Whole Exome Sequencing of a Multiplex Family of Indian Origin Identifies Variants in the RAI1 and FLII Genes within the 17p11.2 Region in Siblings with Autism and Smith Magenis Syndrome

Autism spectrum disorders (ASDs) featured with impairment in social and communication behavior are highly known for their genetic origin, illustrated about three decades ago [1]. In addition to the genetic heterogeneity observed in individuals with ASD, clinical heterogeneity, stemming from multiple comorbidities and syndromic features, further complicates efforts to establish genotype-phenotype correlations [2]. Approximately 25% of ASD subjects present syndromic features with a known genetic cause [3]. These features can be attributed to copy number variations (CNVs) or single nucleotide polymorphisms (SNPs). The SFARI database currently lists 251 genes associated with syndromic forms of ASDs (https://gene.sfari.org). Given the heterogeneity of ASD, the literature suggests implementing whole exome sequencing (WES) as the primary diagnostic tool [4]. This cost-effective method can detect both CNVs and SNVs and has shown greater diagnostic potential compared to Chromosomal Microarray Analysis [5].

The current study leverages the utility of WES to uncover the probable pathogenic variants in an Indian multiplex family with two siblings (female-S1 and male-S2) diagnosed with ASDs along with suspicious syndromic form. Our computational pipeline and functional analysis found haploinsufficiency of the RAI1 gene in the female sibling (S1) and FLII gene in the male sibling (S2), both of which correspond to the Smith Magenis syndrome (SMS) phenotypes [6]. Clinically, SMS patients show ASDs characteristics accompanied with minor physical anomalies (MPAs), developmental delay, self-harming behavior, neurological disorder and mild to moderate intellectual disabilities [7]. The identified genes in the current study are encompassed at Chr17p11.2 which is called the SMS critical region [8]. Our WES findings were consistent with the clinical findings that imply SMS features in both siblings.

The multiplex family with two affected siblings presenting with autistic phenotype was recruited for the study at the Centre for Advanced Research and Excellence in Autism and Developmental Disorders (CARE-ADD), St. John’s Research Institute, Bengaluru, Karnataka, India. The study was approved by the Institutional Ethical Board (IEC Approval No. 100/2022). Subjects were diagnosed with ASDs at the age of 3 years by the INCLEN Diagnostic Tool for Autism Spectrum Disorder (INDT-ASD) based on DSM IV-TR. The siblings were recruited for the current study at the age of 22 and 13 years; hence, they were reassessed by using the All India Institute of Medical Sciences (AIIMS)-Modified-INDT-ASD Tool based on DSM 5 criteria [9] that indicated both the siblings continue to meet the criteria for autism according to DSM-5. Severity was estimated by the Indian Scale for Assessment of Autism (ISAA) [10] that showed moderate level of severity in both the siblings. Vineland Adaptive Behavior Scale-2 (VABS-2) was administered to measure the adaptive behaviors in ASDs subjects. Sibling S1 obtained an adaptive behavior composite of 20 on the VABS, indicating a low level of adaptive skills. The VABS profile of S1 is indicative of a global delay in skills. Sibling S2 obtained an adaptive behavior composite of 49 on the VABS, indicating a low level of adaptive skills. VABS profile of S2 is indicative of a global delay in skills, with relatively better motor skills, despite being in the moderately low level. The Autism Spectrum Quotient (AQ) was used to assess autistic traits in parents. The parents’ autism AQ scores were 6 (father) and 17 (mother), indicating the absence of significant traits for broad autism phenotypes. Demographic details along with family medical history were also collected. Later, physical examination for typical SMS syndromic features was conducted on both the siblings as recommended [11]. Detailed clinical information is given in the online supplementary file (for all online suppl. material, see https://doi.org/10.1159/000539400).

Written informed consent was obtained from the parents for all investigations. The genomic DNA samples of the mother (M), father (F) and the two siblings (S1 Female, S2 Male) were subjected to WES using SureSelectXT Human All Exon (V6) and sequenced on Illumina HiSeq series to generate 2 × 150 bp sequence reads at 80–100X coverage. The relatedness check between probands and parents was performed using Ancestry and Kinship (AKT) and VCFTools. Variant calling was performed using GATK and VarScan to identify genetic variants including SNPs and insertions/deletions (INDELs). The raw sequencing data was subject to quality assessment using FASTQC, and subsequent steps including adapter trimming and quality filtering (Phred Score ≤20) were performed using TrimGalore and Cutadapt.

Both de novo and inherited variants were annotated using Variant Effect Predictor (VEP) http://www.ensembl.org/vep to identify SNPs. Stringent parameters like depth (DP ≥20), Genotype Quality (GQ ≥20), Variant Allele Frequency (VAF ≥0.5) and allelic depth at the mutated base (ADA ≥5) were applied to get the list of genes with damaging variants. All “high” impact (frameshift, stop gain, stop loss, and splice-site variants) and “moderate” non-synonymous variants (nsSNPs) predicted to be damaging by ≥15/25 (i.e., ≥60%) of VEP tools were considered. Variants were filtered further using the global minor allele frequency (GMAF ≤0.01) to remove the common variants across populations using 1000 Genomes, gnomAD, ExAC and Genome-Asia [12]. Further, SFARI database, AutDb (http://autism.mindspec.org/) and de novo database (https://denovo-db.gs.washington.edu/denovo-db/) were used to filter known ASD genes. Subsequently, functional analysis was done, and later genes were annotated for biological process, human and animal phenotype with adjusted p value <0.05.

Validation of the RAI1 gene was carried out by amplifying the desired region by conventional PCR. Primers were designed spanning 250 bp upstream and downstream of the putative variant. The amplified product was run and visualized by agarose gel electrophoresis. Direct sequencing of the putative gene was carried out using BigDye terminator v3.1 Cycle Sequencing Kit and the ABI 3500 Genetic analyzer with 8-capillary (Applied Biosystems^™). The chromatogram data obtained was analyzed for the confirmation of the variant.

The two siblings of an Indian multiplex family born of a non-consanguineous marriage, siblings S1 (Female) and S2 (Male) were recruited for the study at the ages of 22 and 13 years, respectively, of severity. The WES analysis identified eight high-risk genes carrying protein-damaging variant that have been reported in other autism databases (Table 1). Functional analysis predicted that identified genes were highly enriched for cellular mechanism and developmental morphogenesis with adjusted p value <0.05 by enrichment analysis (online suppl. Table S1).

Further, annotating the genes with respect to the human phenotype (HP) and mouse phenotype (MP) offered the most significant clues with regard to clinical phenotypes that were noticed in both the siblings S1 and S2. Exclusively top ten Human phenotype terms highlighted RAI1 and FLII as potential genes with correlation to clinical phenotype (online suppl. Table S2). More interestingly, both FLII and RAI1 were enriched for several human phenotype terms associated with clinical characteristics for SMS (HP:0100716 self-injurious behavior; HP:0001250 seizures; Scoliosis HP:0002650; abnormality of speech or vocalization HP:0002167). Our WES analysis did not detect any INDELS at the 17p11.2 region. This finding was further confirmed by Sanger sequencing validation of exon 3 of the RAI1 gene, which also showed no nucleotide deletions as mentioned below. This implicated rare heterozygous missense variant c.5320C>T (p.Arg1774Trp) in RAI1 gene with CADD score of 24.1 could be a potential gene responsible for syndromic features in S1. Similarly, FLII gene with missense variant c.2030A>C (p.Glu677Ala) with CADD score of 29.0 in S2 could be responsible for the observed features. The correlation between the genotype and phenotype led us to evaluate the subjects for SMS phenotype.

Sibling S1 exhibited behavior and neurological abnormalities along with craniofacial abnormalities like broad square face with deep-set eyes, a bulbous nose, full cheeks, short philtrum, malocclusion, and long eyelashes. Clinician also witnessed several scars on the forearms and body from skin-picking behavior. Another observed feature was the presence of a coarse, low-pitched voice. Further, sibling S2 was positive for features like macrocephaly, broad square-shaped face with malar hypoplasia, prognathism, a wide nasal bridge and short philtrum, deep-set eyes, and a bulbous nose. Hypertelorism and long eyelashes were also present. Malocclusion was observed with a high-arched palate. When compared to sibling S1, sibling S2 had a maximum number of craniofacial abnormalities than neurological disorder as described in detail in Table 2. These clinical findings suggest that two different genes RAI1 and FLII, located at the critical region of SMS, may be the probable candidate genes for the respective phenotype.

The study intensely focused on RAI1 gene and validation by Sanger sequencing was carried out on both the siblings and parents to confirm the true positive and true negative variants that were identified by WES. Forward primer: 5′TGC​CAA​AAC​CCG​GCC​AAC​TTC​A 3′; reverse primer 3′ TTA​CCA​TGT​CCA​CGG​CCA​CCT​T 5′ was used to amplify the amplicon size of 514 bp. The amplified product was visualized by running 1% agarose gel electrophoresis (Fig. 1b). Sanger sequencing confirmed that the variant identified by WES was a true positive variant in sibling S1 which was found to be maternally transmitted and true negative in sibling S2 (Fig. 1c).

To the best of our knowledge, this would be the first WES report with genetic and clinical validation on autistic subjects of Indian origin carrying syndromic features. The computational pipeline and functional analysis with respect to clinical phenotype identified promising evidence for RAI1 and FLII genes that are located on chromosome 17p11.2. The study suggests that these genes could be potential candidates associated with the neurodevelopmental disorder, SMS [13, 14] observed in the subjects. Validation of the RAI1 gene by Sanger sequencing confirmed the variant as true positive and maternally inherited in S1 that was coherent with the WES findings. Clinical re-examination of both the siblings showed syndromic features consistent with SMS.

SMS is characterized by typical and distinctive clinical features including craniofacial abnormalities, intellectual disabilities, cognitive impairment, and behavior problems. More than 70% of SMS patients have 17p11.2 del [15]; however, 10% of patients have been reported with point mutations (frameshift and nonsense) resulting in haploinsufficiency of the RAI1 gene [16]. We identified a clinically novel missense variant in the RAI1 gene (NM_030665.4:c.5320C>Tp.Arg1774Trp) at exon3, known for its mutational hotspot. Further, the sibling S1 seems to have inherited the said variant from the mother, contrary to it being a sporadic de novo mutation [17]. This could be due to the phenomenon of maternal mosaicism as supported by the literature [18]. While maternal mosaicism was not directly investigated, the presence of both wild-type and mutant peaks in the Sangers’ sequencing results of the maternal blood sample may suggest an indication of mosaicism. Nevertheless, it is crucial to note that additional evidence is required to substantiate the same. It is also noteworthy that no autistic traits were observed in the mother, likely due to the dosage-sensitive nature of the RAI1 protein.

Rai1 (retinoic acid induced 1) is a nuclear protein of 203 kDa vastly expressed in the brain tissue, with 7 functional domains (Fig. 1d). The missense variant identified in the current study resulted in a change in amino acid from arginine (R) to tryptophan (W) at a highly conserved position (Fig. 1e) that falls between the NBD and PHD of the RAI1 protein (Fig. 1d). Several truncating mutations have been identified in RAI1 gene, be it the N- or the C- terminal, it is found to have varied functional effects at the molecular level rather than its translation to a clinical phenotype [19]. Further, de novo missense mutation 3440G > A(p.R1147Q) located near the NLS domain has been shown to have an effect on the BDNF-enhancer-driven transcription activity [17]. This further seemed to be consistent with other syndromic disorders like the ROHHAD syndrome and Prader-Willi syndrome, which phenotypically overlap with SMS, where BDNF insufficiency was reported [20]. Similarly, in our observation, the sibling S1 with RAI1 haploinsufficiency presented clinically with obesity, BMI of 30.4 kg/m² and cognitive deficits, suggesting a potential involvement of BDNF under expression in these cases.

Further, the study has identified missense variant in the FLII gene as a probably pathogenic variant in the other sibling (S2) who also had prominent features for SMS, transmitted from unaffected father. Upon comparison, the clinical phenotype difference between the siblings S1 and S2 were evident, with the former subject exclusively presenting with seizures disorder and otolaryngologic abnormalities that were not observed in S2. However, the sibling S2 was positive for more than 80% of MPAs when compared with the other sibling S1 clearly indicating the involvement of the FLII gene, present in the SMS critical region [8]. Studies witness phenotype differences between the patients carrying 17p11.2 deletion (short/larger deletion) and RAI1 haploinsufficiency [13]. This implies contribution of other genes (SREBF1, SHMT1, TOP3A) present in the deleted region to the SMS phenotypes. Likewise, FLII was enriched for human phenotype terms like Scoliosis (HP:0002650) and other behavior terms (online suppl. Table S2) that were consistent with phenotype observed in S2. Though the prevalence rate of SMS is 1 in 15,000 [21], it remains a disorder whose molecular mechanisms have not been well explored, hence necessitating extensive exploratory studies to identify molecular mechanisms. We conclude that WES data was effectively used to understand the genetic burden, variability, and potential markers in the affected siblings.

We thank the Indian Council of Medical Research (ICMR) (project ID 2021-15112-Genomics/BMS – 2022) for supporting the fellowship for the first author. We thank the Founder of CARE-ADD late Dr. Ashok M.V. for supporting this project and acknowledge the support of clinicians and nursing research fellows. We thank the subjects for providing consent to participate in the study.

This study was approved by the Institutional Ethics Committee of the St. John’s National Academy of Health Sciences (IEC Study Ref. No. 100/2022). Written informed consent was obtained from the parent of the patient for participation in the study and for publication of clinical details.

The authors have no conflicts of interest to declare.

This study was facilitated by the Philanthropic funds received by the Centre for Advanced Research and Excellence in Autism and Developmental Disorders (CARE-ADD), St. John’s Research Institute, Bangalore, Karnataka, India.

S.D.: patient recruitment, sampling, data analysis, conceptualization and interpretation, design paper, and manuscript preparation. S.V.P.: data analysis and interpretation and material and methods preparation. S.R.: critical review, evaluation, and contribution to the preparation of the discussion. A.H.S.: mentoring and reviewing. A.K.: supervision and mentoring. A.S.N.M.: WES analysis and gene-level filtration analysis. N.B.: clinical evaluation and clinical inputs. M.V.: clinical evaluation and inputs; N.K.C.G.: project head, mentoring, and critical review of the manuscript.

The raw dataset utilized in this research are available in the NCBI repository, accession number PRJNA1036375.