A novel missense COL9A3 variant in a pedigree with multiple lumbar disc herniation

The proband in this study was a 14-year-old boy who had low back pain and numbness in the lower extremities for half a year. The father, who accompanied for his son’s consultation, had radiculopathy in the left lower extremity for over two decades. The proband’s paternal aunt and grandfather were also clinically diagnosed with symptomatic LDH, with available MR images for confirmation. No other skeletal abnormality was observed in this family on radiological images. Written consent for each family member was obtained. This study was approved by the authors’ institutional ethic board.

After blood sample collection, genomic DNA of the proband, his father, paternal aunt and grandfather were extracted using the Blood Genome Column Medium Extraction Kit (Kangweishiji, China). The extracted DNA samples were subjected to quality controlling using Qubit 2.0 fluorimeter and electrophoresis with 0.8% agarose gel. Whole exome library was constructed using Roche Nimble Gen Seq EZ Exome Enrichment Kit V2.0 and Seq EZ Exome Enrichment Kit V2.0 capture probes (Roche, USA). High-throughput sequencing was performed on a Novaseq 6000 instrument (CHIGENE, Beijing, China) [16]. Quality control of whole exome sequencing data, variants calling and variant annotation was performed in the same institution.

Variant prioritization was performed based on guidelines released by ACMG (The American College of Medical Genetics and Genomics). First, co-segregation analysis was performed to exclude SNVs contradictory to the phenotypic data. SNVs were analysed under the assumptions of Mendelian dominant inheritance, recessive inheritance and sex-linked inheritance. Second, only rare variants with MAF < 1% were included for further selection [17, 18]. Candidate variants were selected based on the 1000 Genomes (https://​www.​internationalgen​ome.​org), the Exome Sequencing Project (ESP), the Exome Aggregation Consortium (ExAC, http://​exac.​broadinstitute.​org), Allele Frequency Aggregator (ALFA) and the Genome Aggregation Database (gnomAD, https://​gnomad.​broadinstitute.​org). Third, synonymous variants and non-coding region variants were excluded. In silico prediction tools were applied to analyse pathogenicity of identified missense variants (Sorting Intolerant From Tolerant (SIFT), Polymorphism Phenotyping v2 (Polyphen2), Multivariate Analysis of Protein Polymorphism (MAPP), Mutation Taster, Mendelian Clinically Applicable Pathogenicity (M-CAP), Rare Exome Variant Ensemble Learner (REVEL) and Combined Annotation Dependent Depletion (CADD)) [19‐25]. The deleterious effects of splice variants were predicted by MaxEntScan and dbscSNV [26, 27]. Next, evolutionary conservatism was analysed by phastCONS, phyloP and Genetic Evolutionary Rate Profiling (GERP) [28‐30]. At last, protein function, GO (Gene ontology) annotations, tissue-specific distribution and existing literature were searched to evaluate the remaining SNVs.

Sanger sequencing was performed to validate the identified candidate variants. Sanger sequencing was performed with these primers:

Forward primer 5′-CAGGCGTCCCTGTGAGTATC-3′,

Reverse primer 5′-CATCAAGGCAACCAAATGCCA-3′.

The proband (height 1.65 m, weight 55.6 kg) is a 14-year-old Chinese boy suffering from low back pain and numbness in the lower limbs for 6 months before consultation. Magnetic resonance (MR) imaging revealed LDH at L4/5 and L5/S1 spinal segments (Fig. 1A). Growth and development of the proband were normal in adolescence. The proband did not experience waist trauma or excessive physical labour within 6 months before the onset of clinical symptoms. Blood biochemistry measurements and radiographs of limbs did not present any abnormality on the proband. A follow-up investigation of the LDH-related clinical symptoms and radiographs was carried out on the proband’s immediate family members. The proband’s father, who suffered from left lower limb radiculopathy for decades, had L3/4 and L4/5 LDH based on his lumbar spine MR images (Fig. 1B). The proband’s father, paternal aunt and grandfather had back pain and radicular leg pain, which were consistent with MR findings. They were clinically diagnosed with LDH at an age of 35, 30 and 23, respectively (Fig. 1C). None of them reported a history of waist injury before the onset of LDH‐related symptoms nor any other abnormality in the musculoskeletal system, except for the proband’s grandfather, who had degenerative kyphosis after 60 years old.

We performed WES from four members (proband III-I, affected father II-2 and paternal aunt II-1, and unaffected mother II-3) to identify the candidate gene for the phenotypic manifestation of LDH. The work flow is summarized in Fig. 1D. A total of 3624 SNVs were identified after exclusion of non-genic and polymorphic variants. SNVs which did not co-segregate with the phenotype and were more than 1% frequency in the public genomes databases were excluded. Under the assumption of Mendelian dominant inheritance, 179 SNVs were identified. Synonymous variants and variants in non-coding region were then excluded, resulting in 69 SNVs. Besides, several in silico prediction tools invariably predicted that seven of the SNVs might cause damage to the protein, and no SNV near splice site had deleterious effects on the protein. In view of the protein function, GO annotations, tissue-specific distribution and the facts in published literatures, a novel heterozygous missense variant (c.1150C > T, p.Arg384Trp) in COL9A3 was identified, and further confirmed by Sanger sequencing (Fig. 1E).

The MAF of this SNP was less than 0.01 in different databases, ESP (0.00), 1000 Genomes (0.00), ALFA (0.00), GnomAD_exome (0.00004) and ExAC (0.00003). This variant was predicted to be pathogenic by in silico prediction tools, Provean (3.16), SIFT (0.005), Polyphen2_HDIV (1.0), MutationTaster (0.999494), M-CAP (0.705) and REVEL (0.725). In addition, with a CADD score > 20, this variant was also evaluated to be deleterious in GERP, phyloP and phastCons software. According to the variant interpretation guideline of ACMG (PM2, PP1, PP2, PP3 and PP4), this variant (c.1150C > T) was classified as “likely pathogenic” variant [31].

The p.Arg384Trp variant is situated within the collagenous domain, resulting in the substitution of arginine with tryptophan in amino acid sequence (Fig. 2A). Protein sequence alignment revealed that this locus is highly conserved among common species (Fig. 2B). Furthermore, the impacts of the p.Arg384Trp variant on the structure, function, and stability of COL9A3 were analysed using Swiss-Model (Fig. 2C) [32‐34]. According to the prediction, the variant at this site changed the charge property and hydrophilicity of COL9A3 protein due to the substitution of arginine (Basic amino acid) with tryptophan (Aromatic hydrophobic amino acid). In addition, the prediction result in I-Mutant2.0 revealed a decreased stability of COL9A3 protein (Fig. 2D) [35].