assessed to date, including the unique phenotypic consequences of de novo variants, as well as protein-truncating variants (PTVs) that are inherited or of unknown inheritance.

To further delineate BSN-related phenotypes, we leverage the Human Phenotype Ontology (HPO), a standardized framework that harmonizes clinical data across large, heterogeneous cohorts.^18,19,20,21 By mapping phenotypic features to HPO terminology, subtle phenotypic patterns can be uncovered that might otherwise be obscured.^20,21 Previous studies have demonstrated the power of using the HPO in large-scale genetic research, where it has been used to identify novel gene-phenotype associations.^{19,20,21,22,23} For example, AP2M1 (MIM: 601024) was implicated in epilepsy and NDDs through the characterization of individuals carrying de novo variants, highlighting the role of endocytosis in synaptic function.²² Accordingly, by systematically analyzing phenotypic similarities, HPO helps bridge the gap between genotype and phenotype, providing critical insights into the genetic basis of complex disorders.

Here, we applied the HPO framework to a cohort of 29 individuals with BSN variants, including 14 individuals with de novo variants, 13 individuals with PTVs of unknown inheritance, and two individuals with PTVs with paternal inheritance. Affected individuals presented with diverse neurodevelopmental phenotypes, including behavioral abnormalities, delayed speech, learning disabilities, and variable seizure types. By harmonizing phenotypic features through an HPO-based approach, we explored the phenotypic landscape of BSN-related disorders and examined the variability of phenotypes across the age span.

Material and methods

Participant recruitment

We identified individuals with BSN variants through multiple sources, ensuring that all variants were either de novo missense variants or PTVs, in order to have a high level of certainty of the variant’s causality. Parental informed consent for participation in this study was obtained for all subjects, in accordance with the Declaration of Helsinki and with approval from the institutional review boards (IRBs) of the respective institutions.

Two participants (ages 14–15 years) were enrolled in the Epilepsy Genetics Research Project (EGRP, IRB 15–12226) cohort at Children’s Hospital of Philadelphia (CHOP). Both individuals had de novo PTV BSN variants identified through diagnostic trio whole-exome sequencing (WES). Clinical data for these research participants were manually extracted from their electronic medical records (EMRs).

An additional 14 individuals (ages 3–23 years) were identified through GeneMatcher,²⁴ an online platform that facilitates international collaborations by matching researchers and clinicians with overlapping genetic findings. Among these, nine individuals had confirmed de novo BSN variants (two missense and seven PTV), verified by their respective institutions.

Seven individuals (ages 33–83 years) with BSN PTVs were identified through the Penn Medicine BioBank (PMBB).²⁵ The PMBB operates under IRB protocol #813913, with approval from the IRB at the University of Pennsylvania.

One individual (age 3 years) with a missense de novo BSN variant was identified from the Birth Defects Biorepository (BDB) at CHOP. The BDB is an IRB-approved protocol (#18-015525), designed to store and permit access to biological specimens and longitudinal clinical and research data for future studies on birth defects.

Three additional individuals (ages 9–17 years) with BSN PTVs were identified through the Center for Applied Genomics (CAG) at CHOP, a pediatric genomics research program focused on complex traits and rare diseases.

A literature review identified two previously reported individuals with de novo PTVs in BSN.¹⁷ Clinical data from the prior report was translated to HPO terms. While the exact ages of these individuals were not reported, the available clinical data were collected during infancy or early childhood, with seizure outcomes documented up to 3 years of age. Both individuals were included in the analysis of pediatric individuals with BSN variants.

Variant identification and annotation

Variants were identified through trio WES and were confirmed using standardized protocols as described previously.²² Diagnostic sequencing for individuals identified through GeneDx²⁶ (individuals 4, 6, 7, 9, and 10) was conducted using exome capture platforms such as the IDT xGen Exome Research Panel v1.0 or v2.0 (Integrated DNA Technologies) and Twist Bioscience Exome 2.0 (Twist Biosciences), followed by massively parallel sequencing on Illumina platforms with paired-end reads of >100 bp. Sequencing data were aligned to the human genome reference build GRCh37/UCSC hg19, and variants were called using institution-specific pipelines, ensuring high-quality annotation and filtration.

For participants recruited through the EGRP cohort at CHOP, CAG, and PMBB, variant annotations were performed using ANNOVAR.²⁷ Additional variant filtration criteria included allele frequency (AF) <0.005 (based on gnomAD v4)²⁸ and pathogenicity predictors such as combined annotation dependent depletion (CADD) > 15, rare exome variant ensemble learner (REVEL) > 0.2, and genotype quality (GQ) >30.

These thresholds ensured the retention of rare, likely pathogenic variants. Variants identified in GeneMatcher cohorts were confirmed by respective contributing institutions following previously published standards for variant interpretation. The sequencing and annotation methods used across all contributing cohorts were consistent with best practices for genomic analyses and align with protocols previously described in the literature.²²

For the single individual identified through BDB, whole-genome sequencing and data processing were performed by the Genomics Platform at the Broad Institute of MIT and Harvard. DNA libraries were prepared using the Illumina Nextera or Twist exome capture (∼38-Mb target) and sequenced with 150-bp paired-end reads, achieving >85% of targets covered at 20× and a mean target coverage of >55×. Sequencing data were processed through a pipeline based on Picard, with read mapping performed using the burrows-wheeler aligner (BWA) aligner to the human genome build 38 (GRCh38). Variants were called using the genome analysis toolkit (GATK) HaplotypeCaller package version 3.5 following best practices for variant detection.

Phenotypic analysis

Clinical phenotypes from the EGRP cohort at CHOP, PMBB, and CAG participants were confirmed through manual review of EMR, ensuring accurate phenotypic information. All clinical data associated with the research participants were mapped to HPO terms. For phenotyping forms and databases, we manually mapped clinical terms to HPO terms (HPO version 1.2; release format version, 1.2; data-version, releases/2023-10-09; downloaded on 11/10/23) in accordance with prior studies.²³ The phenotypes of all individuals were manually coded by expert reviewers. Phenotypes were first extracted by research staff with clinical and biomedical knowledge and experience with the HPO by using all available clinical and research notes for an individual and by using the most specific HPO terms applicable. These assigned terms were then reviewed and verified by domain experts, either a physician or genetic counselors specialized in epilepsy genetics. In cases of ambiguity and uncertainty, a higher-level, more general HPO term was coded rather than a more specific term.

For each individual, all higher-level (ancestral) HPO terms were derived as previously reported.^22,23,29 This method, known as propagation, results in a base and propagated set of HPO terms for each individual.^22,23,29 The propagated HPO dataset from the entire cohort was used to generate baseline frequencies (f) for all HPO terms. Information content (IC) of each term was defined as the −log2(f), with a higher IC value reflecting a more specific and less frequently encountered HPO term in the cohort. In the current manuscript, we use a compact internationalized resource identifier (CURIE) to refer to HPO terms, i.e., “HP:0001250” (“Seizures”) abbreviates https://hpo.jax.org/app/browse/term/HP:0001250 in accordance with the Open Biological and Biomedical Ontologies (OBO) Citation and Attribution Policy as previously described.^20,29 For readability of the manuscript, we omit quotation marks for phenotypes expressed in HPO terms, streamline their descriptions, and adjust the grammatical usage of these terms within sentences. When followed by HPO identifier [e.g., “[…] seizures (HP:0001250)”], a phenotype refers to a clinical term coded in HPO terms rather than a more general reference to this phenotype.

For PMBB and CAG cohorts, International Classification of Diseases (ICD)-9 and ICD-10- clinical modification (CM) codes were provided as part of the datasets and were translated into HPO terms using a predefined mapping table.^30,31 Longitudinal clinical data for these research participants were also provided and incorporated into the phenotypic analysis. Additionally, the ICD-to-HPO mapping process and longitudinal data integration underwent quality-control steps to ensure robust phenotypic alignment across cohorts. This standardization facilitated a uniform phenotypic analysis across all cohorts.

Data integration and processing

All datasets were curated to ensure consistency in variant annotation and phenotypic mapping. This process included manual validation steps to ensure accuracy in EMR data extraction, ICD-to-HPO mapping, variant filtration, and cohort selection.

Statistical and computational analyses

All computations were performed using the R statistical framework. To assess the association between BSN variants and phenotypic features, we utilized statistical and computational methods aligned with those detailed in our previous publication.²⁹ Volcano plots were generated to visualize association results, plotting –log₁₀(p value) against log₂(odds ratio), and deriving p values through Fisher’s exact tests.

Phenotypic similarity (sim) analyses were conducted using the simmax algorithm due to its established use in prior studies.^22,29 Permutation testing (100,000 iterations) was employed to validate the statistical significance of phenotypic clustering, ensuring observed similarities exceeded those expected by chance. Specifically, the median similarity score for each gene was compared to a null distribution derived from random permutations of phenotypic data. The denovolyzeR tool was used to determine the probability of n de novo variants in a given gene.³²

Results

Identification of two de novo BSN frameshift variants in individuals with early-onset seizures

We identified two individuals with de novo frameshift variants in the BSN through clinical exome sequencing. The BSN (NM_003458.4) variants in individual #1 (c.8158_8162delACGGA; p.Thr2720Alafs^∗38) and individual #2 (c.867dupG; p.Pro290Alafs^∗27) were absent in gnomAD (Figure 1). BSN is predicted to be highly intolerant to loss-of-function variation, with a probability of loss-of-function intolerance (pLI) score of 1.²⁸

Figure 1.

Overview of BSN variants identified in 29 individuals

(A) Diagram of presynaptic active-zone assembly with synaptic vesicle fusion machinery proteins, showing BSN (red) as a key scaffolding protein in synaptic vesicle positioning and release.

(B) BSN with variant distribution, where de novo variants (top) include both missense (blue) and protein-truncating variants (PTVs, red), while “Other” shows that inherited and unknown inheritance variants (bottom) consist of PTVs only.

Both individuals presented with febrile seizures before 18 months of age. They remained seizure free until the ages of 7 and 8 years old, respectively, when individual #1 had a first unprovoked bilateral tonic-clonic seizure, and individual #2 presented with absence seizures. For individual #1, seizures were infrequent initially and were managed with levetiracetam. By 12 years of age, individual #1 started to have monthly bilateral tonic-clonic seizures, accompanied by a decline in academic performance. Individual #2 had infrequent absence seizures, followed by focal impaired-awareness seizures and generalized tonic-clonic seizures. Both individuals exhibited behavioral abnormalities at 6 years and were diagnosed with attention deficit hyperactivity disorder (ADHD [MIM: 143465]) at 5 and 11 years, respectively. Individual #1 had early developmental delays, particularly in language, and was diagnosed with autism at 5 years. Both individuals had learning disabilities that necessitated specialized schooling (Table 1).

Table 1.

Clinical and genetic features in 14 individuals with de novo BSN-related disorders

	Age at last evaluation	Sex	Variant	Exon	Epilepsy/seizure types	Seizure frequency	Age at seizure onset	Seizure outcome	Developmental features	Other notable features	EEG features	MRI features
Individual #1 local cohort	14 years	M	c.8158_8162delACGGA; p.Thr2720Alafs∗38	5	FS, BTC	frequent between 18 months and 3 years. one GTCS 6 years later	18 months	SF > 5 years while on levetiracetam, relapse at 9 years	global DD, ADHD, autism, SLDD	behavioral abnormality	generalized spikes	cerebellar atrophy
Individual #2 local cohort	15 years	M	c.867dupG; p.Pro290Alafs∗27	3	FS, GTCS, AS; nocturnal seizures	2 per month	11 months	SF < 1 year on clonazepam and topiramate	DD, ADHD, ODD, LD, behavioral concern	abnormal lab findings, sleep disturbance, obesity	temporal sharp waves, focal epileptiform discharges, focal spike waves	unremarkable
Individual #3 GeneMatcher	3 years	M	c.10255C>T; p.Gln3419∗	6	FMS; CFS; BTC with focal and generalized onset seizures	2 FSE. 1 GTCS, multiple AS	1 year	SF < 1 year on levetiracetam	global DD, speech developmental stagnation at onset of seizures	ataxia, gait imbalance, atypical behavior	interictal abnormality	suspicion of focal cortical dysplasia
Individual #4 GeneMatcher	19 years	F	c.8095G>T; p.Glu2699∗	5	seizure	2 lifetime seizures	8 years	SF	global DD, mild ID, SLDD, FTT, LD	tachycardia	unremarkable	haziness of the gray-white matter interface in the right anterior temporal pole
Individual #5 GeneMatcher	17 years	M	c.7916C>G; p.Ser2639∗	5	no seizures	NA	NA	N/A	global DD, mild ID	hypotonia, atypical behavior, macrocephaly	unremarkable	abnormal cerebral white matter morphology
Individual #6 GeneMatcher	6 years	F	c.9499C>A; p.Pro3167Thr	6	no seizures	N/A	N/A	N/A	DD, SLDD, LD	hypotonia, gait imbalance, sleep disturbance, abnormal emotion/affect behavior	N/A	unremarkable
Individual #7 GeneMatcher	7 years	M	c.620C>A; p.Pro207His	2	no seizures	N/A	N/A	N/A	moderate DD, autism, SLDD, LD	hypotonia, atypical behavior, sleep disturbance	N/A	unremarkable
Individual #8 GeneMatcher	9 years	F	c.4138delA; p.Thr1380Profs∗19	5	staring episodes; AS	daily or every other day	N/I	SF w/o medication	moderate DD, autistic behavior	hypotonia, hypertonia, obstructive sleep apnea, atypical behavior	unremarkable	unremarkable
Individual #9 GeneMatcher	15 years	M	c.8614C>T; p.Gln2872∗	5	no seizures	N/A	N/A	N/A	DD, mild ID, ADHD, SLD	obesity, scoliosis, tall stature	N/I	N/I
Individual #10 GeneMatcher	15 years	M	c.8614C>T; p.Gln2872∗	5	no seizures	N/A	N/A	N/A	DD, ADHD, SLD	scoliosis, tall stature	N/I	N/I
Individual #11 GeneMatcher	23 years	M	c.7126G>T; p.Glu2376∗	5	no seizures	N/A	N/A	N/A	DD, ID, FTT, SLD	growth failure	N/I	N/I
Individual 12 (Ye. T et al.17)	N/I	M	c.3322G>T; p.Glu1108∗	5	no seizures	FS 5–6 times/yr	toddler	N/I	normal	N/A	unremarkable	unremarkable
Individual 13 (Ye. T et al.17)	N/I	F	c.7351C>T; p.Gln2451∗	5	FS, FIAS	FS 3–4 times/yr, CPS 4 times/month since 3 years	infancy	N/I	normal	N/A	generalized spike-and-slow waves	unremarkable
Individual #14 BDB	3 years	F	c.5869G>A; p.Ala1957Thr	5	seizure, encephalopathy	N/I	N/I	N/I	global DD, dysphagia	cerebral visual impairment	N/I	N/I

AS; Absence seizure; BTC, bilateral tonic-clonic seizure; CFS, complex febrile; CPS, Complex partial seizures; GTCS, generalized tonic-clonic seizure; FIAS, focal impaired awareness seizures; FM, focal motor seizure; FS, febrile seizure; FSE, Febrile status epilepticus; FMS, focal motor seizure; DD, developmental delay; FTT, failure to thrive; ID, intellectual disability; LD, learning disability; SLDD, speech and language development delay; SLD, specific learning disability; SF, seizure free; N/A, not available; N/I, not informed; w/o, without.

Individuals with overlapping neurodevelopmental features carry de novo variants in BSN

We identified 12 additional individuals with confirmed de novo BSN variants absent from gnomAD: nine individuals through a collaborative network facilitated through GeneMatcher, one individual through a local biobank (BDB), and two individuals previously reported in the literature (Table 1).^17,24,26 The specific variants in these 12 individuals included nine PTVs and three missense variants, which were distributed across all the functional domains of BSN (Figure 1).

We identified overlapping seizure and developmental features in these additional individuals with de novo BSN variants, consistent with those observed in both individuals #1 and #2 (Table 1). Clinically, 10 out of 12 individuals presented with developmental delays, and epilepsy or febrile seizures were observed in five out of 12 individuals, with a median seizure onset of 16 months (range, 1–8 years). Seizure types varied; three had febrile seizures at onset, and two progressed to bilateral tonic-clonic seizures. Two individuals had epileptic encephalopathy or atypical absence seizures. At the most recent clinical follow-up, three individuals had achieved seizure freedom for at least a year, with a median duration of 6 years (range, 1–10 years). Seizure freedom was typically achieved by a median age of 4.2 years (range, 1–4 years). Three individuals continued to have active epilepsy during the study period, with one individual experiencing seizure recurrence after 8 months of seizure freedom.

Additional clinical features associated with de novo variants in BSN included hypotonia (4/12) and growth abnormalities (3/12), encompassing both growth failure and tall stature. Three out of 12 individuals for whom brain imaging was available had non-specific findings, including suspicion of focal cortical dysplasia (individual #3), haziness of the gray-white matter interface (individual #4), and abnormal cerebral white matter morphology (individual #5). Notably, of the two individuals identified through our neurogenetics clinic, individual #1 also exhibited mild cerebellar atrophy, a feature consistent with findings observed in the additional individuals as described.

Rare BSN PTVs show variable expressivity and incomplete penetrance

To investigate a potential gene-disease relationship, we analyzed the phenotypes of individuals with rare BSN variants absent from gnomAD that were identified via GeneMatcher (n = 5) and additional biobank databases (n = 10, CAG and PMBB; Table S1).^24,26 In total, 15 individuals were found to have PTVs (Figure 1). Of the 15 individuals with PTVs, 13 out of 15 individuals had variants of unknown inheritance, while two out of 15 individuals had paternally inherited variants (Table S1). Individual #19, who inherited a frameshift variant from his father, had multiple febrile seizures and bilateral tonic-clonic seizures, whereas the father only had a single febrile seizure with no other neurological symptoms. Individual #18 had speech and language delays and hyperactivity; this individual inherited a nonsense variant from his father, who did not have seizures or other neurological features. Inheritance of BSN variants in both individuals suggests potential incomplete penetrance and variable expressivity.

When comparing individuals with PTVs of known or unknown inheritance, phenotypic features overlapped with those seen in individuals with confirmed de novo BSN variants, such as delayed speech and language development (4/15) and specific learning disabilities (3/15). Seizures were present in five out of 15 individuals, with febrile seizures as the initial presentation in three out of five individuals.

Among the seven adults with BSN PTVs identified through biobank databases, three out of seven individuals did not have neurological phenotypes recorded in their EMRs. Of the four out of seven individuals with neurological phenotypes in the EMR, a single individual (individual #29) had a seizure-related ICD-10-CM code documented (Table S1), while the remaining individuals had sleep apnea (G47.33), cerebral edema (G93.2), and abnormal movement (R25.2). This suggests that BSN-related phenotypes are comparatively mild in adulthood with incomplete penetrance.

Comparative phenotyping of BSN variants identifies age-related differences

In our combined cohort of 29 individuals, we annotated 455 HPO terms across 15 phenotypic categories (Tables 2 and S2; Figure S1), referred to as base terms. The most common base HPO terms were global developmental delay (HP:0001263; 45%), obesity (HP:0001513; 34%), specific learning disability (HP:0001328; 34%), and delayed speech and language development (HP:0000750; 27%). The median number of HPO terms assigned per individual was 13, with a range of 1–73 terms (Table 2). Through structured data harmonization and propagation, we derived 1,637 HPO terms across 616 distinct phenotypic categories, allowing for a comprehensive analysis of clinical manifestations associated with BSN variants (Table 2; Figure S1).¹⁹ The most common HPO terms after propagation were abnormality of mental function (HP:0001249; 69%) and neurodevelopmental abnormality (HP:0012759; 55%; Figure 2A; Table S3). Next, we compared three groups of individuals with BSN variants to assess whether inheritance and age impacted phenotypic expression (Figures 2B–2D), including (1) de novo variants (n = 14), (2) pediatric PTVs (n = 8), and (3) adult PTVs (n = 7). Both the de novo cohort and PTV pediatric cohorts exhibited more cognitive and seizure-related HPO terms compared to the PTV adult group. When comparing all children with de novo variants (missense and PTV variants) to adults with PTVs, this observation remained consistent (p = 0.049, Figure S3). The frequency of global developmental delay (HP:0001263, p = 6.21e−08) was notably higher in those in the de novo cohort (86%) compared to those with pediatric PTVs (50%), suggesting a potential association of developmental delays associated with de novo BSN variants, although recruitment bias cannot be ruled out. While both pediatric groups (de novo cohort and PTV pediatric cohort) displayed similar frequencies for disinhibition (HP:0000734, 36%), hyperactivity (HP:0000752, 36%), specific learning disability (HP:0001328, 43%), and seizures (HP:000125, 57%), certain traits showed notable differences. For instance, delayed speech and language development (HP:0000750) were more prevalent in the PTV pediatric cohort (50%, p = 0.004) compared to the de novo cohort (29%, p = 0.004). Additionally, atypical behavior (HP:0000708) was observed more frequently in the de novo cohort (71%, p = 0.004) than in those with PTVs (50%, p = 0.004). Specific learning disability (HP:0001328) was also slightly more common in the de novo cohort (50%) than in the PTV pediatric cohort (38%), but the difference was not significant (p = 0.12).

Table 2.

Cohort information on 29 individuals with rare BSN variants including de novo variants (n = 14)_and PTVs (n = 15)

Demographic information
Male	14/29 (48.2%)
Female	15/29 (51.8%)
Age distribution, median (range)
Age at assessment (n = 15)	10 years (0.8 months–85 years)
Seizure onset (n = 14)	2.2 years (0.0–36 years)
Seizure offset (n = 8)	9 months (1 month–42 years)
Not applicablea	14/29 (48.2%)
Phenotypic information, median (range)
Number of phenotypic terms per individual	13 terms (1–73 terms)
Number of phenotypic terms per individual after propagation of HPO terms	48 terms (8–224 terms)
Number of distinct phenotypic terms in cohort	274 terms
Number of distinct phenotypic terms in cohort after propagation of HPO terms	616 terms

^aLiterature and biobanks report with limited data.

Figure 2.

Comparison of phenotypic features in BSN cohorts reveals distinct trends by age and inheritance

(A) A radial plot showing phenotypic features in the overall cohort (n = 29). Radial lines reflect the frequency of specific terms within the cohort.

(B–D) (B) Radial graphs displaying phenotypic feature distribution across subgroups, categorized by age and inheritance: de novo cohort (n = 14, B), PTV pediatric cohort (n = 8, C), and PTV adult cohort (n = 7, D).

Certain seizure types were more common in the de novo cohort compared to the PTV pediatric cohort (Figure 2B; Table S3). Notably, focal-onset seizures (HP:0007359, 21%) and focal impaired awareness seizures (HP:0002384, 14%) were present in the de novo cohort but neither were reported in the PTV pediatric cohort. There were no differences between both cohorts in the frequency of febrile seizures (HP:0002373, 25%) and bilateral tonic-clonic seizures (HP:0002069, 38%).

Notably, obesity (HP:0001513) was more common in individuals with pediatric PTVs (38%) compared to the de novo cohort (14%, p = 0.0001), and this observation remained consistent when comparing all pediatric de novo variants to the PTV adult cohort (p = 0.028, Figure S3). Furthermore, in the PTV adult cohort, 72% of individuals exhibited obesity-related features. Information on adults with PTVs in BSN was collected from PMBB, an EMR-linked biobank, which often captures more common medical conditions. Consequently, the PTV adult cohort showed a higher frequency of HPO terms related to common medical conditions in adults, such as abnormal cardiovascular system physiology (HP:0011025, 86%) and abnormality of the respiratory system (HP:0002086, 72%). In contrast, we did not identify a substantial frequency of high-level HPO terms indicative of neurological conditions (Figure S1; Table S3). Only four out of seven (57%) individuals in the PTV adult cohort were found to have an abnormality of the nervous system (HP:0000707).

Association analysis reveals unique phenotypic features in BSN-related disorders

We reconstructed the clinical presentation of BSN-related disorders using 675,109 HPO terms in 14,895 probands with developmental and epileptic encephalopathies (DEEs) and NDDs derived from various data sources including EGRP, Deciphering Developmental Disorders (DDD), and Epi4k (dbGaP) (Figure 3).²⁹ To identify phenotypic features associated with BSN, we performed an association analysis using Fisher’s exact test comparing the frequency of HPO terms in all 29 individuals with variants in BSN to a larger cohort of 1,470 individuals with DEEs and 13,425 individuals with NDDs.²⁹ Prominent phenotypic features associated with BSN variant affected individuals included disinhibition (HP:0000741, p = 3.39e−17), fatigue (HP:0012378, p = 5.27e−14), hypercholesterolemia (HP:0003124, p = 4.01e−10), temper tantrums (HP:0025160, p = 2.18e−05), and febrile seizures (HP:0002373, p = 1.26e−06) as some of the most prominent phenotypes (Figure 3; Table S4). While hypercholesterolemia emerged as a notable feature in this analysis, this association may reflect the contribution of the PTV adult cohort rather than a defining characteristic of BSN-related disorders.

Figure 3.

Phenotypic association analysis identifies disinhibition and fatigue as key features of BSN-related disorders

Volcano plot depicting the frequency of HPO terms in the BSN cohort (red, n = 29) compared to a larger reference group (blue, n = 14,893). Red dots represent terms with odds ratio (OR) >0.5 and p < 0.05, indicating significant phenotypic associations in the BSN cohort, while blue dots represent terms with a lower association in the reference cohort. Dot size reflects term frequency within the respective group.

Phenotypic similarity analysis supports a shared BSN gene-phenotype signature

Following the identification of distinct phenotypic features in individuals with BSN variants through association analysis, we sought to evaluate whether the phenotypic terms linked to individuals carrying BSN variants were sufficiently distinct to establish a discrete gene-specific phenotypic signature. We performed a formal phenotypic similarity analysis to assess the extent of clinical relatedness among individuals with de novo BSN variants compared to those with de novo variants in 256 other NDD-related genes (Figure 4).²⁹ In brief, a phenotypic similarity analysis assesses whether clinical features observed in a subset of individuals are more related than expected by chance within a given cohort. This analysis allowed us to compare the statistical evidence for BSN based on phenotypic similarity with the genetic evidence derived from the relative frequency of de novo variants, which compares the observed versus expected number of de novo variants within a given gene.³² Individuals with de novo BSN variants demonstrated a significant degree of phenotypic similarity (p = 0.00014), suggesting a strong gene-phenotype relationship and consistent phenotypic expression (Figure 4). This indicates that individuals carrying BSN variants had phenotypic features that are more similar than expected by chance, supporting the hypothesis of a phenotypic signature specific to BSN.

Figure 4.

BSN-related disorders have significant phenotypic resemblance in a comparison analysis of genetic and phenotypic evidence across 256 genetic etiologies implicated in NDDs

Scatterplot comparing genetic and phenotypic evidence for de novo variants in BSN (red, n = 14, p = 0.00014) against 257 genetic etiologies. Each data point represents an individual gene, with point size indicating the number of individuals with de novo variants per gene. Dashed blue lines denote the significance threshold of −log10(0.05) for both axes, with genes above these thresholds shown in blue to denote statistical significance in either genetic or phenotypic evidence, while genes below the thresholds are shown in gray. Genetic evidence on the x axis reflects the statistical significance of observed de novo variants, calculated using denovolyzeR, while phenotypic evidence on the y axis represents phenotypic similarity scores generated by sim analysis (simmax), followed by permutation analysis to assess significance. This comparative approach highlights the alignment or divergence between genetic and phenotypic evidence across genes, identifying where one type of evidence deviates from the expected correlation.

Although the phenotypic similarity for individuals carrying de novo variants in BSN was significant, the median sim score was lower compared to established genetic etiologies for epilepsy and NDD, including SCN1A (MIM: 182389) and SCN2A (MIM: 182390), reflecting the variability observed in the phenotypic expression of BSN variants (Figure 4).^19,29

Phenotypic similarity for individuals carrying de novo BSN variants was lower than the phenotypic relatedness assessed by a formal phenotypic similarity analysis in 78 out of 256 other NDD-related conditions caused by de novo variants (Figure 4). This suggests that the majority of NDDs caused by de novo variants are more recognizable than phenotypes related to de novo BSN variants. In fact, the clinical relatedness of individuals carrying de novo BSN variants ranges only in the top 30% of all NDDs assessed through this analysis.

Longitudinal EMR data uncovers neurological and behavioral trajectories in BSN

In order to assess the clinical trajectory of individuals carrying BSN variants, we mapped available EMR data from 12 individuals across a total of 103 patient years. We included longitudinal data from five individuals in the pediatric range (birth to 18 years) and seven adults (18 years and above). This analysis allowed us to recapitulate the longitudinal disease history of BSN-related disorders over a median observation window of 12 years (Figure 5). The longitudinal analysis revealed that neurological phenotypes emerged at a median age of 2 years (range 2 months to 17 years). Seizures were documented in three individuals, starting at a median age of 1 year (Figure 5A). Febrile seizures occurred at 13 months in two individuals, followed by other seizure types, including bilateral tonic-clonic seizures between 8 and 11 years. A single individual ascertained through the PMBB had seizures at 48 years.

Figure 5.

The longitudinal trajectory of clinical features in 12 individuals with BSN variants highlights early neurological manifestations and broad phenotypic spectrum

Distribution of key clinical features (red) over time in 12 individuals with BSN PTVs (n = 10), and de novo PTV variants (n = 2), illustrating age-related progression of features such as epilepsy (A), neurodevelopmental delays (B), behavioral phenotypes (C), and obesity (D). Phenotypic categories were manually mapped to HPO from ICD/ICD-10 codes.

More than half of the cohort (9/12) had features of obesity, with the first recorded instance in EMR ranging widely from 10 to 80 years (Figure 5D). Aside from the single individual with seizures recorded in the EMR at 48 years, all other adults (n = 6) did not have phenotypes related to neurodevelopmental abnormalities across a cumulative time span of 61 patient years (Figure 5B).

Behavioral phenotypes identified through the longitudinal phenotype analysis included atypical behavior (HP:0000708) and autism spectrum disorder (HP:0000717), with ADHD (HP:0007018) diagnosed at ages 3 and 11 years in three individuals (Figure 5C). Neurological features were present in four out of five individuals with longitudinal clinical data between birth and 15 years, including speech delay (HP:0000750), global developmental delay (HP:0001263), and specific learning disabilities (HP:0001328). Seizures were present in 3 out of 12 individuals with longitudinal clinical data available in the pediatric range (birth to 18 years) and between 41 and 51 years in adult individuals (Figures 5A and 5B). Obesity emerged as a common phenotype across both pediatric and adult individuals, further emphasizing the broad phenotypic spectrum associated with BSN-related disorders (Figure 5D).

Overall, our findings suggest that features associated with BSN-related disorders, such as febrile seizures and behavioral abnormalities, emerge during early childhood. These phenotypes diminish in frequency during adolescence and adulthood.

Discussion

In our study, we identify BSN, encoding the presynaptic protein Bassoon, as a gene associated with NDDs and epilepsy, utilizing a multi-faceted approach that combines detailed phenotypic curation through the HPO, assessment of phenotypic similarity through computational approaches across 14,893 individuals with epilepsy and NDDs, and longitudinal phenotyping analysis across 103 patient years. Phenotypic data were gathered maximizing available resources, including disease-specific cohorts focused on epilepsy and NDDs as well as data derived from large biobanks spanning the age spectrum. This strategy allowed us to refine the phenotypic signature of BSN variants and provide new insights into the genetic etiology across different age groups and clinical settings.^20,22,23

We first identified two individuals with de novo frameshift variants in BSN, both of whom presented with febrile seizures early in life, which later evolved into more complex seizure types such as bilateral tonic-clonic seizures and absence seizures. Behavioral features, including ADHD and autism spectrum disorder, became evident in later childhood. Expanding our analysis, we identified 12 additional individuals with de novo BSN variants through collaborative efforts facilitated by GeneMatcher, biobank data, and prior literature.¹⁷ In total, 57% of these individuals exhibited epilepsy, with 83% experiencing febrile seizures and several reporting multiple seizure types, such as bilateral tonic-clonic seizures (14%) and atypical absence seizures (14%).

Our findings parallel those from other genetic etiologies related to NDDs, such as SCN1A and STX1B (MIM: 601485), which exhibit a broad spectrum of seizure phenotypes, often beginning with febrile seizures in early childhood.^33,34,35 This observation reinforces the need to consider BSN within the broader context of genetic etiologies related to childhood epilepsies, as our study included several individuals who transitioned from febrile to generalized seizures during adolescence (Figure 5). These results provide further evidence for the role of BSN in seizure-related NDDs.

The identification of PTVs in BSN suggests haploinsufficiency as a likely mechanism, consistent with other genetic etiologies related to epilepsy and NDDs. Insights from Bsn-knockout mouse models strengthen this hypothesis,¹³ emphasizing the role of BSN in maintaining synaptic function and regulating hyperactivity of neuronal networks that may result in seizures.^14,36,37,38 Homozygous Bsn-knockout mice develop spontaneous seizures, underscoring the importance of BSN in regulating normal synaptic activity.¹² Furthermore, constitutive Bsn mutants and GABAergic neuron-specific knockouts (Bsn^Dlx5/6cKO) exhibit severe epilepsy, reinforcing the pathogenic link between Bsn disruption and epilepsy.³⁹ Additionally, the presence of both missense and PTVs distributed across the gene suggests a broader disruption of protein function that may variably affect synaptic processes (Figure 1).

Our cohort analysis, which included individuals with inherited BSN PTVs, provided key insights into the variability of phenotypic expression associated with this gene. Notably, 85% of the adults with BSN PTVs were asymptomatic or only had mild neurodevelopmental phenotypes, contrasting with the more obvious presentations in pediatric individuals. This incomplete penetrance and variable expressivity have been observed in other genetic etiologies, such as DEPDC5 (MIM: 614191), NPRL3 (MIM: 600928), and PRRT2 (MIM: 614386), which also show variability in clinical presentations and a relatively high proportion of asymptomatic individuals.^{40,41,42,43,44} Importantly, the differences observed between pediatric and adult presentations could be influenced by cohort ascertainment bias, as pediatric cohorts often focus on disease-specific phenotypes, whereas broader genetic studies include more diverse populations. By including both community-based ascertainment of individuals with de novo variants through GeneMatcher as well as inclusion of individuals with variants in large pediatric and adult biorepositories, we believe that we have overcome such a recruitment bias and present a holistic view of the phenotypic consequences of disruptive BSN variants.

Our approach allowed us to obtain a larger overview of associated phenotypes in the 29 individuals carrying rare BSN variants harmonizing phenotypic data through the HPO ontology. This approach allowed us to identify both specific features in a relevant subset of individuals, such as febrile seizures (HP:0002373, 25%) and maladaptive behavior (HP:5200241, 35%), as well as more generalized, higher-level terms present in the majority of individuals, such as abnormality of mental function (HP:0001249, 69%) and global developmental delay (HP:0001263, 55%). The use of the HPO framework enabled us to standardize phenotypic descriptions across various cohorts, which is crucial for comparing phenotypic data in genetic studies. Our data showed that neurodevelopmental abnormalities (HP:0012759, 86%) and atypical behavior (HP:0000708, 71%) were highly prevalent in individuals with de novo BSN variants. These features were also present in individuals with PTVs, although with lower frequency, highlighting the variable expressivity of BSN variants.

Using the same HPO-based framework to compare the 611 phenotypic features in 14 individuals with BSN de novo variants to 674,767 phenotypic annotations in 14,907 individuals with DEE and NDD, we identified specific features associated with BSN-related disorders that include disinhibition, fatigue, and febrile seizures. Furthermore, a formal phenotypic similarity analysis supported the presence of a gene-specific phenotypic signature, emphasizing that clinical features linked to disruptive BSN variants are more similar than expected by chance. Identifying a gene-specific signature related to BSN is critical for future clinical and therapeutic studies, and this phenotypic profiling approach has provided valuable insights for genetic etiologies such as SCN2A and GRIN2A (MIM: 138253), where a combination of de novo variants and phenotypic clustering has helped refine the role in NDDs.^19,20,29,45 BSN demonstrated moderate phenotypic similarity, suggesting that, despite phenotypic variability, many individuals carrying de novo BSN variants have recognizable phenotypic features. The similarity scores for BSN were only higher than those seen for 179 out of 256 other NDD-related conditions, suggesting greater phenotypic variability than 70% of all other NDD, far removed from the prominent similarity seen in SCN1A-, AP2M1-, or DNM1-related conditions.²⁹ This phenotypic similarity approach provides complementary insight into gene-disease associations, supporting the link of disruptive variants in BSN to NDDs but also quantifying the variability in phenotypic expression compared to other genetic etiologies.

Finally, our longitudinal data for 12 individuals with rare BSN variants illustrated that, while BSN variants can lead to significant neurological manifestations in childhood, only a small subset of individuals had seizures and behavioral issues in adulthood (Figure 5). While our findings suggest that certain features, such as febrile seizures and behavioral abnormalities, tend to emerge in early childhood and may be less frequently documented in adulthood, the extent to which these features change over time remains uncertain. In particular, given that none of the individuals included in our study had an observation period that spanned both childhood and a significant part of adult life, our findings might reflect recruitment bias, with more mildly affected adults identified through large-scale biobanking. However, it is also possible that neurologic features ameliorate over time, making them less prevalent within the adult cohort.

In our study, we acknowledge the uncertainty surrounding obesity as a definitive feature of BSN. Obesity was observed in a notable proportion of our cohort, particularly among individuals with biobank-identified PTVs (Figure 2). Among one of the pediatric sub-cohorts (CAG), three individuals had obesity. However, these individuals were identified through a dedicated recruitment as part of an obesity research study (Figure S2). In contrast, only one individual in the de novo cohort had obesity, and the frequency of 71% in the PMBB adult cohort may reflect broader population trends rather than disease-specific associations. The mechanism underlying the obesity in individuals with BSN de novo variants is unclear, and it can potentially be due to appetite regulation or metabolic issues. Prior GWAS-based studies have implicated BSN PTVs in severe adult-onset obesity, type 2 diabetes, and fatty liver disease, and have theorized that dysregulation in neurodevelopment, neurogenesis, and neuronal oxidative phosphorylation lead to increased appetite drive, thereby creating the link between BSN and obesity.^46,47 Our study recapitulates these findings by identifying febrile seizures as a significant phenotypic feature and noting the occurrence of obesity in subsets of our cohort.⁴⁸ These observations suggest a complex relationship between BSN variants and metabolic as well as neurodevelopmental phenotypes. Further studies are needed to determine whether these associations reflect direct effects of BSN disruption or cohort-specific biases, highlighting the need for integrating diverse datasets to better define the phenotypic spectrum and broader effects of BSN variants.

Our analysis highlights phenotypic patterns shaped by the various cohorts examined in our study, underscoring how cohort selection can influence the clinical features reported in genetic studies. By harmonizing data across multiple datasets, we strengthen our understanding of BSN and illustrate the importance of utilizing a wide range of study cohorts in genetic research. This unique approach enabled us to identify milder presentations of the condition that might otherwise go undetected, further highlighting the significance of integrating findings across varied populations.

A notable challenge in interpreting our findings is the distinct phenotypic presentations observed between adult and pediatric cohorts. This heterogeneity raises concerns about potential confounding factors. However, we would like to emphasize that the observed differences between cohorts actually provide important insights into the variable penetrance and expressivity of disruptive BSN variants. The variability in presentation suggests that our study outlines the extreme phenotypic presentations of rare BSN variants, ranging from unaffected adults to severe NDDs with early-onset epilepsy. This variability of clinical presentations depending on recruitment strategy and study cohort corroborates findings in other genetic etiologies, such as those involving NPRL3, DEPDC5, PRRT2, and KCNQ2, which highlight the challenges of delineating the full phenotypic range in variably penetrant genes.^{40,41,43,44,49}

In summary, our findings position BSN as a candidate gene for NDDs, demonstrating the critical interplay between genetic variants and their phenotypic manifestations. The wide range of clinical features associated with BSN variants delineate a distinct class of synaptic disorder that contrasts the relative homogeneous condition of other genetic etiologies linked to presynaptic function. These findings provide insight into the pathophysiology of NDDs and underscore the necessity for in-depth phenotypic studies to inform relevant outcomes in gene-specific therapeutic strategies.

Data and code availability

Primary data for this analysis are available in the supplemental information. Computer code for all analyses is available at https://github.com/staguzman/BSN/.

Consortia

The members of PMBB are Daniel J. Rader, MD; Marylyn D. Ritchie, PhD; JoEllen Weaver, MPH; Giorgio Sirugo, MD, PhD; Afiya Poindexter; Yi-An Ko, PhD; Kyle P. Nerz; Meghan Livingstone; Fred Vadivieso; Stephanie DerOhannessian; Teo Tran; Julia Stephanowski; Salma Santos; Ned Haubein, PhD; Joseph Dunn; Anurag Verma; PhD; Colleen Morse Kripke, MS, DPT, MSA; Marjorie Risman, MS; Renae Judy, BS; Colin Wollack, MS; Shefali S. Verma, PhD; Scott Damrauer, MD; Yuki Bradford, MS; Scott Dudek, MS; Theodore Drivas, MD, PhD.

The members of CHOP BDB are Stacy Woyciechowski, MS; J. William Gaynor, MD; Janine McNelia; Monica Molina, MS; Hongbo Xie, PhD; Teran Oung; Erika Diaz, MS; Donna Stephan, MD.

Acknowledgments

We thank the following who made this study possible: the biobank participants from CAG and PMBB, BDB, the research and clinical teams, and GeneMatcher. We acknowledge the PMBB for providing data and thank the patient-participants of Penn Medicine who consented to participate in this research program. We would also like to thank the PMBB team and Regeneron Genetics Center for providing genetic-variant data for analysis. The PMBB is approved under IRB protocol (#813913) and supported by Perelman School of Medicine at University of Pennsylvania, a gift from the Smilow family, and the National Center for Advancing Translational Sciences of the National Institutes of Health under CTSA award number UL1TR001878. The CHOP Birth Defects Biorepository (BDB) is supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant UL1TR001878. Individual 3 was identified as part of the Acute Care Genomics study research study, funded by the Australian Government’s Medical Research Future Fund (grant number GHFM76747). We acknowledge the use of data from the DDD project. The DDD study presents independent research commissioned by the Health Innovation Challenge Fund (grant number HICF-1009-003). This study makes use of DECIPHER, which is funded by Wellcome (grant number WT223718/Z/21/Z). See Nature⁵⁰ for full acknowledgment. I.H. is supported by the National Institute of Neurological Disorders and Stroke (R01 NS131512, R01 NS127830, and U24 NS120854) and the Hartwell Foundation (Individual Biomedical Research Award).

Figure 1 and Figure S1 were created with BioRender. https://BioRender.com/n50t398.

Author contributions

S.G.G., S.M.R., and I.H. contributed to the conceptualization of the study. Data curation was performed by S.G.G. and S.M.R. S.G.G. and S.G. conducted the analysis, while S.G.G. and S.G. developed the methods. The original draft was written by S.G.G., S.G., S.M.R., and I.H., and all authors (S.G.G., S.G., S.M.R., and I.H.) participated in reviewing and editing the manuscript.

Declaration of interests

The authors declare no competing interests.

Published: May 19, 2025

Web resources

• DECIPHER, http://www.deciphergenomics.org

• GeneDx ClinVar submission page, http://www.ncbi.nlm.nih.gov/clinvar/submitters/26957/