Targeted gene panel for genetic testing of south Indian children with steroid resistant nephrotic syndrome

The Institutional Ethics Committee approved the study and all participants were recruited after informed consent. Twenty five children with idiopathic SRNS (18 males: 7 females) as defined by standard guidelines were included [18]. Socio demographic information, clinical and treatment details were recorded in case record forms. All these children were previously analyzed by Sanger sequencing for all the exons of NPHS2 and exon 8 and 9 of WT1 genes [7, 14].We also included three subjects with pathogenic mutations in NPHS2 reported previously to determine the sensitivity of the targeted-NGS method [7]. Three healthy individuals were included to check sequencing efficiency.

Blood samples (5 ml) were collected from recruited patients and genomic DNA was extracted from peripheral blood leukocytes by the phenol chloroform method [19]. Quantity of the extracted DNA was estimated using Qubit fluorometric assay (Thermofisher scientific, MA, USA).

For targeted next-generation sequencing, we selected a panel of 17 genes associated with SRNS based on their prevalence in clinically diagnosed SRNS patients and mutation frequency in the NS cohorts (Table 1) [4, 5, 20]. The genes selected for the panel accounted for 95–100% of the mutations in two large cohorts of SRNS one of which included Indian children [6, 21]. A total of 359 primers targeting the exonic regions of the selected 17 genes (307 exons) associated with nephrotic syndrome were designed using Ion Ampliseq Designer (Life Technologies, CA, USA). The amplicon size was designed in a range from 125 to 375 bp. The panel consisted of three primer pools amplicon size ranging from 125 to 375 bp and covering 99.6% exon of the selected genes. The uncovered region was mainly repeat rich region making primer designing difficult. An Ion Torrent adapter-ligated library was prepared using the Ion AmpliSeq Library Kit 2.0 (Life Technologies, CA, USA) by following the manufacturer’s protocol. Briefly, 10 ng of DNA was amplified by PCR using the premixed primer pool and Ion AmpliSeq HiFi master mix. After PCR, the amplified targets were treated with FuPa reagent to partially digest primer sequences and phosphorylate the amplicons. For adaptor ligation, amplicons from each sample were combined with a barcode adapter mix that contained Ion P1 adaptor and a unique Ion Xpress Barcode (Life Technologies, CA, USA). The unamplified libraries were purified using AMPure beads (Beckman Coulter, CA, USA) and the purified beads were amplified using Platinum PCR SuperMix High Fidelity and Library Amplification Primer Mix (Life Technologies, CA, USA). The amplified library was purified using AMPure beads. Library quantity and quality was determined using Qubit fluorometric assay and Agilent BioAnalyzer High-Sensitivity DNA kit (Agilent Technologies, CA, USA), respectively.

Template preparation, emulsion PCR, and Ion Sphere Particles (ISP) enrichment were done using the Ion PGM Template OT2 400 kit (Life Technologies, CA, USA) according to the manufacturer’s instructions. Next-generation sequencing was carried out on Ion Torrent Personal Genome Machine sequencer (Life Technologies, CA, USA) using the Ion 318 and 314 Chips (Life Technologies, CA, USA) and Ion PGM Hi-Q Sequencing Kit (Life Technologies, CA, USA) according to the manufacturer’s instructions.

The data from the both sequencing runs were analyzed using the Torrent Suite V5 analysis pipeline. Sequence reads were separated according to their barcodes. Human genome sequence (build GRCh37/hg19) was used as a reference sequence. For each individual barcode, the sequence reads were aligned to this reference sequence with a Torrent Mapping Alignment Program optimized to Ion torrent data using the default alignment algorithm and parameters. After alignment, the variants were annotated to determine their clinical significance by using a combination of frequency, structural prediction, or evidence-based data. The DNA variant regions were piled up with Torrent Variant Caller (TVC) plug-in software to identify missense, nonsense, frameshift, obligatory splice variants and short insertion/deletion (indels) across the targeted subset of the reference using germ-line parameters and low stringency settings. The output variant call format (VCF) file was then annotated using Ion Reporter Software v5.0 (Life Technologies, CA, USA) and variants were further investigated. All the variants were filtered based on their coverage (coverage> 30), variant effect (non-synonymous, frameshift, nonsense), location (to detect splice site variants) and allele frequency in public databases (ExAc (http://​exac.​broadinstitute.​org/​), and 5000 Exome (http://​evs.​gs.​washington.​edu/​EVS/​) < 1%). The filtered variants were visually examined using Integrative Genomics Viewer (IGV) software (http//www.​broadinstitute.​org/​igv), to further filter out variants with possible strand-bias and variants within homopolymeric region. In silico analysis using Sorting Tolerant From Intolerant (SIFT) and Polymorphism Phenotyping v2 (Polyphen-2) tools was performed to predict the potential deleterious effect of the identified missense variants on protein function [22, 23]. Bioinformatics analysis of the strength of predicted splice site variants was performed with neural networks (NNSPLICE 0.9) [24]. The variants were classified as pathogenic, likely pathogenic, uncertain significance, likely benign, or benign according to the stringent criteria of American College of Medical Genetics and Genomics (ACMG) Standards and Guidelines and Sherloc rules [25, 26]. A scoring system developed by Karbassi et al. was used to determine the pathogenicity of VUS identified in this study [27].

The pathogenic and likely pathogenic variants were validated by Sanger sequencing using variant specific primers in patients as well as in healthy individuals (n = 30) (Additional file 1: Table S1).

The clinical details of 25 SRNS patients are presented in Table 2 with detailed phenotyping in Additional file 1: Table S2. The median age of onset of NS was 2.5 years- (0.58–16 years) with a median follow up of 2.5 years.. Majority of the patients were non-responsive to non steroidal immunosuppressant, with only 8 children demonstrating partial response to calcineurin inhibitors (Additional file 1: Table S2).

Two sequencing runs, containing 25 samples (23 patients and 2 healthy individual sample; 318 chip) and 4 samples (2 patients, 1 healthy individual sample and one human standard CEPH DNA sample; 314 chip) were performed. Total 854 M (Q20) and 172 M of Q20 data were obtained per 318 and 314 chips respectively and the coverage was comparable between runs. After filtering out polyclonal, low quality reads, and primer-dimers, the percentage of usable reads were 4.57 M and 0.788 M per 318 and 314 chips respectively (Additional file 1: Table S3). Combining the data derived from two runs, sequencing of the 17 glomerular disease gene panel generated a mean of 0.18 M reads per individual with mean read length of 214 bp. Only 10% of called bases had a quality score of <Q20; About 99% of these reads were mapped to the reference genome (hg19) and 93.9% of mapped reads were on target genes (Additional file 1: Table S4). A mean coverage of 442× was achieved for the genes across all individuals, with 93.1, 63.2 and 17% of the targets having minimum read depth of 20×, 100× and 500× respectively.

Overall, 2916 single-nucleotide variants (SNVs) and indels were identified in the 25 patients and 3 healthy individuals by Torrent Suite software V5, using default germline parameters. These variants were annotated and filtered using the Ion Reporter Software 4.4 with following parameters: inclusion of frameshift, stop loss, missense, nonsense variants and variants located in splice site with a minimum coverage of 20×. After the filtration, a total of 26 variants (23 missense, 2 nonsense and 1 splice site) were identified in 13 genes in 16 subjects (Fig. 1). Among these variants, 1 pathogenic NPSH2 (R71X), 3 likely pathogenic [PLCe1 (R752X), NPHS1 (G968 V) and NPHS2 (splice site variant, g .179521737C > T)] and 2 VUS (LMX1B (V145 M) and NPHS2 (H141Y) were considered clinically relevant. The remaining 20 variants not considered further for annotation included 15 heterozygous VUS in genes with recessive inheritance, two VUS (P973T and P995L) in MYO1E gene in a single patient (SRNS 60) in cis and a likely benign variant (R877Q) in INF2 gene. A homozygous VUS in PLCe1 (G222R) gene in SRNS was also excluded from further annotation, as it was observed in a healthy individual. A variant in ACTN4 gene (R310Q) was excluded from clinically relevant list although it was classified as likely pathogenic based on ACMG criteria. This variant has a very low allele frequency in ExAC database and also has been reported in probands of families with FSGS and individuals with sporadic FSGS [0.0074 (8/1084) controls 0.016 (3/192) sporadic FSGS] [28]. Besides, podocyte transient transfection assay indicates that the mutation inhibited the complex formation between α-actinin-4 and CLP36 causing the podocyte defect although the precise pathways involved were not identified [29]. However, a large number of alternate alleles (n = 1426) have been identified at the same position in general population Although global allele frequency of p.R310Q variant in ACTN4 was < 1%, total allele count was higher (3138) in gnomAD database (Updated version of ExAC, http://​gnomad.​broadinstitute.​org/​variant/​19-39207742-G-A). As per the Sherloc rule (EV0161, https://​www.​ncbi.​nlm.​nih.​gov/​pubmed/​28492532), variants with allele count > 8, is considered as high allele count and the variant is classified as benign. Therefore although the based on the ACMG criteria p.R310Q variant in ACTN4 was classified as pathogenic, it was considered as benign based on the improved and robust variant classification guidelines of Sherloc.

The pathogenic variant in NPSH2 (R71X) gene was observed in a pair of sibling (8%). The likely pathogenic variants in PLCe1 (R752X), NPHS1 (G968 V) and NPHS2 (splice site variant, g .179521737C > T) genes were identified in one subject each (4%) (Table 3). All these variants were novel except for the NPHS2 variant (R71X) [30]. Of the 2 variants, identified by Sangers sequencing previously, one variant (H141Y) was not present in the final filtered variants. A review of the NGS data indicated that the variant was identified by the sequencing but was filtered because of the stringent variant filtration settings (minimum read depth of 30×). A total of 18 reads was obtained for this variant H141Y in NPHS2. The pathogenic and likely pathogenic variants were validated using Sanger sequencing in the respective patients and 30 healthy individual samples and no false positives were identified. The pathogenicity score of two variants (LMX1B; heterozygous, V145 M, andNPHS2; homozygous, H141Y) classified as variants of uncertain significance (VUS) indicated that they could be pathogenic in nature and needed to be explored further for their causality in SRNS (Additional file 1: Table S5).

The clinical features and the renal histology were similar between those with pathogenic or likely pathogenic variants. Response to immunosuppressive treatments was not significantly different between those with pathogenic or likely pathogenic variants and those without an abnormal variant. The homozygous nonsense R71X variant in NPHS2 gene was identified in two siblings (patient SRNS20 and SRNS76). The siblings presented with NS at age of 3.5 and 2.5 years respectively and both showed FSGS on biopsy. Both subjects showed no response to calcineurin inhibitors. The elder sibling (patient SRNS20) progressed to ESRD by the age of 5 years and died at the age of 6 years with sepsis. The younger sibling currently aged 4.5 years (patient SRNS76) is in CKD stage 3 (Additional file 1: Table S2). Their parents were heterozygous for the point mutation (data not shown). A likely pathogenic homozygous nonsense variant R752X in PLCe1 gene was identified in SRNS123 in whom renal biopsy showed DMS (Additional file 1: Table S3). This child presented with symptoms of NS at the age of 1.5 years had renal dysfunction at the time of diagnosis and progressed to ESRD within a year of diagnosis. Similar histopathology has been reported with pathogenic variants in PLCe1 gene [31]. A splice site likely pathogenic variant was observed in NPHS2 gene in patient 83 who also had a homozygous VUS in the same gene (H141Y missense, both parents are heterozygous for this particular variant). This child manifested SRNS at the age of 1.2 years, showed FSGS on renal biopsy and progressed to ESRD, 2 years after diagnosis. SRNS 13 was identified to have the homozygous recessive variant (G968 V) in the NPHS1 gene. The child was diagnosed as SRNS at age of 10 months, with the biopsy report revealing MHC and is in remission at last follow up. Patient 73 in whom a heterozygous variant in LMX1B (dominant inheritance) was observed manifested SRNS at the age of 3.5 years with FSGS on biopsy and progressed to ESRD within 7 years of diagnosis. The risk score suggests pathogenicity.