In consanguineous families with human developmental disorders, autosomal recessive genes are generally identified as the genetic aetiology. We have performed exome sequencing in 12 independent Iranian consanguineous singlet families with syndromic autism and identified four novel variants in IQSEC2, FOXG1, DMD, and CHKB in four families. Among them, only CHKB is an autosomal recessive gene, while IQSEC2 and DMD are X-linked and FOXG1 is an autosomal dominant gene. Interestingly, two of them, DMD and CHKB, are known muscular dystrophy genes. This emphasizes the striking association between autistic spectrum disorder and muscular dystrophy. Autosomal dominant disease genes with homozygous variants were reported in consanguineous Iranian families previously and both parents, who are heterozygous carriers in those families, were healthy [15, 27]. By contrast, Subject 2 with a heterozygous variant in an autosomal dominant gene, FOXG1, in our consanguineous singlet Iranian family was affected with syndromic autism.
ASD aetiology has consistently pointed to robust genetic components; however it has been complicated by extensive genetic heterogeneity. The current study disclosed four novel disease variants in IQSEC2, FOXG1, DMD, and CHKB in four syndromic singlet autism families from consanguineous marriages. Accordingly, two de novo heterozygous variants were found in IQSEC2 (c.1637G > A, p.W546X) and FOXG1 (c.563C > A, p.A188E), respectively, that associate with ASD and a neurodevelopmental disorder. A de novo variant in DMD (c.631T > A, p.L211M) and a novel homozygous variant in CHKB (c.382G > T, p.E128X) are likely involved in neuromuscular disorders and ASD.
The phenotypic spectrum of Subject 1 demonstrates that IQSEC2 is an X-linked gene with a female-specific X-linked dominant phenotype, including intellectual disability, autism, developmental delay, speech delay, stereotypic hand movements, poor eye contact, acrophobia, sleep disorder, and ADHD. High expression of IQSEC2 in the human fetal brain highlights the functional role of this gene in early human development, which is consistent with these clinical features. The truncating variant W546X in IQSEC2 is likely to generate loss-of-function because it will eliminate three functional domains (Sec7, PH, and PDZ binding motif) affected, even if it escapes NMD. According to ACMG classification, this variant belongs to the category of PVS1/PS2/PM2/PP3 and combining these criteria classifies it as pathogenic (Table 2). Moreover, the CADD score is 36, and a score of greater than 30 designates the top of less than 0.1% of most deleteriousness. Careful NGS data analysis with consideration of all modes of inheritance suggests that it may be unrelated to shared recessive genes identified in consanguineous families.
Residue Ala188 in FOXG1 is located in Forkhead domain, which is highly conserved and intolerant. Thus, the de novo variant Glu188 (c.563C > A) in Subject 2 will likely affect DNA binding of this domain. According to protein modeling, its replacement with the bulky Glutamic acid is predicted to cause spatial and steric hindrance, resulting in FOXG1 misfolding. Development delay, ASD, seizures, and intellectual disability found in this 18-year-old Iranian girl can be explained by the high expression pattern of FOXG1 in the human fetal brain (Fig. 6A). Based on ACMG classification, this belongs to the categories of PS2/PM1/PM2/PM5/PP2/PP3 and thus can be classified as pathogenic (Table 2). Moreover, The CADD score is 42 and greater than 40, indicates the top of less than 0.01% of most deleteriousness. A de novo variant also suggests autosomal dominant and unlikely to be an autosomal recessive pattern based on the consanguineous marriage of her parents.
The X-linked variant (c.631T > A, p.L211M) in Subject 3 was a de novo hemizygous missense change in DMD due to its absence in his unaffected parents. Two major phenotypes of a neurodevelopmental feature coupled with muscular dystrophy in this Iranian male could be explained by the highest expression of DMD in the skeletal muscle followed by the second highest expression in the human fetal brain (Fig. 6A). The Z score of a missense variant in DMD in the general population is -2.43 based on gnomAD data (https://gnomad.broadinstitute.org/), suggesting more tolerance to the missense variation. However, this does not preclude the possibility of missense mutations in this gene including L211M reported here.
There are multiple lines of evidence suggesting the pathogenicity of this DMD missense variant in this proband with development delay, autism, intellectual disability, muscular dystrophy, and hypotonia. Firstly, there are totally 5,159 mutations reported in DMD based on the HGMD website (Human Gene Mutation Database, http://www.hgmd.cf.ac.uk/ac/index.php). Among them, 989 are missense or nonsense mutations. Approximately one third of these are reported missense mutations, which suggests > 300 reported missense mutations in DMD in human patients. Secondly, L211M is absent in both parents and thus de novo. It is also absent in gnomAD and thus defined as PM2 based on ACMG guidelines . Notably an elevated serum CPK level and the clinical details suggested dystrophinopathy as third evidence. Furthermore, in silico protein modeling of the residue L211M in Fig. 4 shows the disruption of the 3-dimensional structure in the mutant DMD. Taken together, pathogenicity has been determined as “likely pathogenic” in Table 2 by the entire body of evidence (PS2/PM2/PP3). Without our result in Subject 3, clinicians or geneticists would never have thought that an ASD male with muscular dystrophy from a consanguineous marriage could have a mutation in the X-linked DMD gene.
In Subject 4, the homozygous nonsense variant, c.382G > T (p.E128X) in exon 3 of CHKB was heterozygous in both healthy carrier parents. This premature stop codon was predicted to result in a truncated protein. Even if a C-terminal truncated protein escapes NMD to some extent, CHKB dimers, which are required for choline kinase function, would not be generated (Fig. 5). The CADD score scale was 32, which indicates the top of less than 0.1% of most deleteriousness along with ACMG classification criteria using combined criteria of PVS1/PM2/PM3/PP1/PP3 indicated that it is pathogenic (Table 2). Clearly, this variant follows an autosomal recessive inheritance pattern, based on the family history and in line with our hypothesis for consanguineous families. CHKB had higher levels of mRNA expression in skeletal muscles and brain tissues (Fig. 6A), which is consistent with its involvement in muscular dystrophy and autism.
In the remaining eight families, we identified six polymorphic variants in six families, among whom three heterozygous variants were found in one of the healthy parents, whereas another three heterozygous variants were found in both healthy parents in the other three families. In two families no candidate variants were found.
RTT and autism
Rett syndrome (RTT, MIM 312750) is a rare neurodevelopmental disorder, which affects females almost exclusively, occurring in 1 in 10,000–15,000 female live births, and which is responsible for around 10% of female severe intellectual disabilities of genetic origin . Using a systematic gene screen approach from RTT families, in whom exclusion mapping studies mapped the disease gene locus to Xq28, MeCP2 (methyl-CpG-binding protein 2) was identified as a disease gene. Clinical features of typical RTT appear to be enigmatic for the normal perinatal period until 6–18 months of age, followed by loss of speech and purposeful hand use, and developing microcephaly, seizures, and autism. IQSEC2 mutations were identified in some individuals with either classic RTT or clinical symptoms similar to RTT . whereas FOXG1 mutations underlie atypical RTT . So, from a genetic point of view, IQSEC2 and FOXG1 are clearly responsible for atypical and variant RTTs .
IQSEC2, identified in the families with intellectual disability by systematic X chromosome exome sequencing in 2010 , is an X-linked gene (Xp11.22) having been named for two protein domains, IQ-like and Sec7 domains (Fig. 2). IQSEC2 is a member of GEF (guanine nucleotide exchange factor) for the ADP-Ribosylation Factor (ARF) family of small GTP-binding proteins (ARFGEF) . According to the high expression level of IQSEC2 in all brain regions (highest in the hippocampus), its disruption represents a significant cause of intellectual disability . IQSEC2 variants are also associated with seizures, behavioural and psychiatric abnormalities. A recent review reported 70 different types of IQSEC2 mutations, in which autism presented in 25% of affected males and 30% of females . Escaping X inactivation to some extent may explain the rather high prevalence of both intellectual disability and autism in heterozygous females. Due to this high level of comorbidity, it has been demonstrated that IQSEC2-associated intellectual disability and autism share common biochemical abnormalities . Males have both missense and truncating pathogenic variants; however, females have mostly truncating variants like in Subject 1 of our study .
A highly regulated process of brain development encompasses the particular spatio-temporal activation of cell signaling clues . Transcription factors play a fundamental role in this process by spreading information from external signaling to the genome. The encoded protein, Forkhead box protein G1, FOXG1 (formerly brain factor 1 [BF-1]) primarily acts as a brain-specific transcriptional repressor, which is involved in brain development . It is expressed in the evolving nervous system, where it plays a critical role in the establishment of the regional subdivision of the developing brain ranging from telencephalon specification, patterning and neuronal differentiation, maintenance, and survival of mature neurons [38, 39]. It is one of the earliest expressed transcription factors that can change the structure of chromatin and permit the binding of other factors. Consequently, it is considered a pioneer transcription factor.
Any alterations in the FOXG1 dosage resulted in a complex group of cellular effects with key consequences for human diseases, including neurodevelopmental disorders. As an example, FOXG1 interacts with global transcriptional co-repressors of the Groucho family as well as with the transcriptional repressor JARID1B. The interaction of these proteins has functional significance for early brain development. FOXG1 also associates indirectly with the histone deacetylase 1 protein (HDAC1), as MeCP2 also does .
Autism and muscular dystrophy
Muscular dystrophy has also been accompanied by neuropsychiatric disorders like ASD and association between autistic spectrum disorders and dystrophinopathy (including Duchenne and Becker muscular dystrophies) has recently been reported . Duchenne muscular dystrophy and megaconial congenital muscular dystrophy manifest a broad spectrum of clinical severity with multisystem involvement. Based on the largest study, only 11 of 351 (3.1%) boys with muscular dystrophy were reported to have ASD , indicating that ASD in DMD is still a rare event.
As one of the first cloned human genes by positional cloning, the DMD gene attracts attention not only for its medical prominence but also for its exceptionally large size. It spans a genomic range of more than 2 Mb and encodes a large protein, dystrophin, containing an N-terminal actin-binding domain, and multiple spectrin repeats alongside a family of N-terminal truncated isoforms by activating independent promoters. Mutations, most commonly deletions, in DMD, located on the short arm of the X chromosome (Xp21.1 to Xp21. 2), lead to Duchenne and Becker muscular dystrophies (DMD and BMD). These have an impact not only on muscle but also on the brain . With seven promoters, DMD has a full-length and shorter isoforms that may play a crucial role in synaptogenesis and axon guidance during brain development; hence, Becker Muscular Dystrophy is associated with neuropsychiatric disorders including ASD, which has been reported in two individuals with DMD mutations . This is also supported by the highest expression of DMD in the skeletal muscle tissues compared to other human organ tissues (Fig. 6A).
Dystrophin and several components of the dystrophin-associated protein complex (DPC) can be found in the soma and dendrites of cortical and hippocampal neurons as well as cerebellar Purkinje cells, where it is associated with the postsynaptic membrane of neurons and is present at high levels in the postsynaptic density (PSD); therefore, it may have a role in the synapse structure or function . Additionally, the full-length isoforms are largely expressed neuronally and localized to punctate structures, which correspond to a group of GABAergic synapses in many brain regions (the cerebellum, cortex, and hippocampus) that are responsible for higher-order functions such as learning and memory. Therefore, cognitive impairment in individuals with DMD mutations is caused by a lack of dystrophin in the neuronal membrane.
The CHKB is located on 22q13.33 that comprises 11 exons, 10 of which (except exon 10) have been implicated in the pathogenesis of megaconial congenital muscular dystrophy (CMD). It encodes the choline kinase β protein, which catalyzes the first step of the de novo biosynthesis of phosphatidylethanolamine and phosphatidylcholine, an important lipid component of the mitochondrial membrane of skeletal muscles and the brain . Then, it plays a vital role in phospholipid biosynthesis.
Autosomal recessive congenital megaconial muscular dystrophy (CMD, MIM 602541) is characterized by neonatal hypotonia, early-onset muscle wasting, intellectual disability, and developmental delays without brain malformation as specific clinical features, which were manifested in Subject 4. It is a rare inborn error of metabolism with a broad spectrum of clinical severity and multi system involvement in which some individuals presented with autism . CMD is characterized by a distinct sparseness or absence of mitochondria in the centers of fibers and enlarged (megaconial) mitochondria in the periphery of muscle . CHKB catalyzes the first step of phosphatidylcholine (PC) biosynthesis and its mutations result in loss of choline kinase activity and decreased levels of PC . In relation to the phenotype-first approach, the clinical diagnosis of mitochondrial disease was primarily considered to be due to the multi-system nature of the condition. Laboratory testing such as pathognomonic features in the muscle biopsy and elevated serum creatine kinase also confirmed phenotypic compatibility with the identified variant.
We have found two muscular dystrophy genes, DMD and CHKB, mutated in our two syndromic autism families. These two autism genes overlap with non-cognitive phenotypes of muscular dystrophy, demonstrating significant genetic relation across neurodevelopmental and neuromuscular disorders. It clearly shows that some genes transcend the boundary of clinical categories, and this result could have broad clinical implications for diagnosis.
Our study shows that some children with autism may have a genetic defect that affects the muscles, and consanguineous singlet autism individuals might have autosomal dominant or X-linked disease genes. Additionally, future research consisting of animal models and functional studies of these four novel variants are essential to understand the pathological role of these variants.
The major limitation of our study should be illustrated in the context of Functional experiment and sample size. However, our finding is mostly pathogenic due to ACMG classification and could be causative of the disorder, Functional studies are still important to understand the pathoethology of the novel variants, mainly a VUS variant in the DMD gene. Although we performed the first study in Iranian patients with ASD and concluded the novel clinical insight to ASD, our sample size is small, twelve families. Notwithstanding that recruiting suitable cases in one type of ASD (Syndromic) is critical with multi-center collaboration with mostly clinical implications and specialists. Due to untrusted behavior from other scientists that created confusion of the families suffering from ASD, and things of that nature, we missed the suitable cases, 30 out of 42. Moreover, it is worth noting that our study established to depict the genetic architecture of syndromic autism in Iranian patients and in the future cohort study we involved more cases.