Introduction

Nephronophthisis (NPHP) should be included in the differential diagnosis of children or adolescents presenting with chronic renal failure (CRF) of unknown etiology, because NPHP is the most common (accounting for ~3%) monogenic autosomal recessive cause of CRF in this age group.1, 2, 3 NPHP is clinically characterized by anemia and growth retardation due to impaired renal function as well as polyuria/nocturia and polydipsia due to decreased renal concentrating ability. Urinalysis often appears normal,4 and blood pressure is typically not high. Commonly, ultrasonography reveals normal-sized or relatively small kidneys that are distinctive from autosomal recessive or autosomal dominant polycystic kidney disease (ARPKD or ADPKD)5 and lack corticomedullary differentiation. NPHP is often accompanied by defects in other organs and tissues, for example, the retina, cerebellum and liver, such as in Joubert syndrome, Senior–Løken syndrome and Meckel–Gruber syndrome (MKS). A collective term, NPHP-related ciliopathy (NPHP-RC), is used to describe this group of diseases, because most of the causative genes in these disorders encode proteins that have a role in the cilium.6, 7, 8

A genetic diagnosis is required for a definitive diagnosis of NPHP-RC, because the clinical features of patients with NPHP-RC are rather non-specific, and their symptoms overlap significantly.8 The most common genetic cause of NPHP is a large deletion of NPHP1,9, 10, 11 which is noted in >20% of patients with NPHP, whereas other genes contribute less than 2–3%.7 Sanger sequencing of commonly mutated NPHP genes detects mutations in less than one-third of patients.8 In addition, the number of NPHP-RC-related genes has been rapidly increasing, with 20 genes having NPHP as part of their gene name at the time of writing, especially after the introduction of next-generation sequencing (NGS) techniques.12, 13, 14 Therefore, the genetic diagnosis of NPHP-RC based on an ‘educated guess at the best candidate gene’ is now changing to adopting high-throughput genome analysis techniques. Through NGS screening of known and candidate genes, a genetic diagnosis can be obtained efficiently in more patients, and novel causative genes can be identified. However, this technology is still evolving, and it is relatively costly and requires advanced bioinformatics support for handling large quantities of data. Therefore, the combination of both traditional Sanger sequencing and NGS might be more practical in obtaining a genetic diagnosis of NPHP-RC or other genetically heterogeneous groups of disorders.

We report here the use of a stratified two-step procedure of Sanger sequencing of selected genes followed by targeted exome sequencing of 34 disease-related genes, which led to a genetic diagnosis for 19 (34.5%) of our cohort of 55 Korean patients with NPHP-RC clinical diagnoses.

Materials and methods

Study population

This study was approved by the independent review board of Seoul National University Hospital (H-0812-002-264), and only those patients who provided written informed consent were screened and included in this study. Patients who were clinically diagnosed with NPHP were included.15 Presentation with incidentally identified CRF in the first three decades of life without evidence of previous renal damage was typical. Other causes of CRF were excluded by past medical history and imaging of the urinary tract in the majority of the cases. Patients with a clinical diagnosis of ADPKD were excluded. One patient (J-86) was included owing to typical findings of Joubert syndrome (cerebellar vermis aplasia, Leber congenital amaurosis and mental retardation) without renal involvement. We performed fundus examinations and abdominal sonography to evaluate the intra-abdominal organs. Renal ultrasonographic findings of increased echogenicity (normal or slightly decreased size for age) and corticomedullary differentiation loss were considered to be typical for NPHP-RC.5 For those individuals with developmental delay and neurological problems, we requested brain imaging. For some patients (n=15), renal pathology was also obtained, which exhibited typical findings of chronic tubular interstitial disease. This study adheres to the Declaration of Helsinki.

The patients were classified according to their age at the time of end-stage renal disease as infantile when younger than 5 years old (n=11) and juvenile when older (n=44). Combinations of NPHP with cerebellar vermis hypoplasia or aplasia were designated as Joubert syndrome (n=8). Retinal involvement with retinitis pigmentosa (RP) or Leber’s congenital amaurosis was designated as Senior–Løken syndrome (n=13), and those patients with multiple problems without cerebellar hypo/aplasia were designated as MKS-like (n=3).

Sanger sequencing

The most probable candidate genes were selected for Sanger sequencing based on known mutation frequencies and specific clinical phenotypes (Figure 1). Peripheral blood mononuclear cells were collected from the patients, and genomic DNA was obtained using a QIA amp DNA Blood Mini Kit (Qiagen, Hilden, Germany). All of the patients (n=55) were screened for large deletions of NPHP1, which is the most common mutation in NPHP, by amplifying each of the exons using PCR; a failure of amplification was considered to be a total homozygous deletion of the NPHP1 gene.10, 16 Among those patients without large deletions of NPHP1, infantile NPHP patients were tested for INVS/NPHP2 mutations (n=11).17 The selection of genes according to phenotype was as follows: IQCB1/NPHP5 for RP (n=19);18 CEP290/NPHP6 for Joubert syndrome (n=8);19 and TMEM67/MKS3 for hepatic fibrosis (n=7).20, 21

Figure 1
figure 1

Strategy for Sanger sequencing. All of the patients were screened for complete deletions of NPHP1, which is the most common mutation in NPHP. Among those without complete NPHP1 deletion, infant NPHP patients were tested for INVS/NPHP2 mutations. IQCB1/NPHP5, CEP290/NPHP6, and TMEM67/MKS3/NPHP11 were tested according to the phenotypes of the patients. Crbll, cerebellar; F, female; M, male; NPHP, nephronophthisis; RP, retinitis pigmentosa; SNHL, sensory neural hearing loss.

Full size image

Targeted exome sequencing

We designed customized targeted exome capture using a SeqCap EZ kit (Roche NimbleGen, Madison, WI, USA) for 34 genes related to NPHP-RC, Bardet–Biedl syndrome and ARPKD (Supplementary Table S1). After capturing the target sequences, the DNA libraries were amplified and sequenced using Illumina HiSeq2000 (Illumina, San Diego, CA, USA).

Alignment, coverage calculation and variant detection

Reads were aligned to the UCSC hg19 reference genome using BWA-0.6.1 with default settings22 for single-nucleotide variation (SNV)/insertion and deletion (indel) detection, and duplicate reads were removed. Variants were identified using Unified Genotyper in GATK-1.3.23 A Perl script and Annovar were used to annotate variants and search the known SNPs and indels from dbSNP v135 and the 1000 Genomes project data (drafted February 2012). Coverage and depth were calculated using the GATK DepthOfCoverage analysis. The significance of variants was assessed in silico using MutationTaster, PolyPhen-2, SIFT and FATHMM. Those variants predicted to be disease causing or not tolerated by two or more programs were considered to be pathogenic mutations.24, 25, 26, 27 Conservation scores (Phylo-P) for each mutated nucleotide were also considered.28 Prediction of alternative splicing by intronic variation was performed using NetGene2 (http://www.cbs.dtu.dk/services/NetGene2/), Ex-skip (http://ex-skip.img.cas.cz/), and BDGP acceptor site prediction (http://www.fruitfly.org/seq_tools/other.html).

Variant calls were obtained using the following filter parameter: minimum base quality of 17 (the default value for the GATK Unified Genotyper). A minimum variant count of 10 was applied for potential truncating mutations (nonsense, frameshift and obligatory splice-site mutations), and non-synonymous missense variants were filtered by a minimum count of 3. Synonymous variants and common dbSNP (v135) variants with a population allele frequency >1% were excluded.

All of the candidate mutations were validated by Sanger sequencing. When available, we verified their co-segregation in the family (J-61 and J-50).

Results

Clinical features of NPHP-RC patients

We recruited 55 unrelated Korean patients with a clinical diagnosis of NPHP (M:F=34:21) who were referred to our laboratory for genetic diagnosis (Supplementary Table S1). The mean age of the patients at end-stage renal disease was 9.0±5.1 years old (median, 8.3 years; range, 0.6–17.9 years). Renal histology from 15 patients exhibited typical findings of tubulointerstitial nephropathy. Nineteen patients (34%) had eye involvement of RP or Leber’s congenital amaurosis, eight patients (14.5%) had a molar tooth sign in the brainstem on brain imaging and seven patients (12.7%) exhibited hepatic fibrosis. Eight patients (14.5%) had siblings with similar symptoms; however, none of their parents had CRF.

First step: Sanger sequencing for genetic diagnosis

As a first step in the stratified approach to determine a genetic diagnosis of NPHP, we selected and screened the patients by Sanger sequencing for candidate genes (Figure 1). A homozygous deletion of NPHP1 was detected in four patients (7.3% (confidence interval (CI, 95%) 0.4–14.2%), and 17% of patients exhibited juvenile isolated NPHP without other organ involvement (n=23); Table 1, in bold). Three (15.8%) of 19 NPHP patients with RP or Leber’s congenital amaurosis (Senior- Løken syndrome, marked with ♣ in Figure 1) carried a common pathogenic homozygous mutation (c.1522_1523dupGA, p.Ala509Lysfs*3) in the IQCB1/NPHP5 genes.29 In screening NPHP patients with cerebellar vermis hypotrophy/atrophy involvement (n=8, marked with ♦ in Figure 1), one familial case of Joubert syndrome had a compound heterozygous mutation in CEP290/NPHP6, a known pathogenic c.1666delA, p.Ile556Phefs*17,30 and an SNV (c.6011-12T>A) that is predicted to cause alternative splicing.31 Four of seven NPHP patients with hepatic fibrosis (marked with ♣ in Figure 1) carried compound heterozygous mutations in TMEM67/MKS3/JBSTS6, a frameshift, truncating mutation and an SNV.32, 33, 34 All four of these patients had Joubert syndrome and congenital hepatic fibrosis compatible with COACH (cerebellar vermis hypoplasia/aplasia, oligophrenia, congenital ataxia, ocular coloboma and hepatic fibrosis) syndrome.35 In total, Sanger sequencing detected genetic aberrations in 12 patients (21.8%) with a clinical diagnosis of NPHP-RC.

Table 1 Pathogenic mutations found in Korean NPHP-RC patients by Sanger sequencing and targeted exome sequencing
Full size table

Second step: Targeted exome sequencing in patients with NPHP-RC

Targeted exome sequencing of 34 NPHP-RC-related genes was applied in 43 patients in whom a genetic diagnosis was not obtained by our first step of Sanger sequencing. We obtained an average 21.8 Mb of mapped data per individual for ~166 kb of targeting and flanking regions (Supplementary Table S2). The mean read depth was 131±55 and 95.4% of the captured target exons exhibited >10-fold coverage. Most likely azpathogenic mutations of various types were identified in 7 of 43 cases (16%) in six genes, including NPHP1, NPHP3, NPHP4, SDCCAG8/NPHP10, TTC21B/NPHP12 and PKHD1 (Figure 2). In three cases (K-8, J-84, K-9, Table 2), a heterozygous known pathogenic mutation in BBS4 was identified that could not explain the genetic pathogenesis but might explain a mild phenotype resembling BBS, as previously reported.36, 37 Heterozygous potentially damaging mutations/variations in at least one of the target NPHP-RC genes were identified in additional 15 patients, including one truncating mutation of NPHP3 in two patients with isolated juvenile NPHP. In the remaining 18 patients, no significant variations of the target genes were identified. In total, we identified most likely pathogenic mutations in nine NPHP-RC-related genes in 19 (34.5%) of 55 Asian NPHP-RC patients using a two-step genetic diagnosis (Figure 2), where the frequency of homozygous deletion of NPHP1 is relatively low (7.3%, CI 0.4–14.2).

Figure 2
figure 2

Results of genetic diagnosis for NPHP-RC. One-third (n=19; 34.5%) of the patients with clinical diagnoses of NPHP-RC obtained a genetic diagnosis by two-step genetic diagnosis using Sanger sequencing (n=12, 21.8%) and targeted exome sequencing (n=7; 12.7%). Four patients with homozygous total deletion of NPHP1, three with IQCB1/NPHP5, one with CEP290/NPHP6, and four with TMEM67/MKS3/NPHP11 were detected using Sanger sequencing. Mutations of other genes were detected using targeted exome sequencing. In addition, heterozygous mutations in NPHP-RC genes were detected in 13 patients (23.6%). NPHP-RC, nephronophthisis-related ciliopathy.

Full size image
Table 2 Probably pathogenic variants in NPHP-RC patients
Full size table

Discussion

The genetic causes of several Mendelian diseases, such as NPHP-RC, RP, and non-syndromic hearing loss, are heterogeneous. For these diseases, providing a precise diagnosis is often difficult until pathogenic mutations are identified. In the present study, we obtained a genetic diagnosis in one-third of patients with a clinical diagnosis of NPHP-RC with a two-step genetic diagnosis using Sanger sequencing followed by high-throughput mutation analysis using NGS after a custom DNA-capture procedure. For two patients (K-7 and O-463), mutation analysis led to correction of their diagnoses from NPHP to ARPKD. Patients with ARPKD may have normal-sized kidneys and therefore may be misdiagnosed with NPHP, as shown here. In addition, additional heterozygous mutations were identified in candidate genes from NGS (data not shown), implying that further study of the respective genes or other closely related genes would enhance the efficacy of genetic diagnosis of NPHP-RC. For the definitive genetic diagnosis of NPHP-RC patients, extensive analysis of trio or family studies of NPHP-RC will be followed to understand the penetrance of genetic alterations and recurrence in additional patients. Systemic functional studies on the variant proteins will also be required to understand the effects of genetic alterations.

Traditionally, genetic diagnosis has been obtained by Sanger sequencing of the best candidate genes based on disease phenotype and frequency. The same approach was applied for our NPHP-RC patients as our first step, and 22% of the patients were given genetic diagnosis with a fair genotype–phenotype correlation. All three patients with congenital blindness due to Leber’s congenital amaurosis were shown to have an identical homozygous indel mutation in IQCB1/NPHP5 (5.5% (CI 0–11.5%) of the total population), suggesting a founder effect. Four patients with Joubert syndrome, RP, hepatic fibrosis and developmental delay were found to have TMEM67/MKS3/JBSTS6 mutations (7.3% (CI 0.4–11.5%) of the total population). These four patients had a similar appearance to each other, with rectangular faces, square jaws, and amiable natures despite mental retardation. Another interesting finding at this step was that the proportion of NPHP1 mutations in these patients was relatively lower than reported.7, 8, 12 It is not clear whether this difference reflects ethnic characteristics of the study populations or is simply derived from a selection bias of this study given the small number of participants. Nonetheless, a majority of the patients were not given a definitive diagnosis, similar to that reported in the literature.7, 8, 12, 38 Those with an atypical phenotype would have been misclassified in this step, and their causative genes would not have been assessed. Therefore, we introduced NGS technology as our second step.

For better cost-effectiveness, we chose targeted exome sequencing (TES) instead of whole-genome sequencing or whole-exome sequencing. TES for disease-related genes with high read depth enables multiplex screening of candidate genes.38 In addition to a well-defined set of known NPHP-RC genes known at the time of study design, we included an extended set of cilia-related genes in our TES (Supplementary Table S2), including ARPKD and genes of interest to the authors. Among the variants detected in our second step of genetic diagnosis, only those that might explain the phenotype were selected as the most likely pathogenic mutations (n=13, Table 2); four known pathogenic mutations, five frameshift or truncating mutations and four missense mutations were predicted to be damaging. The possibility of large deletions or duplications was also considered, and total deletion of one NPHP1 allele was detected in one patient (J-39) in whom the other NPHP1 allele exhibited a heterozygous frame shift mutation. Other than this case (J-39), no genetic diagnosis was made involving the five genes selected for our first step. Given that those with known genetic defects were excluded at this second step, the efficiency of TES could not be assessed. Overall, this second step of genetic diagnosis mapped genetic aberrations in known NPHP-RC genes or PKHD1 in 16% (7 of 43) of the patients. Our results reflect a similar mutation-detection rate to that reported by Halbritter et al.38, where a molecular diagnosis was obtained in 12% of patients using high-throughput mutation screening of 13 NPHP genes in a large population of NPHP-RC patients.

This study has several shortcomings. The number of the patients is small compared with previous reports,38, 39 and segregation analysis was not available in the majority of the cases due to the inability to obtain a sample from the parents. In addition, functional studies of ‘most likely pathogenic’ novel mutations have not been performed to date. In addition, given that the coverage of TES of this study was not perfect (Supplementary Table S3), a second mutation might not be sequenced, thus warranting Sanger sequencing of the respective genes with single candidate variants, Unfortunately, further analysis was not possible in this study because those with single variants were lost to follow-up. Above all, the set of genes used for TES reflects our knowledge at the time of study design; therefore, newly discovered genes, such as ZNF423,40 WDR19,41 ANKS6,42 IFT172,43 CEP83 (ref. 44) and DCDC2,45 were not included. On the basis of this study, we are currently developing our next set of TES by incorporating the recent discovery of NPHP-RC. Improvements in NSG technology are expected to increase the rate of genetic diagnosis.

Although the number of patients of this study was not sufficiently large to determine the distribution of genetic aberration types and loci, this finding helps to delineate the characteristics of Korean or Asian NPHP, which enables a more efficient genetic diagnosis of NPHP-RC in this population. On the basis of this study, we are currently assessing the genetic aberrations of our NPHP-RC patients as follows. Gel electrophoresis of PCR products of NPHP1 is performed as the first step (step 1), and then one or two particular mutations or genes are analyzed if the patient has distinctive extra-renal findings, such as congenital blindness with c.1523_1524insGA of IQCB1/NPHP5 or COACH syndrome with a TMEM67/MKS3/JBSTS6 mutation (step 2). If genetic diagnosis is not obtained with these first two steps, then the TES steps in (step 3) might be replaced by whole-exome sequencing or whole-genome sequencing in the near future. With expanding collective knowledge and rapidly evolving technology, both steps 2 and 3 are expected to simultaneously become more diverse and more precise, thus yielding a better genetic diagnosis of NPHP-RC.