Novel HCFC1 variants identified in patients with ASD/ADHD and previously unreported structural brain malformations reveal the potential for phenotypic expansion

Abstract

Pathogenic variants in the HCFC1 gene, which encodes a transcriptional cofactor, cause cblX syndrome, intellectual disability, or partial focal epilepsy. HCFC1 encodes a multi-domain precursor protein that undergoes proteolytic cleavage to produce N- and C-terminal fragments that interact non-covalently. Pathogenic variants in the N-terminal kelch domain cause cblX syndrome with high penetrance and severity. However, pathogenic variants in the proteolytic cleavage domain cause focal epilepsy, and pathogenic variants in the C-terminal domain (i.e. acidic) are associated with non-syndromic intellectual disability. Previous studies have loosely associated some variants with neuropsychiatric disorders such as schizophrenia and autism spectrum disorder (ASD), but the gene has not been associated with other neuropsychiatric disorders such as attention deficit hyperactivity disorder (ADHD). In this report, we describe the genotype-phenotype observations for 2 novel HCFC1 variants including one non-coding variant. The presence of ASD/ADHD in this case series indicates that neuropsychiatric features merit routine assessment when interpreting HCFC1 variants of unknown significance, pending larger studies and functional data.

1. Introduction

Pathogenic variants in the HCFC1 gene cause methylmalonic aciduria and homocystinuria, cblX type (cblX) MIM#309541 [1]. Patients present with multiple congenital anomalies characterized by defects in cobalamin (vitamin B12) metabolism, mild to moderate facial dysmorphia, intellectual disability (ID), intractable epilepsy, and failure to thrive. Recent studies have identified missense variants in the HCFC1 gene that do not cause cblX syndrome but are associated with non-syndromic X-linked intellectual disability (ID) (reviewed in [2]). More recently, missense variants were identified in a Chinese cohort of focal epilepsy in the absence of a syndrome [3]. Pathogenic variants that cause cblX are consistently associated with disease (high penetrance) and a recognizable clinical pattern with variable expressivity [1], [4], [5]. For other variant classes, the current data does not define penetrance or expressivity, which constrains variant interpretation.

HCFC1 encodes a multi-domain protein consisting of an N-terminal kelch protein interaction domain, a basic region, a series of HCFC1 pro repeats, an acidic domain, and carboxyl end fibronectin domains [2]. Pathogenic variants in the kelch protein interaction domain cause cblX syndrome [1], whereas variants in the basic, acidic, or fibronectin type III (Fn3) domains do not cause an MCA syndrome but do result in ID [2], [6]. Interestingly, variants in the proteolytic cleavage domain cause focal epilepsy without syndromic phenotypes [3], and a single non-coding variant upstream of the HCFC1 gene was associated with ID in the absence of a syndrome [7]. Thus, pathogenic variants can be located across the protein and in regulatory DNA regions resulting in diverse and sometimes overlapping phenotypes. Because of the large phenotypic spectrum and variable expressivity, it is difficult to determine the causality of newly identified and potentially pathogenic variants. Here we identified 2 new variants in the HCFC1 gene through clinical genetic testing. Detailed phenotypic data was collected for each case. Consistent with previous literature, we observed a considerable phenotypic spectrum that spans structural brain malformations, autism spectrum disorder (ASD), speech delays, and attention deficit hyperactivity disorder (ADHD). The presence of ASD and ADHD phenotypes are not generally associated with variants in HCFC1; however, the variants we report are associated with these phenotypes, suggesting that the phenotypic spectrum for HCFC1 variants can be expanded to include ASD and ADHD phenotypes as well as some previously unreported structural brain malformations.

2. Case series

2.1. Patient 1

Patient 1 was a 5-year-old male with a history of speech delay and regression of speech at 2.5 years. Gross motor development was age appropriate. The subject was diagnosed with ASD at 4-years-old and was enrolled in speech and occupational therapy. Biochemical screening including methylmalonic acid (MMA) (0.18 nmol/ml-ref range ≤0.04 nmol/ml), urine organic acids (notable for very large quantities of 3-hydroxypropionic acid), acylcarnitine profile (no C3 elevation), and homocysteine (Hcy) (5.0 μmol/L: ref. range 4.0–14.0 μmol/L) were performed but non-diagnostic. According to his biochemical genetics report, levels of 3-hydroxypropionic acid were markedly elevated (no range provided) and not suggestive of a specific inborn error of metabolism and attributed to bacterial metabolism. Overnight electroencephalogram (EEG) to assess electrical status epilepticus in sleep was unremarkable. Chromosomal microarray analysis (CMA) and Fragile X analysis were negative, but CMA demonstrated two regions of SNP homozygosity that include 5.8 Mb at 5q31.3q32 (Chr5:142,980,826-148,815,204) and 10.6 Mb at 6p12.1q12 (Chr6:53,368,309-63,977,325). Given the large size of these regions, no specific candidate genes were identified prior to whole exome sequencing (WES). GeneScout was used to identify genes within these regions of homozygosity (ROH) but did not identify any genes, recessively inherited, associated with the phenotypes observed in the patient. Trio based WES identified a hemizygous variant in the HCFC1 gene c.4297G > A inherited from his unaffected mother. Although CMA identified regions of homozygosity, WES did not reveal any homozygous coding variants within these intervals. Therefore, no candidate genes within the ROH regions were implicated by exome analysis. There was no familial history of known neuropsychiatric diagnoses and no evidence of consanguinity. There were no complications during pregnancy and no reported prenatal exposures. The c.4297G > A missense variant causes the nonsynonymous p.Ala1433Thr protein residue change in the proteolytic cleavage domain (Fig. 1B). The p.Ala1433Thr variant has 3 submissions according to dbSNP (rs2065343555) with a minor allele frequency of 0.000004 (1/264690, TOPMED). The c.4297G > A variant was not present in gnomAD (v4.1.0) but a p.Ala1433Pro variant was reported in gnomAD. As shown in Fig. 1C, p.Ala1433 was conserved across species.

Fig. 1.

Conservation and location of variants of unknown significance (VUS). (A) Exon:Intron schematic of the HCFC1 gene. Shaded gray boxes indicate exons 1–26 with unshaded regions indicating untranslated regions at the 5 and 3′ end of the gene. The c.4498-10C > T VUS is located between exons 18 and 19. (B) Schematic of the HCFC1 protein (N to C-terminus) with indicated protein domains. The p.A1433T is in the proteolytic cleavage domain. Disorders associated with each domain are indicated below the diagram. Abbreviations: K = kelch; Fn3 = fibronectin type III domain. Yellow arrow heads indicate proteolytic cleavage sites and the gray shaded region at the C-terminal end is the nuclear localization signal. (C) Sequence conservation of the p.A1433T VUS across multiple species. Asterisks represent conserved amino acid residues in the region.

The HCFC1 c.4297G > A exonic variant was evaluated using the Franklin platform. Franklin's ensemble approach includes high specificity predictors such as REVEL, MetaLR, and AlphaMissense, all of which produced benign in silico predictions for this variant. Several individual tools (VARITY, SIFT, MutationTaster, DANN, and GenoCanyon) predicted deleterious effects; however, only VARITY exceeded a threshold that could support PP3, and ACMG guidelines require multiple concordant high-quality predictors to categorize PP3. Therefore, the overall computational prediction evidence supports BP4. According to Franklin, population frequency was extremely low consistent with a PM2 categorization. Franklin also applied PP2, since missense variation is an established disease mechanism for this gene [1], [4], [5] and HCFC1 is missense constrained with a gnomAD missense score of 5.6341. gnomAD missense Z-scores >3.09 indicate gene-level missense sensitivity. Together (PM2, PP2, BP4), the evidence does not support classification toward either likely pathogenic or likely benign. Thus, the appropriate classification for the HCFC1 c.4297G > A variant remains VUS (Table 1).

Table 1.

Classification for HCFC1 variant NM_005334.3: c.4297G > A (p. Ala1433Thr).

Evidence category	ACMG code	Justification
Population frequency	PM2	Variant is absent or extremely rare at a level consistent with prevalence, penetrance, and X-linked inheritance of HCFC1-related disease.
Functional Data	PP2	Franklin reports that HCFC1 is significantly missense constrained (gnomAD missense Z-score > 3.09; gene missense score = 5.6341), and missense variation is an established disease mechanism for HCFC1. PP2 is applied as a gene level missense sensitivity.
In silico predictions	BP4	Multiple ensemble/meta-predictors (REVEL = 0.12, BayesDel = −0.53, AlphaMissense = 0.169, MetaLR low) support a benign effect. Although individual tools show mixed results, ACMG prioritizes ensemble predictors for PP3/BP4 assignment, thus BP4 applies.
Phenotype specificity	Not applied	The phenotype is heterogeneous for HCFC1 disorders; there is no single clinical presentation that uniquely ties to one genetic etiology. PP4 is not used.
Final expected classification	VUS	The available evidence is insufficient to reach LP/LB. Classification remains VUS pending additional evidence.

2.2. Patient 2

Patient 2 was a male presenting to the genetics clinic at 7-years-old with a history of speech delay, hypotonia, attention deficit hyperactivity disorder (ADHD), anxiety, complex tics, recurrent fevers, and ID. The major developmental milestones were within normal limits and included walking and the onset of the first word at 12 months old. The patient had a history of speech articulation delay and was diagnosed with ASD at 8 years of age with an intellectual quotient (IQ) of 77. The patient had a history of eye rolling, behavioral outbursts, twisting neck motions, and abnormal hand movements; he had a normal EEG. Brain MRI identified partial agenesis of the corpus callosum and polymicrogyria with adjacent subcortical gray matter heterotopia involving the medial inferior left frontal lobe with additional subcortical gray matter heterotopia involving the superior left frontal gyrus. The patient's mother presented late to prenatal care with suspected environmental exposures during pregnancy, but these were not confirmed on newborn toxicology. Family history includes the patient's brother with sensory concerns and ASD, the patient's mother with a history of speech delay, and the patient's father with abnormal movements and tics. The patient was placed in the care of his adoptive mother (paternal grandmother) at 1 year of age. CMA and Fragile X analysis were negative. WES identified a variant in TNFRSF13B (c.260 T > A;p.I87N) and a variant in the noncoding region of HCFC1 (c.4498-10C > T; NM_005334.2). TNFRSF13B is associated with autosomal dominant or recessive common variable immune deficiency [8], a phenotype not reported in the patient, though there were recurrent fevers reported. The c.260 T > A variant has been deposited in dbSNP (rs72553877) with frequency of T = 0.000423 from TOPMED, T = 0.000348 from gnomAD (v4.1.0), and T = 0.000461 from the ExAC database. The variant is reported in ClinVar but with conflicting pathogenicity. According to ClinVar the TNGRSF13B variant has been reported in unaffected parents of sequenced index patients. The HCFC1 c.4498-10C > T; NM_005334.2 has an allele count of 1, an allele frequency of 8.63E-07 with zero hemizygotes in gnomAD (v4.1.0). The variant has been deposited into ClinVar as of May 21, 2024, without phenotypic data and not present in the dbSNP database at the time of this publication. The variant is located between exons 18 and 19 as shown in Fig. 1A. Interestingly, noncoding variants in HCFC1 have been associated with intellectual disability [7]. The patient was adopted and consequently hereditary studies could not be performed.

Franklin variant interpretation categorized the intronic variant (c.4498-10C > T) as PM2 based on absence/extreme rarity in population databases and BP4 based on computational splicing predictions, which indicate no predicted impact. This is appropriate for an intronic variant >2 bp from the canonical splice site under ACMG practice, which allows for BP4 when validated computational evidence supports a benign effect. In our own analysis using SpliceAI, scores were low (Acceptor loss Δ = 0.24, located 119 bp from the acceptor site), consistent with no strong effect on splicing, although mild or context dependent impact cannot be excluded. Notably, while most non-splice altering intronic variants are predicted to be benign, a previous report described a noncoding variant associated with intellectual disability through a transcriptional regulatory mechanism [7]. Therefore, although in silico analysis assigns a BP4, we note this possibility, although no functional or segregation evidence supports such an effect. Considering all available evidence, the overall ACMG classification is VUS (Table 2).

Table 2.

Classification for HCFC1 variant NM_005334.3: c. 4498-10C > T.

Evidence category	ACMG code	Justification
Population frequency	PM2	Variant is absent or extremely rare at a level consistent with prevalence, penetrance, and X-linked inheritance of HCFC1-related disease.
Functional Data	Not applied	This is a noncoding variant; PP2 (gene level missense sensitivity) and PM1 (critical domain) do not apply.
In silico predictions	BP4	Multiple meta-predictors indicate no impact on splicing (SpliceAI = 0.24; dbscSNV ADA = 0; dbscSNV RF = 0.04). ACMG allows for BP4 classification when validated computational evidence supports a benign effect.
Phenotype specificity	Not applied	The phenotype is heterogeneous for HCFC1 disorders; there is no single clinical presentation that uniquely ties to one genetic etiology. PP4 is not used.
Final expected classification	VUS	There are no segregation data, no functional data, and a lack of independent reports to classify as LP/LB.

3. Discussion

The primary disorders/phenotypes associated in the literature with pathogenic variants of HCFC1 are cblX syndrome and ID. cblX syndrome has multiple clinical phenotypes that span biochemical defects such as the accumulation of metabolites, craniofacial dysmorphia, intractable epilepsy, intellectual disability, and neurodevelopmental defects [1], [4], [7], [9]. cblX is 100% penetrant but there are varied severity (expressivity) and heterogeneity among phenotypes. Variants in the HCFC1 gene are also associated with isolated, non-syndromic ID [2], [7], [10] or focal epilepsy [3]. The penetrance of ID and focal epilepsy is unclear. For the present study, familial sequencing was absent for patient 2. Thus, we do not have additional information regarding inheritance. Based on previous studies, assessing penetrance of these variants is likely to be complex even when familial inheritance is completed, particularly when examining broader phenotypes such as ADHD, ASD, and other neuropsychiatric disorders. For example, in patient 1, the variant is a VUS; the mother is a heterozygous carrier and is asymptomatic. As this variant is not classified as pathogenic, the maternal phenotype does not inform penetrance, and the clinical significance remains uncertain pending functional or segregation evidence. The foundational studies establishing pathogenic variants that cause cblX syndrome describe a consistently severe neurodevelopmental phenotype [1], [4]. In contrast, the individuals in this series exhibit a broader range of neurodevelopmental and behavioral features, including ASD/ADHD without metabolic findings. Moreover, when evaluating the relevance of patient 2's neuroimaging studies in the context of known HCFC1-associated disease, we identified a cohort of males with missense variants with severe neurological involvement [1], including gyral cortical malformations requiring surgical resection, but agenesis of the corpus callosum and polymicrogyria were not specifically described. Similarly, Gérard and colleagues, described related males with mutation of HCFC1 and profound neurodevelopmental defects, but no agenesis of the corpus callosum and polymicrogyria were reported despite detailed clinical descriptions [4]. Thus, the specific combination of agenesis of the corpus callosum and polymicrogyria observed in Patient 2 of this series has not been previously documented. Our patient's neuroimaging may represent expansion of the known structural brain phenotypes or evidence for an additional, yet undetected genetic etiology, for which whole genome sequencing might be required. We also identified a variant in the TNFRSF13B gene. Variants in TNFRSF13B are associated with common variable deficiency (MIM#240500) and IgA deficiency (MIM#609529). These are primary immunodeficiency disorders whose phenotypes were not detected in patient 2. Common variable disorder can present later in life (20–30 years old) and therefore, we do not know the overarching effects of this variant. But since this gene is not associated with the structural or neurological phenotypes, we conclude that these cases, may provide additional clinical observations that may help refine the phenotypic spectrum associated with HCFC1 variants.

To our knowledge, there is one prior report linking HCFC1 variants with neuropsychiatric disorders [11] in which Piton and colleagues noted an association with schizophrenia and ASD. Here we report two new variants in the HCFC1 gene, both of which are present in subjects diagnosed with ASD or ADHD. Interestingly, one variant is in the non-coding region, and it is very difficult to predict its exact impact on protein function. Prior work has shown that other non-coding variants can act through transcriptional regulatory mechanisms, for example, disruption of the YY1 binding site led to HCFC1 over-expression and ID [7]. Overexpression of Hcfc1 in murine brain cells results in reduced neural stem cell growth and proliferation and abnormal neurite outgrowth [7]. This hypothesis is further supported by additional studies in zebrafish and mice, which show that knockdown Hcfc1 disrupts neural development and increases neural progenitor cell proliferation [7], [10], [12]. Given these observations, we report the non-coding allele as a VUS and emphasize that functional assays will be required to define its mechanism and clinical relevance, particularly since in silico prediction tools classify this variant as benign.

Patient 1 carries a missense variant located in the proteolytic cleavage domain of HCFC1 (amino acid residue 1433). Variants in this domain have been shown to disrupt proteolytic processing of HCFC1, leading to focal epilepsy in the absence of metabolic abnormalities [3]. Patient 1 of this series partially fits this category. For example, clinically, both MMA and Hcy were not elevated. In contrast with previously reported variants, the p.Ala1433Thr variant is associated with ASD in the absence of focal epilepsy. Although the individuals described by He and colleagues did not exhibit ASD, the presence of neurodevelopmental diagnoses in our patient highlights the broader potential for clinical variability observed among HCFC1 variants. Interestingly, cleavage is facilitated by O-GlcNAcylation (OGT) [13], and independent reports show that dysfunction of OGT is associated with ASD and schizophrenia [14]. Schizophrenia was previously associated with variants in HCFC1 [11], and our case series associates ASD with HCFC1 variants.

4. Conclusions

We describe two HCFC1 variants, one intronic and one exonic, each currently classified as VUS. In our case series, affected individuals carry neurodevelopmental diagnoses (ASD or ADHD) and one patient (patient 2) exhibits structural brain malformations not previously reported in the core HCFC1/cblX literature. Prior reports document severe neurological involvement and variable biochemical findings with cortical malformation in at least one case. These reports do not specifically report ASD, ADHD, or the structural phenotypes present in patient 2. Accordingly, our data are presented as descriptive clinical observations that complement existing reports. Publishing carefully phenotyped cases of VUSs in HCFC1 has practical value for the field. First, it broadens the phenotypic reference against which future variants are interpreted. Second, it enables case aggregation that can clarify whether specific neuropsychiatric or additional structural features co-occur with specific variant classes. Finally, we hope this case series and others like it will motivate the field to produce targeted functional assays, especially for noncoding alleles. In sum, we are unable to assign pathogenicity of these variants with current evidence. However, we used available evidence and ACMG-aligned documentation to inform future classification of HCFC1 VUSs.

CRediT authorship contribution statement

Victoria L. Castro: Writing – review & editing, Writing – original draft, Investigation, Data curation, Conceptualization. Carlos E. Palomino: Writing – review & editing, Formal analysis, Data curation. Jessica O'Shea: Writing – review & editing, Project administration, Methodology, Investigation, Formal analysis, Data curation. Elizabeth Ames: Writing – review & editing, Project administration, Methodology, Formal analysis, Data curation. Lauren Frazier: Writing – review & editing, Project administration, Methodology, Formal analysis, Conceptualization. Amanda Pritchard: Writing – review & editing, Resources, Investigation, Formal analysis, Data curation. Harry Lesmana: Writing – review & editing, Project administration, Methodology, Investigation, Formal analysis. German Rosas-Acosta: Writing – review & editing, Validation, Investigation, Data curation. Marlana J. Swann: Writing – review & editing, Methodology, Formal analysis. Anita M. Quintana: Writing – review & editing, Writing – original draft, Project administration, Methodology, Formal analysis, Conceptualization.

Ethics statement

This study was classified as exempt by the University of Texas El Paso's Institutional Review Board under protocol submission 1,925,515–2. Only de-identified phenotype and genotype data was shared. Subjects were consented in the clinical setting according to University of Michigan Institutional Policies. WES was provided by GeneDx, a CLIA certified laboratory. According to GeneDx protocols, quality metrics for WES mean depth 82×; 98.6% of targeted bases covered at >10×. All analyses were compared with the published human genome build GRCh37/UCSC hg19. Identified variants are reported as recommended by ACMG guidelines [15], and all reported variants are confirmed by appropriate orthogonal methods. Each variant was classified as a variant of unknown significance (VUS) by GeneDx.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

Authors report no conflicts of interest.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.