High prevalence of multilocus pathogenic variation in neurodevelopmental disorders in the Turkish population

Abstract

Neurodevelopmental disorders (NDDs) are clinically and genetically heterogenous; many such disorders are secondary to perturbation in brain development and/or function. The prevalence of NDDs is > 3%, resulting in significant sociocultural and economic challenges to society. With recent advances in family-based genomics, rare-variant analyses, and further exploration of the Clan Genomics hypothesis, there has been a logarithmic explosion in neurogenetic "disease-associated genes" molecular etiology and biology of NDDs; however, the majority of NDDs remain molecularly undiagnosed. We applied genome-wide screening technologies, including exome sequencing (ES) and whole-genome sequencing (WGS), to identify the molecular etiology of 234 newly enrolled subjects and 20 previously unsolved Turkish NDD families. In 176 of the 234 studied families (75.2%), a plausible and genetically parsimonious molecular etiology was identified. Out of 176 solved families, deleterious variants were identified in 218 distinct genes, further documenting the enormous genetic heterogeneity and diverse perturbations in human biology underlying NDDs. We propose 86 candidate disease-trait-associated genes for an NDD phenotype. Importantly, on the basis of objective and internally established variant prioritization criteria, we identified 51 families (51/176 = 28.9%) with multilocus pathogenic variation (MPV), mostly driven by runs of homozygosity (ROHs) - reflecting genomic segments/haplotypes that are identical-by-descent. Furthermore, with the use of additional bioinformatic tools and expansion of ES to additional family members, we established a molecular diagnosis in 5 out of 20 families (25%) who remained undiagnosed in our previously studied NDD cohort emanating from Turkey.

Keywords: Alu-Alu mediated rearrangement (AAMR); exome reanalysis; identity-by-descent (IBD); multilocus pathogenic variation; neurodevelopmental disorders; runs of homozygosity (ROH); whole-genome sequencing.

Figure 1

Family recruitment and summary result of exome sequencing (A) The workflow of ES analysis in this study: 234 newly enrolled families (TBM2 cohort, highlighted in pink) and 20 undiagnosed families from the TBM1 cohort (highlighted in blue) were analyzed (highlighted in gray boxes). (B) Bar plots showing the distribution of total AOH sizes, used as a surrogate measure for ROH, with no reported consanguinity (upper) and with reported consanguinity (lower). Blue and gray bars represent solved and unsolved individuals, respectively. A total AOH size of 100 Mb is marked by a red line. (C) A bar plot showing the molecular findings in the TBM2 study (234 families in total). Abbreviations are as follows: AOH, absence of heterozygosity; CNV, copy number variant; Comp Htz, compound-heterozygous; Hemi, hemizygous; Hmz, homozygous; and Htz, heterozygous; ROH, runs of homozygosity.

Figure 2

Genes identified in this study (A) Comprehensive gene list in alphabetical order. Genes that are identified in one family in this study are shown with bold letters (205 genes in total). (B) Venn diagram showing overlapping genes identified in the TBM1 and TBM2 cohorts. Note that the number (16) of overlapping genes in TBM1 and TBM2 is very small; this represents the wide genetic heterogeneity in NDDs. (C) Summary of genes categorized on the basis of the number of families reported in the literature. Genes that are identified both in TBM1 and TBM2 are in light green shaded boxes. Genes with red font color represent previously proposed candidate disease-trait-associated NDD genes from our group. Underlined text indicates genes identified in more than one family within the TBM2 cohort. The # symbol across several genes shows that there were reports of individuals with pathogenic variants with different modes of inheritance.

Figure 3

Segregation results, images, and protein conservation of the individuals with potentially pathogenic variants in ESAM and COPB1 (A) Pedigree structure of family BH13684 (HOU4550) and segregation of the frameshift variant in ESAM. The variant is predicted to be subject to degradation by nonsense-mediated decay (NMD). (B) Image of BAB12311. (C) Brain MRIs for BAB12311 showing severe cerebral atrophy and dilation of both ventricles. (D) Expression of ESAM in various tissues. All tissues noted in the figure are colored (e.g., the brain is colored as yellow). The graph data have been obtained from the Genotype-Tissue Expression (GTEx) Portal. (E) Pedigree structure of family BH11518 (HOU4031) and Sanger confirmation of a de novo COPB1 variant. (F) Image of BAB11010 showing synophrys, as well as a short and smooth philtrum. (G) Evolutionary conservation of the altered amino acid residues at position 347 is shown. (H) Expression of COPB1 in various tissues from the GTEx Portal.

Figure 4

Three families with homozygous exonic deletions identified in the TBM1 and TBM2 studies (A) A pedigree from a family with a homozygous exonic deletion (two exons) in BLM. The family was included with limited clinical and molecular data in our previous report. (B) BAB4987 at four years old. (C) A deletion plot in the BLM region constructed using ES data from BAB4987. The y axis represents the RPKM values on a log scale. The red line connects RPKM values at the exons in BAB4987, and the black line demonstrates the RPKM information from individual ES samples with similar experimental conditions. The blue dashed line represents the cut-off for homozygous deletion (0.21). (D) The Alu-Alu mediated rearrangement (AAMR) scores given by the AluAluCNVPredictor tool for BLM are 0.9 for OMIM genes (not shown) and 0.874 for RefSeq genes. The AAMR score is a potential measure of susceptibility to genomic instability based on Alu repetitive element sequence directly oriented pairs flanking exons for that given gene, and a score > 0.6 implicates relative susceptibility to AAMR. (E) The AOH lot for BAB4987 demonstrates the deletion is located within a 20 Mb AOH block on chromosome 15, marked by gray zones. (F) A schematic representation of a portion of BLM. Breakpoint sequence analysis for BLM deletion using the MultAlin alignment tool is also shown. The proximal reference sequence and proband breakpoint sequences that match the proximal reference sequence are shown in red, the distal reference sequence and proband breakpoint sequences that match the distal reference sequence are shown in blue, and microhomology at the junction is shown in green. The 6 Kb deletion in BAB4987 presumably results from Alu-Alu mediated rearrangement between directly oriented AluSx. (G) Pedigrees from two families with a homozygous exonic deletion in SNX14. (H) Deletion plots in the SNX14 region constructed from ES data from affected individuals (BAB3498 and BAB13562). (I) AOH plots for BAB3498 and BAB13562 demonstrate 34.8 and 6.3 Mb blocks of AOH on chromosome 6, marked by gray zones, and both show that SNX14 (red vertical line) is located within the AOH block. (J) Schematic representation of SNX14. The breakpoint junction of the 64 Kb 25-exon deletion in BAB3498 is blunt end, which suggests potentially non-homologous end joining (NHEJ) as a mechanism. The 17 Kb deletion involving 9 exons identified in BAB13562 presumably results from Alu-Alu mediated rearrangement between directly-oriented AluSz.

Figure 5

Digenic variants in ASTN1 and ASNT2 in an individual with NDD (A) Images of the proband BAB10738. (B and C) Axial (B) and sagittal (C) T2WI brain MRI showed severe hydrocephalus. (D) Log₂ ratio of read depth from short-read 40x WGS with vertical pedigree (left) aligned with personal genome data from the ASTN2 genomic region. The corresponding individuals (top to bottom) are BAB10739 (mother), BAB10740 (father), and BAB10738 (proband). The blue horizontal line reveals average depth of coverage consistent with normal N = 2 copy number state and heterozygous deletion; note the deletion is observed only in the proband panel (bottom) as a de novo variant allele. WGS data revealed a de novo 14.2 kb deletion of ASTN2 (NG_021409.2: g.119405571_119419749del [hg19]). (E) Log₁₀ reads count per kilobase of capture region, per million mapped reads (RPKM) from research exome sequencing using HMZDelFinder-generated data; the red line represents the proband RPKM value at the designated exon-capture region (blue rectangle), whereas black lines represent RPKM values from individual ES samples using similar experimental conditions. (F) The identified de novo single exonic deletion (exon 16) was independently confirmed by droplet digital PCR (ddPCR). (G) Canonical transcript (NM_014010.5) and two alternative transcripts (NM_198186.3 and NM_001184735.1) of ASTN2 (top) with the Sanger-confirmed junction sequence aligned with reference sequences at the bottom. The sequence in blue represents a junction sequence aligned with the proximal reference sequence, and the sequence in red represents the one aligned with the distal reference. Note the blunt-end junction and 16 bp deletion in proximity to the junction. (H) Pedigree and Sanger tracing showing a paternally inherited missense ASTN1 variant. (I) Evolutionary conservation of the altered amino acid residues at position 925 is shown.

Figure 6

Families potentially affected by multilocus pathogenic variation (A) Families affected with multilocus pathogenic variations (MPVs) in TBM2 (51 families). Blue, orange, and green bars represent known + known genes, known + possible disease-associated genes, and candidate disease-associated + candidate disease-associated genes, respectively. Note that Hmz + Hmz and Hmz + Hmz + Hmz represent ∼70% of families with MPV. Abbreviations are as follows: CNV, copy number variant; Comp Htz, compound-heterozygous; Hemi, hemizygous; and Hmz, homozygous. (B) Pedigree and Sanger confirmation of the LAMA1 and FSHR homozygous variants in family BH13142 (HOU4478). Abbreviations are as follows: DD/ID, developmental delay and intellectual disability and HH, hypogonadotropic hypogonadism. (C) Pedigree and Sanger confirmation of the identified de novo FOXG1 and TERF2 variants. Both FOXG1 and TERF2 variants possibly contributed to severe microcephaly seen in BAB12622. (D) Pedigree and Sanger segregation study of the homozygous ADSL and TNRC6B variants. ADHD, attention deficit hyperactivity disorder. (E) AOH plot showing variants in ADSL and TNRC6B are located with the same AOH region on chromosome 22. (F) Photographs for BAB8895 and BAB8898. (G) Brain MRI for BAB8895 at a year old showed mild cerebral atrophy. (H) Evolutionary conservation of the altered amino acid residues at positions 713 of TNRC6B and 426 of ADSL are shown.

Figure 7

Statistical analysis of individuals with single and/or multilocus pathogenic variation (A) Box plot showing the distribution of total AOH sizes in unsolved and solved individuals. The median total AOH sizes for unsolved and solved individuals are 153 and 223 Mb, respectively. ^∗∗ denotes p value < 0.01 (t test). (B) Box plot showing the distribution of total AOH sizes in (i) affected individuals with a single molecular diagnosis harboring either de novo, compound-heterozygous, or hemizygous variants; (ii) affected individuals with a single molecular diagnosis harboring homozygous variants; (iii) affected individuals with MPV with no homozygous variants; and (iv) affected individuals with MPV with homozygous variants. ^∗∗∗ denotes p value < 0.001 (t test). NS, not significant. (C) A heatmap demonstrating the distribution of total AOH sizes and AOH blocks harboring potentially pathogenic variants. (D) AOH plots on chromosome 13 in BH10203-1 and −4. Note that the AOH blocks around a SSH3 in two affected subjects showed a marked difference in size. (E) Pedigree and molecular findings in BH10204 (HOU3583). (F) Photo for the two affected individuals. Note that growth restriction is more severe in BAB9650, and BAB9651 had sparse hair. Both BAB9650 and BAB9651 have microcephaly. (G) AOH plots around variants in LRAP6, PLK4, and KIFC3. Note that the father (BAB9653) also has the homozygous LPAR6 variant.