Exome sequencing of Saudi Arabian patients with ADPKD

Abstract Purpose: Autosomal dominant polycystic kidney disease (ADPKD) is characterized by progressive development of kidney cysts and enlargement and dysfunction of the kidneys. The Consortium of Radiologic Imaging Studies of the Polycystic Kidney Disease (CRISP) cohort revealed that 89.1% had either a PKD1 or PKD2 mutation. Of the CRISP patients with a genetic cause detected, mutations in PKD1 accounted for 85%, while mutations in the PKD2 accounted for the remaining 15%. Here, we report exome sequencing of 16 Saudi patients diagnosed with ADPKD and 16 ethnically matched controls. Methods: Exome sequencing was performed using combinatorial probe-anchor synthesis and improved DNA Nanoballs technology on BGISEQ-500 sequencers (BGI, China) using the BGI Exome V4 (59 Mb) Kit. Identified variants were validated with Sanger sequencing. Results: With the exception of GC-rich exon 1, we obtained excellent coverage of PKD1 (mean read depth = 88) including both duplicated and non-duplicated regions. Of nine patients with typical ADPKD presentations (bilateral symmetrical kidney involvement, positive family history, concordant imaging, and kidney function), four had protein truncating PKD1 mutations, one had a PKD1 missense mutation, and one had a PKD2 mutation. These variants have not been previously observed in the Saudi population. In seven clinically diagnosed ADPKD cases but with atypical features, no PKD1 or PKD2 mutations were identified, but rare predicted pathogenic heterozygous variants were found in cystogenic candidate genes including PKHD1, PKD1L3, EGF, CFTR, and TSC2. Conclusions: Mutations in PKD1 and PKD2 are the most common cause of ADPKD in Saudi patients with typical ADPKD. Abbreviations: ADPKD: Autosomal dominant polycystic kidney disease; CFTR: Cystic fibrosis transmembrane conductance regulator; EGF: Epidermal growth factor; MCIC: Mayo Clinic Imaging Classification; PKD: Polycystic kidney disease; TSC2: Tuberous sclerosis complex 2


Introduction
Polycystic kidney disease (PKD) is the most common inherited multi-systemic disease characterized by the progressive development of kidney cysts with consequent enlargement of the kidney and progression toward end-stage renal disease (ESRD) [1,2]. Kidney cysts range in size from microscopic to many centimeters in diameter, leading to inflammation, regional ischemia, cytokine release, tubular obstruction, and subsequent loss of kidney function. Two forms of PKD include autosomal dominant polycystic kidney disease (ADPKD) and autosomal recessive polycystic kidney disease (ARPKD) that are typically distinguishable by their pattern of inheritance, as well as age of onset, hepatic fibrosis, arterial hypertension, kidney morphology, and cyst location [3]. ADPKD is the most prevalent inherited form of kidney disease with an incidence of 1:500 to 1:1000 individuals and is observed in approximately 7%-10% of patients with ESRD [4,5]. Extra-renal formation of cysts, mainly in the liver and pancreas, abnormalities in connective tissues, and aortic and intracranial aneurysms have been reported in ADPKD patients [6]. Hypertension is prevalent among ADPKD patients and usually precedes loss of glomerular filtration rate [7]. Additional cystic kidney phenotypes exist that could be misdiagnosed as ADPKD including multiple simple cysts, autosomal dominant tubulointerstitial nephritis, tuberous sclerosis complex, medullary sponge kidney, HNF1B nephropathy, nephronopthisis, and others [8]. Genetic testing could be useful to clarify the diagnosis in patients with features inconsistent with typical ADPKD.
Mutations in two genes, PKD1 and PKD2, are implicated in ADPKD, but despite comprehensive screening, up to 15% of suspected ADPKD patients have no mutation detected. Mutations in two additional genes, GANAB and DNAJB11, may account for an additional 1% of ADPKD patients [9,10]. PKD1 codes for a large integral glycoprotein, polycystin 1, with an extracellular and intracellular domain suggesting that it plays a role in cell-to-cell interaction and in cellular signaling [11]. PKD2 encodes a smaller transmembrane protein, polycystin 2, which appears to be a divalent cation channel; however, this theory has been put into question by a recent study [12]. In the Consortium of Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) cohort, 89.1% had either a PKD1 or PKD2 mutation. Of CRISP patients with a genetic cause detected, mutations in PKD1 accounted for 85%, while mutations in the PKD2 accounted for the remaining 15% [11]. In clinically ascertained cohorts, the proportion of PKD2 mutations appears greater, representing 30% of patients [13]. Over 2,300 pathogenic PKD1 and 300 pathogenic PKD2 germline mutations have been identified, though primarily in European populations [12]. Although mutations in PKD1 and PKD2 appear fully penetrant, variable expressivity exists with disease severity varying between family members with the same mutation, implying the possibility of additional modifier genes [14].
In the last ten years, exome sequencing of 180,000 exons has been proven to be a powerful technique for the identification of rare genetic variants. The PKD1 gene contains 46 exons with a large duplicated region spanning exons 1 to 33 [12]. The duplicated regions share a high degree of sequence identity with six pseudogenes adjacent to PKD1 on chromosome 16. PKD2 is a relatively smaller gene with 15 exons [12]. Mutation screening of PKD1 with long-range polymerase chain reaction (PCR) to avoid the pseudogene sequence is technically challenging, laborious, and costly [15]. Recent efforts have successfully employed highthroughput targeted gene panel sequencing-based mutation screening of PKD1, while others have reported poor read depth and coverage of PKD1 with exomebased approaches [16,17]. We attempted exome sequencing in 16 clinically diagnosed ADPKD patients and 16 normal controls from the Eastern Province of Saudi Arabia to identify mutations leading to the development of ADPKD in our population.

Patient selection
This case-control study included 16 ADPKD probands ascertained in the nephrology clinic at King Fahd Hospital of the University, Al-Khobar and King Fahd Military Medical Complex, Dhahran, in the Eastern Province of Saudi Arabia. Patient demographic and clinical characteristics are provided in Table 1. We attempted to exclude all patients with syndromic causes of multiple renal cysts, such as tuberous sclerosis, von Hippel-Lindau disease, and familial polythelia with multiple renal cysts from the study. Clinical diagnosis of ADPKD was obtained using computerized tomography (CT) imaging ( Figure 1). Estimated glomerular function (eGFR) was calculated from serum creatinine using the CKD-EPI equation. ADPKD diagnosis was based upon the unified Pei-Ravine diagnostic criteria when family history was present [18]. The presence of multiple cysts with dilated collecting ducts, enlarged renal outlines, increased renal echogenicity, loss of corticomedullar differentiation, and positive family history was documented. In three patients, it was not possible to confirm whether there was a positive family history of ADPKD due to unavailability of parental medical records. Total kidney volume was calculated using the ellipsoid equation, and Mayo Clinic imaging classification (MCIC) was calculated using their web-based tool [19]. Patients were evaluated for typical features of ADPKD including the following: bilateral symmetrical involvement of both kidneys (i.e., MCIC class 1), the presence of positive family history, and concordance between kidney imaging and rate of loss of kidney function. A total of 16 healthy controls with no family history of ADPKD or related conditions were randomly selected from the Eastern Province population. This study was approved by the Ethical Committee of Imam Abdulrahman Bin Faisal University in accordance with the 1964 Helsinki Declaration and its later amendments. Informed written consent in English, with a verified translation in Arabic, was obtained from all participants in accordance with the Institutional Review Board (IRB # 2014-01-274).

Exome sequencing and data analysis
Peripheral blood samples were collected in EDTA tubes and stored at 4 C before extraction of genomic DNA using Gentra Puregene Blood kits (Qiagen, USA) according to the manufacturer's protocol. BGISEQ-500 Paired Library preparation was performed in accordance with the instruction protocol (BGI, China) described elsewhere [20]. We performed whole exome sequencing of all samples on the combinatorial probe-anchor synthesis and improved DNA Nanoballs technology-based BGISEQ-500 sequencers (BGI, China) using the BGI Exome V4 (59 Mb) Kit (BGI, China). Exome data were mapped to the human reference genome (NCBI build 37.1, UCSC hg19) using Burrows-Wheeler Aligner (BWA-MEM, version 0.7.10) [21]. Variant calling was performed using Genome Analysis Tool Kit (GATK, version 3.3) after removal of the duplicate reads by Picard [22]. All variants were annotated by ANNOVAR and Variant Effect Predictor (VEP) [23]. Variants with a read depth > 20 were extracted for further analysis. The variants which were identified through the above pipeline were further filtered to eliminate benign variants with MAF (minor allele frequency)>0.1% in the 1,000 Genomes Project, dbSNP, exome aggregation consortium (ExAC), ESP-6500, and internal BGI databases containing exome data on >100 000 subjects [24][25][26]. Finally, the candidate variants were checked against the Human Gene Mutation Database (HGMD), the Mayo PKD Mutation Database, and Clinvar, if mutations had been previously published [27][28][29].

Validation by Sanger sequencing
Sanger sequencing was used to validate all mutations identified through high-throughput exome sequencing (Supplementary Figure 1). PCR primers were designed with Primer3. The first step in the PCR was performed at 95 C for 5 min, then followed by 30 cycles of denaturation at 95 C for 30 s, annealing at 55 C for 30 s, extension at 72 C for 30 s, and a final extension at 72 C for 7 min. The PCR products were sequenced using ABI 3730 xl DNA Analyzer, and the results were compared with the reference sequence (UCSC hg19).

Results
The majority of recruited patients were male (69%), with a mean age of 46 years and a mean age of diagnosis of 34 years. The mean age of the female patients was 50 years, with a mean age of diagnosis of 36 years. Hypertension was reported in nine of the total patient number. Nine of the 16 patients had typical ADPKD with MCIC class 1 imaging findings, positive family history, and consistent MCIC and kidney function. Seven patients were found to have atypical ADPKD features including the following: Six of these seven patients had evidence of renal atrophy with reduced eGFR (P2, P9, P11, P12, P15, and P16), and three of the patients had no known family history (P8, P15, and P16). After reviewing the radiological liver images of patients, no cysts were found to be present in the liver. Exome sequencing was performed in 16 patients clinically diagnosed with ADPKD and 16 ethnically matched controls. Of the 59 million bases captured, approximately 50%, 62.8%, 77.3%, 90.5%, and 96.7% of bases were sequenced at a depth of at least 200Â, 150Â, 100Â, 50Â, and 20Â, respectively. The mean read depth for PKD1 was 88 (standard deviation ¼ 49), with adequate coverage of all exons of the duplicated and non-duplicated region with the exception of GC-rich exon 1 (Supplementary Table 1).

Mutations identified in patients with ADPKD
In the 16 clinically diagnosed Saudi Arabian ADPKD patients, four PKD1 mutations were identified in five patients and one PKD2 mutation was identified in one patient. PKD1 mutations included three protein truncating mutations (p.Gln3206 Ã , p.Q3514 Ã , and p.Arg2163 Ã ) and one missense mutation (p.Val4146Ile). All observed mutations were located within the duplicated region and were validated by Sanger sequencing. The PKD2 mutation was a protein truncating mutation (p.Arg464 Ã ). PKD1 and PKD2 mutations were identified in six of nine (66.6%) patients with typical ADPKD presentations, while no PKD1 or PKD2 mutations were found in the seven patients with an atypical ADPKD  Six of the seven patients had no PKD1 or PKD2 mutation detected, but with a predicted pathogenic variant observed in genes involved in cystic pathways, including PKHD1, PKD1L3, CFTR, EGF, and TSC2, and four had atypical presentations with small kidney size and eGFR decline (P11, P12, P15, and P16). One patient carried a heterozygous PKHD1 missense mutation (p.Arg1909Trp), one patient carried a PKD1L3 nonsense mutation (p.Ser1168 Ã ), two patients carried a CTFR missense mutation (p.Ala120Thr), one patient carried a EGF missense mutation (p.Gly366Asp), and one patient carried a TSC2 missense mutation (p.Ala985Val). Medical records of patients carrying these cyst-pathway genes did not display evidence of lung or liver disease. Progressive kidney disease was present in all the atypical patients, with five of the seven patients having CKD stage 5 and the remaining two having CKD stage 3 or lower. The patient with a rare predicted pathogenic PKHD1 variant had been diagnosed with ADPKD based on CT imaging and a strong positive family history of the disease including the patient's mother, three brothers, and daughter. The patient with a PKD1L3 variant had a family history of the disease and had progressed to CKD stage 5 but the patient did not display evidence of liver disease. No patients were compound digenic heterozygotes, meaning no patient carried more than one rare variant in a cyst-pathway gene. Of the four ADPKD patients with no mutation detected in PKD1, PKD2, or any cystpathway gene, two were at CKD stage 5 but had small atrophic kidneys (P2 and P9), one was at stage 3 with typical ADPKD features (P4), and one had very mild disease with CKD stage 1 at age 74 (P8). Pedigree analysis revealed that 13 of the 16 ADPKD patients had a positive family history of ADPKD. It was not possible to confirm whether or not there was a positive family history in patients P8, P15, and P16 due to the unavailability of parental medical records as these patients had moved into the Eastern Province from rural areas in southern Saudi Arabia. While these three patients did not have a PKD1 or PKD2 mutation, P15 carried the CFTR mutation and P16 carried the TSC2 mutation. No mutations were identified in ADPKD or cyst pathway genes in the control group.
Patients with PKD1 and PKD2 mutations demonstrated an earlier age of diagnosis (mean 31.7 ± 11.2 years vs. 43.0 ± 14.3 years), a higher total kidney volume, and higher MCIC, while patients with no PKD1 or PKD2 mutation detected had significantly older age of diagnosis, smaller total kidney volume, and lower risk MCIC.
Based on available clinical and genetic parameters (the presence of PKD1 or PKD2 mutation) the 'predicting renal outcomes in ADPKD' (PROPKD) score was calculated to predict the risk of ESRD and determine the median annual reduction of eGFR (Table 1). Of the six patients with either a PKD1 or PKD2 mutation, two patients were low risk, two were intermediate, and two were at high risk of a decline in kidney function.

Discussion
ADPKD is the most prevalent inherited form of kidney disease and is observed in approximately 7-10% of patients with end-stage renal disease. Variable expressivity is common in ADPKD, with age at presentation and kidney disease severity greatly varying depending on the mutated gene, the mutation class, and the influence of other modifying genes [11]. Over 2,300 PKD1 and 300 PKD2 germline mutations have been identified in different populations, and the list is growing [12,30]. Clinical diagnosis of ADPKD is based upon the presence of kidney cysts on imaging in the context of a positive family history [31], typically with ultrasound, but also by CT or MRI. Positive family history is an important condition as it raises the pretest probability of ADPKD to 50%. Kidney enlargement is also an important consideration in ADPKD diagnosis, but is not a part of current diagnostic criteria. Moreover, diagnosis using imaging criteria alone is difficult in young patients (age < 30) as cysts and renal enlargement may not yet be apparent. Genetic diagnostics is thus especially valuable in evaluating young patients and those patients with no family history of ADPKD. Cases with equivocal or unusual presentations, early-onset, or severe presentations, and for improving risk stratification are emerging indications for genetic testing [31,32]. Furthermore, molecular diagnosis of ADPKD may contribute to family planning decisions.
Our study found PKD1 and PKD2 mutations in six of the nine (66.6%) patients with typical ADPKD features. No PKD1 or PKD2 mutation was detected in the seven patients with atypical features of ADPKD, calling the diagnosis into question. The Toronto Genetic Epidemiology Study of PKD (TGESP), a clinically ascertained cohort including primarily ADPKD patients of European descent, identified 60% (131 of 220 families) with PKD1 mutations and 26% (57 of 220 families) with PKD2 mutations using long-range PCR and Sanger sequencing [32]. Kinoshita and colleagues found putative pathological PKD1 and PKD2 mutations in 89% (90 of 101) of Japanese ADPKD patients with a long-range PCR and next-generation sequencing platform [33]. The lower rate of mutation detection in the current study is either due to technical artifacts leading to missed mutations in PKD1 or PKD2 or the presence of ADPKD phenocopies incorrectly diagnosed as ADPKD [34]. Somatic mosaicism is another potential cause for unidentified mutations, especially in the context of a patient without a family history [14]. While there have been reports of success using targeted gene panel sequencing in ADPKD, a recent report indicated low read depth and call quality over PKD1 exons in the duplicated region using the Agilent TruSeq and SureSelect kits and Illumina HiSeq2000 platform [15][16][17]. We observed adequate read depth through both duplicated and non-duplicated regions of PKD1 using the BGI pipeline, and all identified variants were validated by Sanger sequencing, but further validation to ensure the sensitivity of this approach is required. Exome sequencing technology is likely to continue to improve, but may be replaced with whole-genome sequencing strategies in the near future [35].
While previous reports have identified other variants of PKD1 and PKD2 in the Saudi population, we are reporting here for the first time five variants, namely c.12436G>A, c. 9616 C>T, c.6487C>T, c.10540C>T in PKD1, and c.1390C>T in PKD2 in the Saudi population [36,37]. These variants have been previously identified in European populations [38,39]. Audr ezet et al. [14] reported a number of variants in PKD1 and PKD2 in French and Tunisian populations. Mutations in PKD1 and PKD2 are associated with variable renal survival times in ADPKD patients [36,40]. The age of onset of ESRD in patients with PKD1 mutations is approximately 58 years, whereas in patients with PKD2 mutations, the age of onset is approximately 79 years. The age of onset of ESRD is also influenced by the type of PKD1 mutation itself [36,40,41].
Protein-truncating PKD1 mutations lead to a younger age at onset of ESRD than carriers of non-truncating mutations. In the present study, the four patients with a truncating PKD1 mutation had a more severe kidney phenotype than the patient with a non-truncating mutation. The c.5725C>T mutation in the PKHD1 gene, which is usually associated with ARPKD, has been reported in the Saudi population in association with ARPKD [42]. The subject carrying this mutation has a very strong family history of ADPKD (five members had ADPKD) and was at CKD stage 3. This patient did not present with any other organ disease. The c.3503C>G mutation in the PKD1L3 gene, which is usually associated with sour ageusia, was identified in one subject. This subject carrying this mutation had one family member who had been diagnosed with ADPKD but the patient himself did not present with any other organ disease.
Numerous pathways have been implicated in cyst pathogenesis and progression. Epidermal growth factor (EGF) is a small mitogenic protein involved in normal cellular proliferation, acceleration of fluid secretion, renal electrolyte homeostasis, and expansion of renal cysts [43]. EGF protein is highly expressed in cyst epithelia of ADPKD patients, and primary cultures of cyst epithelia are hyperesponsive to mitogenic stimulation by EGF, which may indicate that EGF plays a role in promoting cyst formation [44]. One ADPKD patient in the present cohort was found to have a mutation in the EGF gene; however, the influence of this mutation on the phenotype of the disease could not be confirmed.
CFTR and purinergic channels in the apical membrane of the renal tubules mediate the conduction of chloride and bicarbonate across the plasma membrane, and it has been postulated that CFTR plays a role in cyst growth where inhibitors of CFTR retard cyst enlargement in animal model [45][46][47]. Moreover, it has also been postulated that inhibitors of CFTR may slow down the progression of polycystic kidney disease [45,46]. Two patients in the present cohort were found to have a mutation in the CFTR gene but the causative mutation of ADPKD in these two patients could not be identified.
Tuberous sclerosis complex (TSC) is an autosomal dominant disease characterized by facial angiofibromas, benign brain tumors, seizures, developmental delay, cardiac rhabdomyomas, pulmonary lymphangioleiomyomatosis, and renal angiomyolipomas [48]. TSC is caused by a pathogenic variant in either TSC1 or TSC2. The TSC2 gene is located adjacent to PKD1 on chromosome 16p13.3. A contiguous deletion syndrome, where both TSC2 and PKD1 genes are deleted, has been reported to lead to a severe phenotype including signs and symptoms of TSC and PKD [49][50][51]. One patient in the present cohort was found to have a likely pathogenic variant in the TSC2 gene, but no classical ADPKD mutation. This patient did not show any signs or symptoms of TSC in the brain, heart, lung, eye, or kidney.
One limitation of this study was that obtaining pedigree analysis of some patients was not possible as these patients had migrated to the Eastern Province from isolated rural areas in other provinces of Saudi Arabia where such information would be impossible to obtain. Furthermore, the patients themselves were unable to provide such information.

Conclusions
Exome sequencing of DNA obtained from Saudi patients who had typical ADPKD revealed that mutations in PKD1 and PKD2 are the most common cause of typical ADPKD in this population. We also observed four rare variants that have the potential to contribute to an ADPKD phenotype. Further examination of exome sequencing techniques is required before clinical application of this technique in ADPKD diagnosis. However, genetic testing for ADPKD can improve diagnostic precision, informed prognosis, and support family planning.