Long read sequencing and expression studies of AHDC1 deletions in Xia-Gibbs syndrome reveal a novel genetic regulatory mechanism

Abstract

Xia-Gibbs syndrome is a rare mendelian disorder characterized by Development Delay (DD), intellectual disability (ID), and hypotonia. Individuals with XGS typically harbor de novo protein-truncating mutations in the AT-Hook DNA binding motif containing 1 (AHDC1) gene, although some missense mutations can also cause XGS. Large de novo heterozygous deletions that encompass the AHDC1 gene have also been ascribed as diagnostic for the disorder, without substantial evidence to support their pathogenicity. We analyzed 19 individuals with large contiguous deletions involving AHDC1, along with other genes. One individual bore the smallest known contiguous AHDC1 deletion (~350 Kb), encompassing eight other genes within chr1p36.11 (Feline Gardner-Rasheed, IFI6, FAM76A, STX12, PPP1R8, THEMIS2, RPA2, SMPDL3B) and terminating within the first intron of AHDC1. The breakpoint junctions and phase of the deletion were identified using both short and long read sequencing (Oxford Nanopore). Quantification of RNA expression patterns in whole blood revealed that AHDC1 exhibited a mono-allelic expression pattern with no deficiency in overall AHDC1 expression levels, in contrast to the other deleted genes, which exhibited a 50% reduction in mRNA expression. These results suggest that AHDC1 expression in this individual is compensated by a novel regulatory mechanism and advances understanding of mutational and regulatory mechanisms in neurodevelopmental disorders.

1 |. INTRODUCTION

Neurodevelopmental disorders (NDDs) are a class of genetic maladies affecting 3% of children worldwide, that typically manifest in early infancy and are associated with dysfunction of normal brain development, leading to a spectrum of conditions including intellectual disability (ID), developmental delay (DD), autism spectrum disorder, or attention-deficit/hyperactivity disorder (ADHD) (Gilissen et al., 2014; Parenti et al., 2020). NDDs have heterogeneous clinical manifestations (Niemi et al., 2018; Tărlungeanu & Novarino, 2018) and have benefited tremendously from the implementation of next-generation sequencing that provides direct molecular diagnostics (Bowling et al., 2017; Harripaul et al., 2017; Martínez et al., 2017; Mitani et al., 2021; Palumbo et al., 2021; Smaili et al., 2020; Snoeijen-Schouwenaars et al., 2019; Vissers et al., 2017; Vrijenhoek et al., 2018). The mutational spectra associated with NDD genes include chromosomal aberrations, copy number variants (CNV), small insertions or deletions (indels) (<50 bp), and point mutations (protein-truncating and missense) (Cardoso et al., 2019). Despite advances in mutation discovery, the full repertoire of mutational mechanisms has not been fully elucidated for many disorders. As different possible mutation classes can characteristically impact the predicted functional biology of an affected gene, a full analysis of mutational spectra can greatly inform disease mechanisms. In particular, the opportunity to discern mutations that lead to loss-of-gene function (LoF) alleles, hypomorphic or null alleles resulting in haploinsufficiency, versus a dominant negative (antimorphic), or other gain-of-function (GoF) mechanisms (neomorphic or hypermorphic), can lead to entirely different models for function or even more importantly, therapeutics.

Xia-Gibbs syndrome (XGS; MIM# 615829) is an autosomal dominant NDD typically caused by de novo pathogenic truncating mutations in the AT-Hook DNA binding motif containing 1 (AHDC1) gene (Cardoso-Dos-Santos et al., 2020; Cheng et al., 2019; Díaz-Ordoñez et al., 2019; García-Acero & Acosta, 2017; Gumus, 2020; He et al., 2020; Jiang et al., 2018; Khayat, Li, et al., 2021; Mubungu et al., 2020; D. R. Murdock et al., 2019; Ritter et al., 2018; Xia et al., 2014; Yang et al., 2015, 2019). In the 8 years since the characterization of the syndrome, more than 300 individuals affected with XGS have been identified through concerted efforts worldwide. In addition, we established a secure XGS Registry in 2014, updated now with detailed mutation and phenotype information for more than 100 families who have provided clinical records and have consented for research (Jiang et al., 2018). Starting in early infancy, XGS individuals exhibit severe neurodevelopmental phenotypes, including delayed motor milestones, hypotonia, brain structure anomalies, dysmorphic features, sleep apnea, and speech delay, together with a variable presentation of secondary features (Xia et al., 2014). Affected individuals exhibit considerable phenotypic heterogeneity with respect to both penetrance and expressivity of clinical features, resulting in an overall complex presentation, hence all diagnoses to date have been dependent on DNA sequence-based approaches (Cardoso-Dos-Santos et al., 2020; Jiang et al., 2018; Khayat, Li, et al., 2021; Ritter et al., 2018; Xia et al., 2014).

The AHDC1 protein may participate in key interactions in the cell nucleus and homozygous deletion of the gene in model systems has led to the recent suggestion that AHDC1 acts as a regulator of early epithelial morphogenesis (Collier et al., 2022). The mechanism by which AHDC1 pathogenic variants result in XGS is unknown, however, with prior suggestions of both dominant-negative effects and haploinsufficiency (Table 1). The AHDC1 gene has a total of seven exons. This constitutes 5 noncoding 5-prime exons, followed by a single 4.9-kb coding exon, and a single downstream noncoding 3-prime exon (Xia et al., 2014). The single 4.9-kb coding exon encodes the 1603 amino acid protein (Xia et al., 2014). As AHDC1 contains a single coding exon, it is predicted that the mutant mRNA will escape the nonsense-mediated decay (NMD) surveillance mechanism and will result in the expression of the truncated versions of the protein (Lindeboom et al., 2016; Neu-Yilik et al., 2011). Further, Khayat, Li, et al. (2021) reported variable phenotypic effects, depending on the position of the truncating pathogenic variant, consistent with a dominant-negative effect. Most individuals clinically diagnosed with XGS harbor de novo protein-truncating mutations predicted to lead to truncated versions of the AHDC1 protein (Jiang et al., 2018; Khayat, Li, et al., 2021). In addition to the individuals with de novo protein-truncating mutations, at least 10 individuals have been reported as diagnosed with XGS based upon the presence of de novo missense mutations in AHDC1 (Gumus, 2020; Khayat, Hu, et al., 2021). Mapping these missense mutations to the protein sequence identifies at least two sensitive domains that overlap with regions of predicted functional motifs (Khayat, Hu, et al., 2021). The reported 10 pathogenic missense variants that result in XGS are predicted to result in full-length protein products and are also likely to act via a dominant-negative or other GoF mechanism (Khayat, Hu, et al., 2021).

TABLE 1.

Catalog of mutation class and literature references suggesting potential disease mechanisms driving XGS pathogenesis

Disease mechanism	Mutation class	References
Dominant-negative, GoF	Truncations	Jiang et al. (2018) and Khayat, Li, et al. (2021)
Dominant-negative, GoF	Missense	Gumus (2020) and Khayat, Hu, et al. (2021)
Haploinsufficiency	Large deletions	Focus of this paper

Abbreviations: GoF, gain-of-function; XGS, Xia-Gibbs syndrome.

In contrast, there have also been reports of individuals with large de novo deletions encompassing the entire AHDC1 locus who have been diagnosed with XGS (Park et al., 2017; Ritter et al., 2017; Wang et al., 2020). The observation of this class of mutation in XGS is important as it also suggests that AHDC1 haploinsufficiency may be a potential disease mechanism. In general, these individuals with large contiguous AHDC1 genomic deletions present with broad symptoms with some overlapping with XGS, and the diagnosis is rendered by the involvement of AHDC1 in the deletion interval. Most of these reported deletions span tens to thousands of kilobases, however, also involving multiple other linked gene loci. It is therefore possible that one or more of these linked loci may contribute as a “driver gene” to the XGS-like phenotypes, undermining support for an XGS model based upon haploinsufficiency of AHDC1. The central focus of this study is the investigation of large de novo deletions harboring AHDC1 to decipher whether haploinsufficiency could be a potential disease mechanism for XGS (Table 1).

Previously, the detailed analysis of such large genomic deletions, along with other structural variations (SVs), has been challenged by the short length of sequence reads (<300 bases), relative to the size of the genomic lesions. Limitations include a lack of sensitivity and specificity for accurate detection of SVs in regions where there are repetitive sequences. Low copy repeats or complex rearrangements and an inability to phase the haplotypes containing the deletions over long distances also complicate these studies (Mahmoud et al., 2019). Recent advances in genomic methods have now afforded the possibility of more precise SV characterization via the generation of longer sequence reads, enabling the resolution of SV alleles, including clinically relevant highly repetitive regions and complex human genome intervals (Chen et al., 2020; Chin et al., 2020), thereby enhancing our understanding of SVs, when compared to Illumina short-read data (Aganezov et al., 2020; Ho et al., 2020; Sedlazeck et al., 2018). Technologies include the HiFi read system of Pacific Biosciences and the long read (10 Kb–2 Mb) methods of Oxford Nanopore Technologies (ONT) (De Coster et al., 2021).

To further elucidate the primary disease mechanism by which AHDC1 mutations cause XGS and to address the central dilemma of haploinsufficiency versus other models as the mechanism of disease causation, we studied a total of 19 individuals reported with large contiguous AHDC1 deletions from the literature and clinical population databases. We also focused on one deletion in an individual (XGS28-P-R, Ritter et al., 2018), for whom detailed clinical data and research consent was available from the local XGS-Registry (Jiang et al., 2018). This latter individual exhibited a relatively mild phenotype and harbors the smallest known AHDC1 contiguous deletion of this class, thus affording a unique opportunity to conduct molecular studies that characterized the gene expression of AHDC1.

2 |. METHODS

2.1 |. Ethics, consent, and clinical data

This study and collection of patient clinical records were approved by the Baylor College of Medicine Institutional Review Board (H-39945). Written informed consent was obtained from participants before inclusion and sample collection in the study. Patient records are currently stored in an XGS Registry (Jiang et al., 2018), utilizing a Research Electronic Data Capture (RedCap) HIPAA-compliant environment (Harris et al., 2019). Parents of the proband (XGS28-P-R) have provided additional consent to participate in the XGS Registry and subsequent laboratory research. They have provided genetic reports as well as clinical surveys for analysis. The DECIPHER (https://www.deciphergenomics.org/about/stats) and DGV (http://dgv.tcag.ca/dgv/app/home) deletion cases were ascertained from their respective public databases (Firth et al., 2009; MacDonald et al., 2014).

2.2 |. Participant samples

We cataloged 19 AHDC1 genomic deletions, including two individuals with deletions reported in the literature (Park et al., 2017; Wang et al., 2020), nine deletions recorded in the DECIPHER database (Firth et al., 2009), seven deletions recorded in the DGV database (MacDonald et al., 2014) and one deletion in an individual (XGS28-P-R, Ritter et al., 2018) for whom detailed clinical data and research consent was available from the local XGS-Registry (Jiang et al., 2018).

2.3 |. Genomic DNA analysis

DNAs for this study were derived from individuals in the XGS Registry with previously reported AHDC1 truncation mutations and from individual XGS28-P-R, who is the focus of this study. Genomic DNAs were extracted from the whole blood of the proband using the Gentra Puregene kit (Qiagen; catalog no. 158667) according to the manufacturer’s instructions. Whole-genome sequencing (WGS) was performed on an Illumina NovaSeq to generate 150 bp paired-end reads, aligned using BWA-MEM to reference genome GRCh37/UCSC hg19 (NC_000001.10). Whole-genome long read sequencing was performed using the ONT sequencing platform. ONT base calling utilized Guppy version 4.3.4 + ecb2805 and subsequent analyses utilized PRINCESS version 1.0 (Mahmoud et al., 2021) with the “all” option to map reads and identify SVs, SNVs, and phase them against reference GRCh37/UCSC hg19 (NC_000001.10). We used the preset parameters for PRINCESS for running the ONT data analysis. PRINCESS starts by aligning reads using the appropriate parameters based on sequencing technology using Minimap2 version 2.17 (Li, 2018), followed by calling SVs using Sniffles version 1.12 (Sedlazeck et al., 2018), PRINCESS identified SNVs and indels using Clair3 (Luo et al., 2020), and phased the SNVs using WhatsHap version 0.18 (Patterson et al., 2015). PRINCESS phased the SVs using Princess tools.

2.4 |. RNA extraction and RT-PCR

Total RNA was extracted from blood cells using the RNeasy Mini kit (Qiagen; catalog no. 74104) according to the manufacturer’s protocol. RNA was digested using RNase-free DNase I to remove any contaminating genomic DNA. The RNA was quantified using a Qubit 3.0 Fluorometer. RNA quality was assessed to ensure proper quality before cDNA synthesis using RNA integrity number (RIN) metrics from the Agilent bioanalyzer. SuperScript^® III Reverse Transcriptase (Invitrogen, #18080-093) enzyme was used for cDNA conversion from 1 ug RNA input using a combination of random and oligo dT primers. The cDNA was quantified using the Qubit 3.0 Fluorometer to ensure high quality. RT-PCR was performed amplifying the exons 1, 2, 3, 6, and 7 junctions of AHDC1. Oligonucleotide primers were designed to target the exon–exon junctions using the Primer-Blast NCBI tool for the following target regions.

2.5 |. Identification of expressed alleles by restriction enzyme digestion analysis and Sanger dideoxy sequencing

To evaluate allele-specific expression, RT-PCR products were analyzed following treatment with restriction endonucleases that targeted sites containing either AHDC1 mutations or polymorphic sites that either created or ablated the enzyme recognition sites. Digested amplicon products were visualized by agarose gel electrophoresis. The cDNA amplicons used as templates for restriction digestion were also analyzed by Sanger sequencing to confirm results.

2.6 |. Estimation of gene expression for the heterozygous gene deletions using RNA sequencing (RNA-seq)

Total RNA was isolated from patient blood using the RNeasy Mini Kit (Qiagen; catalog no. 74104) according to the manufacturer’s protocol.RNA was digested using RNase-free DNase I to remove any contaminating genomic DNA and quality was assessed by the RIN using the Agilent 2100 Bioanalyzer. Library preparation was performed using Illumina TruSeq Stranded Total RNA with Ribo-Zero Globin/Illumina TruSeq Stranded mRNA (using poly-A selection). Approximately 100 million reads were generated per sample on the Illumina Novaseq. The sequencing reads were demultiplexed and the fastq files were aligned to the hg19 reference genome (NC_000001.10) using the STAR aligner (v. 2.5.3a) (Dobin et al., 2013). The duplicate marking was performed using Picard Tools (v. 2.18.4), RNA-SeQC v1.1.8. was used for quantifying postalignment quality metrics.

2.7 |. Quantification of gene expression using quantitative PCR (qPCR)

To quantify the expression of AHDC1 and other surrounding deleted genes, a real-time qPCR assay was performed using the 2X Kapa SYBR Green PCR Master Mix (KAPA SYBR^® qPCR Kit; catalog no. KK4601) in accordance with the manufacturer’s instructions. qPCR reactions were performed in 20 μl reactions in triplicate, for each sample. Each 20 μl reaction contains 10 μl of 2X KAPA SYBR^® FAST qPCR Master Mix Universal, 4 μl of 10 μM of each forward and reverse primer and 50 ng cDNA. Melt curve analysis was performed to check for the presence of primer dimers and the specificity of the reaction. A relative quantification approach using the 2^−ΔΔCt method was performed to quantify the expression levels of targets. (Livak & Schmittgen, 2001). The relative quantification of AHDC1 in the parent, proband, and wild-type controls was calculated and then normalized to the internal reference gene. Using this approach, the fold-change differences were calculated between the parent, proband, and wild-type controls, and the relative levels of AHDC1 expression were determined.

2.8 |. Sensitive quantification of gene expression using droplet digital PCR (ddPCR)

ddPCR was performed using the QX200 ddPCR EvaGreen SuperMix (Bio-Rad; catalog no. 1864034) in a QX200 ddPCR System (Bio-Rad; catalog no. 1864001). The primers used for RT-PCR were adopted for ddPCR. PCR conditions include the following parameters: 95°C for 5 min, 35 amplification cycles of 95°C for 10s, 60°C for 30s, 72 for 1 min and finally hold at 4°C using the BIO-RAD CFX96 Touch Real-Time PCR Detection System (Bio-Rad). The droplet reader Q × 200 Droplet Reader (Bio-Rad; catalog no. 1864003) was used to read the droplets. The QuantaSoft Analysis Pro Software (v1.0; Bio-Rad) was used for data analysis and quantification of the template concentration, according to the manufacturer’s instructions.

3 |. RESULTS

3.1 |. AHDC1 deletions from databases and literature

We first cataloged the 19 AHDC1 contiguous deletion cases reported to date, including nine recorded in the DECIPHER database (Figure 1, orange) (Firth et al., 2009), seven reports in the DGV database (Figure 1, green) (MacDonald et al., 2014), two cases described in the literature (Figure 1, blue) (Park et al., 2017; Wang et al., 2020) and one housed in the internal XGS Registry (Jiang et al., 2018). The majority of the deletions reported in the publicly accessible databases (DECIPHER and DGV) involve multiple flanking genes (Supporting Information: Table S1) (Firth et al., 2009; MacDonald et al., 2014). These flanking genes, within these large genomic deletions, were analyzed to determine their possible involvement in Mendelian disease. Eight genes (EPB41, FCN3, KDF1, MECR, NROB2, PIGV, SLC9A1, TMEM222) were reportedly associated with a disease phenotype in OMIM (1998; https://omim.org/) (Guissart, et al., 2015; Heimer et al., 2016; Liu et al., 2021; McKusick, 1998; Munthe-Fog et al., 2009). We also noted genes with a high intolerance to mutations, as assessed using the pLI scoring metric in the gnomAD database (https://gnomad.broadinstitute.org/) (Karczewski et al., 2020). Those with pLI scores (>0.7) include SLC9A1(107310), WASF2 (605875), AHDC1 (615790), STX12 (606892), PPP1R8 (602636), EYA3 (601655), RCC1 (179710), TAF12 (600773), GMEB1 (604409), YTHDF2 (610640) and SRSF4 (601940). In contrast, the gene with moderate pLI scores (≥0.5 and <0.7) include RPA2 (179836), and genes with low pLI scores (<0.5) include Feline Gardner-Rasheed (FGR) (164940) and PTPRU (602454) (Supporting Information: Table S1).

FIGURE 1.

Schematic representation of the location of the large contiguous AHDC1 deletions. The figure shows the AHDC1 contiguous deletion cases reported to date in the literature, clinical databases (DECIPHER, DGV) and one housed in the internal XGS Registry for whom detailed clinical data and research consent was available (Firth et al., 2009; Jiang et al., 2018; MacDonald et al., 2014). AHDC1, AT-Hook DNA binding motif containing 1; XGS, Xia-Gibbs syndrome.

There were no obvious correlations between the extent of the recorded deletions and the reported phenotypes among the nine entries that were reported in DECIPHER, which involved from 20 to 60 genes per deletion (Table 2). One DECIPHER entry (#274212), was of particular initial interest as it would represent the only known deletion that was solely contained within AHDC1. However, the referring physician for this individual indicated that the phenotype was reportedly not in concordance with the typical clinical presentation as seen in XGS and further, the deletion was undetectable by follow-up laboratory molecular methods, and the initial observation was thus determined to be artifactual.

TABLE 2.

Phenotypic spectrum and genotype information for individuals with AHDC1 deletions

Patient information	XGS28-P-R	Park et al. (2017)	Wang et al. (2020) (Patient 1)	Wang et al. (2020) (Patient 2)	Decipher patient: 1803	Decipher patient: 251385
Variant	Deletion	Deletion	Deletion	Duplication	Deletion	Deletion
Size	357 KB	1 MB	575 KB	491KB	2.45 MB	1.97 MB
Chromosome	1	1	1	1	1	1
Genomic location	27,589,832– 27,949,621  (GRCh37/hgl9)	27,877,5682–8,882,184  (GRCh37/hgl9)	27,376,659–27,951,704  (GRCh37/hgl9)	27,712,904–28,203,653  (GRCh37/hgl9)	27,358,936–29,807,278  (GRCh37/hgl9)	27,237,958–29,205,507  (GRCh37/hgl9)
Inheritance	De novo heterozygous	De novo heterozygous	De novo heterozygous	De novo heterozygous	De novo heterozygous	De novo heterozygous
Age of symptoms onset	8 months	12 months	8 years	2 years, 8 months	3 years	5 years
Gender	Female	Male	Male	Male	Female	Female
Comprehensive skills and language
Speech delay†	+	+	+	+	−	−
Autism diagnosis†	−	−	−	−	−	−
ADHD	−	−	−	+	−	−
Musculoskeletal
Delayed motor milestones†	+	+	+	+	−	−
Scoliosis†	−	−	−	−	−	−
Hypotonia†	−	+	+	−	+	+
Obstructive sleep apnea†	+	−	+	+	−	−
Growth deficiencies	−	+	−	−	−	−
Laryngomalacia†	−	−	+	+	−	−
Neurologic
Corpus callosum thinning	+	+	+	+	−	−
Seizures†	−	−	+	−	−	−
Intelllectual disability†	−	+	+	+	+	+
Ataxia†	−	−	−	+	−	−
Delayed myelination	−	−	+	+	−	−
Behavorial concerns	−	−	+	+	−	−
Head and neck
Dysmorphic facial features†	−	−	+	+	+	+
Microcephaly/macrocephaly	−	−	+	+	−	−
Hydrocephalus	−	−	−	+	−	−
Vision and Hearing
Wearing glasses or contacts	−	−	−	−	−	−
Strabismus†	−	−	−	−	−	−
Hearing loss	−	−	+	−	−	−
Other features
Hypertonia	+	−	−	−	−	−
Urinary incontinence	−	−	+	+	−	−
Insensltivity to pain	−	−	−	+	−	−
Short stature	−	−	+	−	−	−
Micrognathia	−	−	−	−	+	−
Teeth dysplasia	−	−	−	−	−	−
Other notable features	Lipoma, heart defects			Overgrowth, brain white matter dysplasia, urinary incontinence, drooling insensitivity to pain	Heart defects, micrognathia	Hip dislocation, abnormality of the skin
Patient information	Decipher patient: 258404	Decipher patient: 274096	Decipher patient: 274212	Decipher patient: 276526	Decipher patient: 303404	Decipher patient: 395510	Decipher patient: 400951
Variant	Deletion	Deletion	Deletion	Deletion	Deletion	Deletion	Deletion
Size	2.7 MB	658.29 KB	5.02 KB	1.05 MB	1.25 MB	20.68 MB	1.07 MB
Chromosome	1	1	1	1	1	1	1
Genomic location	27,786,502–30,630,253  (GRCh37/hgl9)	27,835,713–28,494,006  (GRCh37/hgl9)	27,873,727–27,878,743  (GRCh37/hg19)	27,780,057–28,827,017  (GRCh37/hg19)	27,589,560–28,838,774  (GRCh37/hg19)	6,965,734–27,915,858  (GRCh37/hg19)	27,333,076–28,401,804  (GRCh37/hg19)
Inheritance	De novo heterozygous	De novo heterozygous	Unknown heterozygous	De novo heterozygous	De novo heterozygous	De novo heterozygous	De novo heterozygous
Age of symptoms onset	1 year	18 years	2 years, 3 months	5 years	<1 year	1 year	<1 year
Gender	Female	Male	Male	Female	Female	Female	Female
Comprehensive skills and language
Speech delay†	−	−	−	−	+	−	−
Autism diagnosis†	−	−	−	−	+	−	−
ADHD	−	−	−	−	−	−	−
Musculoskeletal
Delayed motor milestones†	+	+	+	−	−	−	−
Scoliosis†	−	+	−	−	−	−	−
Hypotonia†	−	−	−	+	−	+	−
Obstructive sleep apnea†	−	−	−	−	−	−	−
Growth deficiencies	−	+	−	−	−	−	−
Laryngomalacia†	+	−	−	−	−	−	−
Neurologic
Corpus callosum thinning	−	−	−	−	−	−	−
Seizures†	−	+	+	−	−	−	−
Intellectual disability†	−	−	−	−	−	+	−
Ataxia†	−	−	−	−	−	−	−
Delayed myelination	−	−	−	−	−	−	−
Behavorial concerns	−	−	−	−	−	−	−
Head and neck
Dysmorphic facial features†	−	−	+	+	−	−	+
Microcephaly/macrocephaly	−	−	−	+	−	+	−
Hydrocephalus	−	−	−	−	+	−	−
Vision and Hearing
Wearing glasses or contacts	−	−	−	−	−	−	−
Strabismus†	−	−	−	−	−	−	−
Hearing loss	+	−	−	−	−	−	−
Other features
Hypertonia	−	−	−	−	−	−	−
Urinary incontinence	−	−	−	−	−	−	−
Insensltivity to pain	−	−	−	−	−	−	−
Short stature	+	+	−	−	+	+	−
Micrognathia	−	−	−	−	−	−	−
Teeth dysplasia	−	−	−	−	−	−	−
Other notable features	Plagiocephaly	Hip dislocation	Short palm, Polyphagia, 2–3 toe syndactyly.		Plagiocephaly, bilateral talipes equinovarus, congenital craniofacial dysostosis, hypothyroidism, left unicoronal synostosis, recurrent patellar dislocation, renal hypoplasia, right unilambdoid synostosis.	Abnormality of upper respiratory tract, depressed nasal bridge, generalized hirsutism, high palate, hypertelorism, hypoplastic philtrum, long philtrum, neuroblastoma, prominent occiput, protruding tongue, wide intermamillary distance.	Abnormality of the skin, cataract, delayed closure of the anterior fontanel, depressed nasal bridge, frontal bossing, long philtrum, posteriorly rotated ears, short nose.

Note: Phenotypic information and genotype is presented for individuals with AHDC1 deletions. Phenotypes are categorized into high-level categories: comprehensive skills and language, musculoskeletal, neurologic, head and neck, visión and hearing, and other features. XGS core features are marked with † and derived from Figure 2 (Khayat, Hu, et al., 2021).

Abbreviations: ADHD, attention deficit hyperactivity disorder; AHDC1, AT-hook DNA binding motif containing 1; NR, not reported; + feature present; − feature absent.

The deletions recorded in the DGV database were relatively smaller than those in DECIPHER. These are recorded as likely benign in the DGV database due to their inclusion in multiple population-based studies as control data sets. These data sets were used as part of case-control studies for assessing the role of CNVs in genetic disease (Table 3). The participants in these control data sets were stated as neurologically normal adults, that is, neurotypical adults, who are “disease-free” individuals included from various diverse populations who provided their informed consent.

TABLE 3.

AHDC1 copy number variation in the control population (DGV database)

Chr	Start	End	Size	Type	Subtype	Genes deleted	References
1	27772009	27945359	173351	CNV	Deletion	AHDC1, FGR, WASF2	Itsara et al. (2009)
1	27849300	27937456	88157	CNV	Deletion	AHDC1	Itsara et al. (2009)
1	27859036	27876482	17447	CNV	Deletion	AHDC1, FGR, IFI6	Shaikh et al. (2009)
1	27921423	27922378	956	CNV	Deletion	AHDC1	Conrad et al. (2010)
1	27883154	27980196	97043	CNV	Deletion	AHDC1, FGR	Teague et al. (2010)
1	27921465	27922763	1299	CNV	Duplication	AHDC1	1000 Genomes Project Consortium et al. (2010)
1	27905350	27910031	4682	CNV	Deletion	AHDC1	Mills et al. (2011)
1	27786334	27945359	159026	CNV	Deletion	AHDC1, FGR, WASF2	Cooper et al. (2011)
1	27849300	27958245	108946	CNV	Deletion	AHDC1, FGR	Cooper et al. (2011)
1	27895301	27943600	48300	CNV	Deletion	AHDC1, FGR	Dogan et al. (2014)
1	27913800	27916300	2501	CNV	Duplication	AHDC1	Alsmadi et al. (2014)
1	27887013	27955385	68373	CNV	Duplication	AHDC1, FGR	Coe et al. (2014)

Abbreviations: AHDC1, AT-hook DNA binding motif containing 1; CNV, copy number variants; FGR, Feline Gardner-Rasheed.

The two additional deletions reported in the literature indicate de novo mutations that have been ascribed to XGS due to overlap in the clinical presentation (Park et al., 2017; Wang et al., 2020). These genomic deletions each include additional genes along with the complete heterozygous deletion of AHDC1. A subset of these genes has been reported in OMIM as Mendelian disease genes (SLC9A1, FCN3, TMEM222).

3.2 |. AHDC1 deletion in the XGS registry

One individual (XGS28-P-R) within our internal XGS Registry (Jiang et al., 2018) presented with the smallest known AHDC1 deletion. XGS28-P-R is a 5-year-old female, molecularly diagnosed at 8 months of age with XGS, following global DD, hypoplastic corpus callosum, hypertonia and intramuscular lipoma. XGS28-P-R had a normal electroencephalogram and no indication of seizures, scoliosis, or laryngomalacia, which are some of the core features of XGS (Xia et al., 2014). The follow-up clinical survey reported recent developmental improvements with independent walking, climbing, and other motor movement activities. XGS28-P-R is adept at communication and language using many signs, follows many commands, and communicates without words. Overall, this individual’s clinical phenotype is mild when compared to other XGS individuals.

3.3 |. WGS delineates the precise deletion boundaries

The initial molecular diagnosis of XGS in proband XGS28-P-R was via a whole-genome single nucleotide polymorphism (SNP) array in a clinical genetics laboratory, which detected an interstitial large de novo deletion (1p36.11 → p35.3) spanning eight genes (AHDC1, FGR, G1P3, STX12, PPP1R8, THEMIS2, RPA2, and SMPDL3B) (Ritter et al., 2018). The array-predicted breakpoint coordinates fell within the 5ʹUTR of the AHDC1 gene, and on the other end, mapped within the SMPDL3B gene (Supporting Information: Figure S1). We performed WGS to delineate the precise deletion boundaries using the Illumina platform (Supporting Information: Figures S1 and S2). WGS revealed the expected de novo heterozygous deletion, spanning 357Kb in size (NC_000001.10 (hg19), chr1: 27,589,832-27,949,621) encompassing the eight flanking genes and terminating within the first intron of AHDC1. In comparison with WGS, the array coordinates varied by 30 bp on the left breakpoint and approximately 3 kb, on the right breakpoint.

3.4 |. Validation of breakpoints and phasing analysis using Oxford nanopore technology (ONT)

To further resolve the haplotype containing the deletion, we performed whole-genome long read DNA sequencing using ONT. The ONT sequence data yielded 26x coverage with 9111 bp average read length and a maximum of 1,389,036 bp. The long reads generated by the ONT platform confirmed the precise breakpoint sites identified by Illumina sequencing and also characterized other associated genomic variations in regions flanking the deletion site (Figure 2). A common SNP c.2145, C>A (NC_000001.10:g.27876482C>A) found within exon 6 of AHDC1, was heterozygous in individual XGS28-P-R and is located within a restriction endonuclease recognition site (Figure 3a). This heterozygous SNP facilitated subsequent experiments characterizing gene expression, as the ONT data established that the nonreference allele ablated the restriction enzyme recognition site. In contrast, the normal (nonmutant) haplotype harbored the reference allele at the SNP site that completed the sequence of the restriction site and was linked to the deletion (Figure 3, Supporting Information: Figure S2).

FIGURE 2.

Breakpoint junction analysis for ~375 Kb deletion in the proband (a) visualization of aligned Oxford Nanopore Technology (ONT) sequencing reads through the~375 Kb deletion by Integrative genomic visualization (IGV). (b) IGV visualization for the soft clipped bases (i.e., breakpoint junction sequences) detected by short-read whole-genome sequencing (WGS) by Illumina (top panel) and long read WGS by ONT (bottom panel). The IGV screen on the left is showing the breakpoint junction sequence at the end of the deletion (green arrow), and the right IGV screen is showing the breakpoint junction sequence at the start of the deletion (red arrow). (c) Schematic representation of the precise location of the deletion breakpoints which fall within the first intron of AHDC1 and 3rd intron of SMPDL3B gene. (d) Sequences at the junction of breakpoints. The proband sequence read (middle) was mapped to two regions (above and below) in the reference genome, that are located at the start (red) and the end (green) of the deletion. Breakpoint junction sequence included one base of microhomology (A highlighted in blue).

FIGURE 3.

Validation of deletion breakpoints and phasing analysis using Oxford Nanopore Technology. (a) Strategy for long read sequencing to obtain full coverage of the locus harboring the heterozygous deletion, additional 1 Mb coverage of flanking regions on both ends, and coverage of the heterozygous single nucleotide polymorphism in AHDC1 to aid with phasing the deletion. (b) Integrative genomic visualization software using zoomed out version shows the 357 Kb deletion and read aligning on respective alleles. The deletion is located on allele A (brown reads) using zoomed-in version.

3.5 |. Restriction enzyme digestion analysis identifies RNA expression patterns of AHDC1 alleles

The pattern of normal allelic expression at the AHDC1 locus was first analyzed utilizing RNA extracted from cells derived from healthy individuals. Restriction enzyme digestion of PCR amplified cDNAs was targeted to the site of the common single nucleotide variant (SNV) in exon 6 of AHDC1. The analysis of allelic expression in the normal cells revealed the expected normal biallelic pattern of RNA expression at the AHDC1 locus (Supporting Information: Figure S3).

3.6 |. Analysis of the expression of individual alleles in truncation cases establishes that AHDC1 escapes NMD

The AHDC1 gene has a single coding exon, and it is predicted that AHDC1 mutant mRNA will escape surveillance by the mechanism of NMD (Lindeboom et al., 2016; Neu-Yilik et al., 2011). We verified the lack of NMD using PCR amplification and restriction digestion analysis of cDNA, using RNA extracted from both normal cells and cells harboring truncation mutations within the coding exon of AHDC1. We evaluated two truncation mutations that recurred in the population of known XGS individuals: (NM_001371928.1: c.2373_2374del, p.Cys791TrpfsTer57) and (NM_001371928.1:c.2908C>T, p.Gln970Ter) (Supporting Information: Figures S3 and S4). Each mutation created a new restriction enzyme recognition site and therefore, the restriction enzyme digestion was expected to reveal novel allele-specific expression patterns. Equal representation of the expression of each allele was observed in each sample (Figures 4 and 5a).

FIGURE 4.

Analysis of allele-specific expression using restriction enzyme digestion in the truncation case (XGS1-P-R). (A) RT-PCR was performed for amplification of AHDC1 target regions in the proband (XGS1-P-R) with AHDC1 truncation mutation (NM_001371928.1:c.2373_2374del, p.Cys791TrpfsTer57). (a) The gel shows the cDNA amplicons of the expected size for the AHDC1 target regions. (b) The table shows the expected size of the PCR amplicons. (B) Schematic representation of the primer design and target sites. (C) Restriction digestion results in Truncation case (XGS1-P-R). (a) The gel shows the separation of the restriction digestion fragments and (b) table with the status for each sample respectively. Interpretation of the pattern of the restriction digestion for proband and control samples (Supporting Information: Figure S3).

FIGURE 5.

Analysis of allele-specific expression using restriction enzyme digestion in the truncation case (XGS11-P-R) and the deletion case (XGS28-P-R). (A) Analysis of allele-specific expression using restriction enzyme digestion in the truncation case (XGS11-P-R) with truncation mutation (NM_001371928.1:c.2908C>T, p.Gln970Ter). (a) The gel shows the separation of the restriction digestion fragments in the truncation case (XGS11-P-R). (b) The table shows the expected size of the PCR amplicons for each sample. Interpretation of the pattern of the restriction digestion for proband and control samples. (B) Analysis of allele-specific expression using restriction enzyme digestion in the deletion case (XGS28-P-R). (a) Gel shows the separation of the restriction digestion fragments (b) the table shows the expected size of the PCR amplicons for each sample. Interpretation of the pattern of the restriction digestion for proband and control samples (Supporting Information: Figures S4 and S5).

3.7 |. Analysis of allele-specific expression using restriction enzyme digestion in the deletion case (XGS28-P-R)

The individual (XGS28-P-R) harbored a deletion that involved only part of the AHDC1 locus, with uncertain effects on AHDC1 expression. To test the relative expression of AHDC1 alleles for each haplotype, RNA was isolated from the blood leukocytes of the proband and analyzed by RT-PCR targeted to exon 6 of the AHDC1 (Supporting Information: Figure S5). The amplified cDNA was treated with a Sma I restriction enzyme that recognized the heterozygous, polymorphic SNP site in exon 6. The RT-PCR products from the proband sample revealed the presence of a single uncut band, which confirms that only one allele carrying the C>A heterozygous SNP (lacking the restriction site) is being expressed in the patient blood (Figure 5b). Results were confirmed by the orthogonal method of Sanger sequencing of the cDNA amplicon (Supporting Information: Figure S6).

3.8 |. Quantification of gene expression of deleted genes

The individual (XGS28-P-R) represented an atypical presentation of XGS, with relatively mild symptoms. Hence, it was of further interest to precisely characterize the level of expression of the single active AHDC1 allele. To examine the residual AHDC1 expression levels in XGS28-P-R, we quantified gene dosage using RNA-seq, qPCR and ddPCR at the AHDC1 locus, along with the profiles of other genes within this large deletion.

3.9 |. RNAseq and qPCR assay identify no deficiency in AHDC1 expression levels in the proband

We first applied RNAseq to characterize the gene expression across the site of the deletion. These data were not informative for AHDC1, due to the inherently low AHDC1 expression levels (Supporting Information: Figure S7), but showed the expected ~50% expression of the flanking genes within the deletion (Figure 6a). To enable a more sensitive measurement of the AHDC1 expression, we first utilized real-time qPCR using primers spanning exon junctions and across the length of the gene (Figure 6b). Surprisingly, these data showed that the proband had equivalent levels of expression at the AHDC1 locus when compared to the parent and the wild-type controls (Supporting Information: Figures S8a and S8b). In contrast, three adjacent genes (FGR, THEMIS2, STX12) that were also hemizygous as a result of the deletion had the expected ~50% of normal expression (40%, 40%, 60%, respectively) (Supporting Information: Figure S8c and Table S2). As the observation of near-normal expression of the AHDC1 was unexpected, and because of the overall low level of expression of that gene in blood cells, we opted to pursue more sensitive detection methods to validate these findings.

FIGURE 6.

Expression analysis using RNAseq and ddPCR. (a) RNAseq analysis reported reduction in gene expression for genes heterozygously deleted in the proband. However, AHDC1 had low counts, so expression differences were inconclusive. (b) Schematic representation of AHDC1 gene structure and primer design strategy for qPCR and ddPCR assay. (C) ddPCR analysis for AHDC1 target regions reported similar gene expression levels in proband when compared to healthy control. ddPCR gene expression quantification for AHDC1 and controls were in concordance with qPCR and RNAseq results, reporting that there is no deficiency in AHDC1 blood expression. G2, G3 = positive controls for assay validation, expecting 50% decrease in expression from RNA-seq and qPCR. AHDC1, AT-Hook DNA binding motif containing 1; ddPCR, droplet digital PCR; qPCR, quantitative PCR.

3.10 |. Orthogonal validation using ddPCR provided accurate quantification of low abundance transcripts

ddPCR has increased sensitivity for quantifying low abundance transcripts and was performed for orthogonal validation of AHDC1 expression results from qPCR. The results of the ddPCR were concordant with the qPCR data for the normal control, proband XGS28-P-R, and the proband’s parent samples. Expression results were also assessed for a subset of the flanking deleted genes and yielded the expected 50% reduction in expression levels (Figure 6c). Overall, the ddPCR supported the finding of no deficiency in overall AHDC1 expression in proband XGS28-P-R, relative to normal controls. All the expressed AHDC1 RNA in individual XGS28-P-R was from the normal gene copy, and none was detected from the allele that was part of the contiguous AHDC1 gene deletion. These results suggested a pattern of AHDC1 regulation in XGS28-P-R supporting a model of genetic compensation as an underlying mechanism operating in this individual.

4 |. DISCUSSION

The primary mechanism by which AHDC1 pathogenic mutations cause XGS is unknown, with some prior evidence supporting dominant-negative, or GoF models (Khayat, Li, et al., 2021) and other evidence favoring a haploinsufficiency disease mechanism (Park et al., 2017; Wang et al., 2020). We experimentally confirmed that truncation mutations in the single coding exon of the AHDC1 gene result in the expression of a mutant mRNA that escapes NMD, consistent with a dominant-negative or GoF disease mechanism, mediated by the expression of the truncated protein forms (Khayat, Li, et al., 2021). When combined with the previous reports of disease in individuals with AHDC1 missense mutations, and the knowledge of phenotypic heterogeneity governed by the mutation position, these factors weigh heavily towards support of dominant-negative or GoF disease models (Jiang et al., 2018; Khayat, Hu, et al., 2021).

In further support of dominant-negative or GoF disease models, we found evidence that contradicted the alternative suggestions of a role for haploinsufficiency in this disorder. Examination of reports of XGS resulting from contiguous deletions encompassing AHDC1 did not substantiate those diagnoses. Although some of the evidence required to fully assess the impact of these reports from public databases was incomplete, examination of available phenotype data and analysis of the other genes in the region offered important clues, but failed to support the previous suggestions that large contiguous AHDC1 deletions cause XGS.

First, the seven reported deletions in the DGV database may have informed this study, but the lack of phenotypic descriptors precluded a full assessment of the consequences of mutation in AHDC1. The samples were from groups of “healthy controls,” suggesting AHDC1 deletion in these individuals is benign, however that may also reflect little control over the full assessment of the health of the study participants. While it is unlikely that severely affected individuals would have been included in these studies, it is entirely possible that more mildly affected individuals might have been included. It was of interest that the average size of the contiguous deletions in the DGV cohort was less than in the DECIPHER group, and therefore involved fewer AHDC1 flanking genes that might be associated with Mendelian disease (Table 3). Overall, however, the data from DGV were generally uninformative for the current study.

Next, the nine contiguous AHDC1 deletions reported in DECIPHER also involve multiple flanking genes already associated with Mendelian diseases (Supporting Information: Table S1). In addition, other flanking genes exhibited high mutational intolerance using the pLI scoring metric and hence are also strong disease-causing candidates. This complicates the assignment of mutation of AHDC1 as the primary cause of disease in individual DECIPHER cases. Some predicted phenotypes are shared between XGS and those bearing mutations in the flanking genes already associated with the disease, even though there are no cases where there is sufficient clinical data to rule out XGS and definitively assign any case to another disease locus. Similarly, the two recent reports in the literature of individuals with large de novo genomic deletions spanning AHDC1 suggest a model of haploinsufficiency, but the data are also confounded by the involvement of multiple genes in most of the observed lesions. Detailed follow-up may resolve whether these other loci also contribute to the clinical phenotype (Park et al., 2017; Wang et al., 2020).

The lack of definitive evidence from the large de novo deletions of AHDC1 in the literature or public databases prompted a deeper study of individual XGS28-P-R, who harbored a deletion involving AHDC1 and the smallest number of flanking genes yet recorded. This individual was of further interest because she both demonstrated a mild XGS phenotype and the preliminary data from array hybridization showed that the deletion included only the first, noncoding exon of the multiexon AHDC1 gene. These breakpoints were confirmed by a combination of short and long read sequencing via the ONT platform, which additionally allowed the identification of informative flanking markers that were used in subsequent RNA expression studies. While we anticipated that the mutated AHDC1 allele in individual XGS28-P-R would not support expression of a functional AHDC1 mRNA, activity of either downstream transcription start sites in the gene or else expression resulting from the new proximity of the AHDC1 to the nearest gene across the boundary of the deletion could not be excluded, a priori.

Surprisingly, results from molecular expression studies revealed near-normal gene expression of AHDC1 in individual XGS28-P-R, providing further evidence that AHDC1 deficiency may not be responsible for the proband’s clinical presentation. Although not unprecedented, such genetic compensation resulting in near-normal gene expression driven by one haplotype in individuals with de novo mutations in loci linked to Mendelian disease is less recognized as a factor in inherited disorders, but is previously well-recognized in cancer studies. This example of increased expression of the normal AHDC1 allele in individual XGS28-P-R can be compared to previous studies of deletions in critical genes with reported dosage compensation either due to functional redundancy or upregulation of the wild-type allele, particularly with tumor suppressor genes or transcription factors in cancer (Guidi et al., 2004; Kamikubo, 2018). Guidi and colleagues first reported transcriptional compensation for a loss of a single allele of a tumor suppressor gene (INI1), mediated by an increase in the rate of transcription from the INI1 promoter and tight regulation of INI1 expression levels. Genetic compensation under the control of an autoregulatory mechanism has also been previously reported in the BRN3a gene, belonging to the POU-domain class of transcription factors (Trieu et al., 2003). This autoregulation mechanism compensates for the loss of one allele in BRN3a heterozygotes by increasing transcription of the remaining allele, thereby suppressing haploinsufficiency (Trieu et al., 2003). There have also been recent reports of transcriptional adaptation mechanisms observed in model organisms including Caenorhabditis elegans and zebrafish mutants, with chromatin remodeling or expression modulation by long noncoding RNAs as stated potential mechanisms at play (El-Brolosy & Stainier, 2017; Jakutis & Stainier, 2021; Rouf et al., 2022; Sztal & Stainier, 2020; Zundo, 2021). While the precise mechanism of genetic compensation in individual XGS28-P-R is not known, the full elucidation of the molecular mechanisms involved will add to the growing understanding of genotype-phenotype relationships in the context of transcriptional adaptation in human disease, particularly the role of copy number alterations in cancer and genomic disorders (Bhattacharya et al., 2020; Carelle-Calmels et al., 2009; Lupski, 2022).

As AHDC1 deletion mutation may not be responsible for the clinical presentation of individual XGS28-P-R, it is possible that hemizygosity of the flanking genes (STX12, PPP1R8, and RPA2) that are heavily intolerant to LoF mutations and harbor 50% reduction in their gene expression levels, may be involved. While these flanking genes are not yet directly associated with a Mendelian disease gene in OMIM, there is indirect evidence for their phenotypic associations and potential candidate disease contributions (Table 4). Thus, the attribution of their role in contributing to the clinical phenotype in this individual cannot be excluded. The potential contribution of pathogenicity conferred from other multilocus variations should also be considered.

TABLE 4.

AHDC1 flanking genes deleted in XGS28-P-R

Deleted genes	pLI score	Gene function	Mouse phenotypes	Candidate phenotypic associations	Related pathways
FGR	0.48	(FGR) Sarcoma Viral Oncogene. Plays role in transferring phosphorus-containing groups and in protein tyrosine kinase activity	Mice homozygous for a knock-out allele exhibit a partial reduction in hemorrhage following induction of a local Shwartzman reaction, and show enhanced natural killer (NK)-cell receptor-induced IFN-γ production in cells	Severe pre-eclampsia and placental insufficiency	Phagocytosis, integrin signaling pathway
IFI6	0.23	Play a critical role in the regulation of apoptosis	Unknown	Hepatitis B, dengue virus	Interferon γ signaling and innate immunity pathways
FAM76A	0	Predicted to play a role in regulating transcription	Unknown	Adrenal tumor, esophageal tumor and soft tissue/muscle tissue tumor	Transcription regulation pathways
STX12	0.9	Syntaxin protein, regulates protein transport between late endosomes and trans-golgi network	Unknown	Frontotemporal dementia	Protein trafficking pathways
PPP1R8	0.99	Involved in RNA degradation, protein binding and protein phosphatase regulatory activity	Knock-out allele exhibit severe growth retardation and impaired cell proliferation and die between embryonic days 6.5 and 7.5.	Distal arthrogryposis Type 1, mesenteric lymphadenitis	Beta-adrenergic signaling, activation of cAMP-dependent PKA pathway.
THEMIS2	0	Plays a role in cell differentiation and T-cell maturation.	Unknown	Endometrial adenocarcinoma	Immune response pathways
RPA2	0.5	Replication protein, crucial for cellular response to replication stress and DNA damage.	Unknown	Cancer, ataxia-telangiectasia	Multiple DNA repair pathways, double-strand break repair via homologous recombination, DNA replication
SMPDL3B	0	Involved in immune response and membrane dynamics.	Knockout elevated serum 116 levels and higher numbers of immune cells.	Diabetic kidney disease	Insulin receptor signaling pathway

Abbreviations: AHDC1, AT-hook DNA binding motif containing 1; FGR, feline gardner-rasheed; PKA, protein kinase A.

Overall, this study leads to three conclusions. First, based on the evidence presented here, together with results from the examination of prior published data and public databases, there is little support for haploinsufficiency as a pathomechanism of how AHDC1 mutations lead to XGS (Jiang et al., 2018; Khayat, Li, et al., 2021).

Second, it is unlikely that AHDC1 deficiency is the cause of clinical symptoms in individual XGS28-P-R. We found that the expression of the remaining normal AHDC1 allele was elevated and likely compensated for the loss of activity of the mutant allele. When considered along with the fact that hemizygosity of multiple genes flanking AHDC1 could contribute to general phenotypes shared by many with DD, it is premature to diagnose this individual with XGS.

Third, the compensation mechanism revealed from this study could be a more general phenomenon protecting individuals with AHDC1 deletions from developing XGS and may also explain the incomplete penetrance reported in some other genetic disorders. An important question is therefore whether this compensation mechanism at the AHDC1 locus is a global phenomenon or constrained to specific tissues. Conducting targeted gene perturbations from relevant tissue types using appropriate cellular models would provide insights into this regulatory mechanism. Collectively, this knowledge will inform factors governing AHDC1 gene regulation and advance understanding of the gene, its function, and how it is regulated.