Novel frameshift mutation in LIS1 gene is a probable cause of lissencephaly: a case report

Lissencephaly (LIS) is a subtype of malformations of cortical development (MCD), which are a heterogenous group of disorders with diverse phenotypic and genotypic presentations. Patients with LIS may present different degrees of developmental delays, seizures, severe psychomotor impairment, muscle spasticity or hypotonia [1].

Lissencephaly is a disorder caused by a defect in neuronal migration, which occurs between 12 and 24 weeks of gestation and results in a lack of development of brain folds (gyri) and grooves (sulci) [2]. Neuronal migration is a complex process, which requires the coordination of many gene products.

LIS1 and DCX were the first genes that were associated with LIS, discovered in 1993 and 1998, respectively [3, 4]. In the past years with the advent of new molecular genomics technologies, many additional genes were found. These LIS-related genes include ACTB, ACTG1, ARX, CDK5, CRADD, DYNC1H1, KIF2A, KIF5C, NDE1/NDEL1, TUBA1A, TUBA8, TUBB, TUBB2B, TUBB3, TUBG1, RELN and VLDLR. Many of these 19 LIS-associated genes are related to microtubule structural proteins (tubulin) or microtubule-associated proteins [5].

The PAFAH1B1 gene (Genbank accession number: NM_000430), located at chromosome 17p13.3, encodes the alpha subunit of the 1B isoform of the platelet-activation factor acetylhydrolase regulatory, a highly conserved protein of 410 amino acids, known as LIS1 or PAFAH1B1 [6]. It has two protein coding transcripts and several non-coding ones. LIS1/PAFAH1B1 forms the non-catalytic subunit of the G protein-like heterotrimeric cytosolic platelet-activating factor acetylhydrolase (PAF-AH) brain isoform Ib (PAFAH1B1) [7]. Along with two other subunits, PAFAH1B2 and PAFAH1B3, LIS1/PAFAH1B1 forms a trimeric complex which regulates the level of platelet activating factor (PAF) in the brain, by catalyzing the removal of the acetyl group at the SN-2 position of platelet-activating factor [8, 9]. The regulation of optimal concentrations of PAF in the brain may be critical for correct neuronal migration, essential for normal brain development and function. LIS1/PAFAH1B1 has also been shown to play a central role in the organization of the cytoskeleton, which in turn affects neuronal proliferation and migration [6]. Mutations in this gene have previously been associated with cortical brain malformation in children (Table 1).

Here, we comment on the case of a seven years old boy with an undiagnosed rare disease, with non-specific symptoms that could be compatible with LIS, but with an unclear presentation. Whole genome sequencing (WGS) of the patient was performed in the context of a genomics project (urugenomes.org) and sequence data was analyzed following a bioinformatics pipeline which concluded with a set of annotated and prioritized variants. A novel candidate frameshift variant was found that fits with the boy’s phenotype. To support the pathogenicity of the variant we used computational prediction tools and made segregation analysis with Sanger sequencing. To the best of our knowledge this is the first time this variant is reported [11] and it is the most likely cause of the patient’s disease.

The index case is a seven years old boy with perinatal clinical records of poorly controlled pregnancy, homelessness and multiple drugs abuse through all gestation. Both parents are healthy and non-consanguineous. Early term delivery, low birthweight with microcephaly at birth with a head circumference of 32 cm (Z score -3).

The patient develops a spastic bilateral cerebral palsy, Gross Motor Function Classification System (GMFSC) IV, Bimanual fine motor function (BFMF) IVa, Communication Function Classification System (CFCS) IV. Associated with a profound intellectual disability, visual and auditory sensory deficit, pharmacoresistant epilepsy with generalized tonic clonic seizures, and congenital microcephaly with a head circumference growth always under -3 standard deviations. No dysmorphic signs were detected.

Cerebral magnetic resonance imaging (MRI) shows a diffuse lissencephaly-pachygyria spectrum with main affectation at posterior brain areas (Fig. 1).

Whole genome sequencing (WGS) of the patient was performed and sequence data was analyzed following a bioinformatic pipeline which included analysis of the quality of reads [12], mapping onto human reference genome (hg19) [13], mark of duplicates, sorting and variant calling [14]. The variants obtained were annotated [15] and then subjected to different sets of filters to detect potentially causative mutations (see Supplementary Material). After these filters were applied, we obtained 40 homozygous or hemizygous variants with population frequency less than 1% located at splicing sites or coding regions, 458 heterozygous variants with population frequency less than 1% and 439 heterozygous variants with population frequency less than 0.5% and located at splicing sites or coding regions.

Among these prioritized variants we found a potential causative mutation in heterozygous state in the LIS1/PAFAH1B1 gene. This gene was previously associated with the phenotype (LIS), especially with an autosomal dominant mechanism of inheritance. The potential causative mutation found is a frameshift insertion of a single nucleotide in exon 8 (PAFAH1B1:NM_000430:exon8:c.681dupG:p.(Lys228Glufs*28)), that lies between the first 23% to 55% of the protein depending on the transcript, according to the SIFT Indel tool [16]. The frameshift indel was reported as damaging with a confidence score of 0.858, and causing a nonsense mediated decay (NMD) response. It generates a premature stop codon 27 amino acids later, causing the loss of 156 amino acids.

If the mutated gene evaded NMD and led to a final product, this would be a 254 amino acids protein instead of the wild type 410 residues. A crystal structure has been described for LIS1 complexed to the brain cytosolic PAF-AH [17]. The complex shows that LIS1 folds into a beta propeller and interacts as a homodimer with a PAF-AH homodimer. From 14 reported surface interacting residues with PAF-AH, 8 are missing from our patient’s hypothetical protein. We predict that the mutated LIS1 could have self-aggregation tendencies, as the 27 new residues composing the shorter C-terminal region, not only would not allow the correct folding into a complete beta-propeller, but in addition would be highly disordered. As a qualitative indicator for this, the homology model in this C-terminal region has very low quality, in particular for the ‘HRTQRMGTY’ amino acid stretch, according to QMEANDiscO scoring function [18]. Even without aggregation and assuming the protein could fold into a ‘half propeller’, this protein would be unable to productively interact with PAF-AH as well as its additional molecular partners, notably dynein and a number of dynein-associated proteins. Indeed, LIS1 has been described as a molecular hub at a crossroad of several pathways, coupling PAF signaling to dynein regulation [17]. We expect all these functions to be hampered or inexistent in the protein product coded by this allelet.

The unaffected mother was sequenced at the proposed variant position and no mutation was detected. This is considered to support the hypothesis of a de novo mutation in the patient. Unfortunately the father of the patient (also unaffected) was not available for analysis.

Figure 2 A shows an IGV view of the candidate position. 37 reads are covering that location with a good quality. Additionally, the variant was confirmed with Sanger Sequencing in the patient (Fig. 2B, top) and was not seen in the mother (Fig. 2B, bottom)

We found a novel probably causative frameshift variant in a patient with a previously undiagnosed rare disease using WGS. Previous genetic tests (sequencing of MECP2 and ARX genes, and methylation analysis for Angelman syndrome) were performed with inconclusive results. This is expected since Lissencephaly and epileptic encephalopathy are highly heterogeneous genetic disorders in their etiology: ie. different genes are associated with several presentations of this pathology. For example, RELN gene is affected in the Norman-Roberts syndrome (LIS2) [19], heterozygous mutations in TUBA1A are responsible for the LIS3 syndrome [20], homozygous mutations in the NDE1 gene are associated to LIS4 [21], among many others (LAMB1 to LIS5 [22], KATNB1 to LIS6 [23], CDK5 to LIS7 [24], TMTC to LIS8 [25], MACF1 LIS9 [26], CEP85L LIS10 [27]). Additionally X-linked forms of Lissencephaly are caused by DCX and ARX genes [28]. Hence, usually WGS or WES are accurate strategies for assessing patients with epileptic encephalopathy. However, in this case if we had had the MRI results (fairly consistent with LIS) before we had done the NGS sequencing, we might have end up doing a targeted sequencing approach, such as the PAFAH1B1 gene or at least a subset of genes or WES, instead of doing the complete genome. This being a matter of costs and resources and not crucial for the patient’s diagnosis.

According to ACMG (American College of Medical Genetics and Genomics) variant interpretation guidelines [29] the frameshift variant found corresponds to the PVS1 (pathogenicity very strong) rule. It is a null variant (frameshift) in a gene where loss of function (LOF) is a known mechanism of disease (in ExAC database PAFAH1B1 gene has a maximal probability of being LOF intolerant, pLI = 1 [30]). The frameshift mutation is also classified as a PM2 (moderate evidence of pathogenicity) since it was absent in population databases (1000 Genomes Project, GnomAD, etc.). We also consider applying rule PP4 (Patient’s phenotype is highly specific for a disease with a single genetic etiology), since the MRI findings are very specific for PAFAH1B1-related LIS. According to ACMG rules, the variant is classified as pathogenic, as it belongs to one very strong (PVS1), one moderate (PM2) and one supporting category (PP4).

We could also consider applying PP3 (supporting evidence of pathogenicity) since the pathogenic computational verdict is based on one pathogenic prediction from SIFT Indel Tool, one pathogenic prediction from the conservation score GERP [31] and no benign predictions. However, some studies avoid [32] applying PP3 in LoF variants when PVS1 is valid.

Furthermore there are other disruptive (frameshift or stop codon) variants in the same gene region reported as pathogenic [33], supporting the importance of the region for proper gene product function.

Through this work we were able to find a molecular diagnosis of a rare disease in a seven years old boy with severe and heterogeneous neurological symptoms. We found a de novo novel frameshift mutation in the LIS1/PAFAH1B1 gene that most likely causes a silencing of one allele. This finding shows the benefit of the use of NGS as a diagnosis tool of rare diseases.