Identification of novel PKD1 and PKD2 mutations in a Chinese population with autosomal dominant polycystic kidney disease

Autosomal dominant polycystic kidney disease (ADPKD) is one of the most frequently inherited renal diseases caused by mutations in PKD1 and PKD2. We performed mutational analyses of PKD genes in 49 unrelated patients using direct PCR-sequencing and multiplex ligation-dependent probe amplification (MLPA) for PKD1 and PKD2. RT-PCR analysis was also performed in a family with a novel PKD2 splicing mutation. Disease-causing mutations were identified in 44 (89.8%) of the patients: 42 (95.5%) of the patients showed mutations in PKD1, and 2 (4.5%) showed mutations in PKD2. Ten nonsense, 17 frameshift, 4 splicing and one in-frame mutation were found in 32 of the patients. Large rearrangements were found in 3 patients, and missense mutations were found in 9 patients. Approximately 61.4% (27/44) of the mutations are first reported with a known mutation rate of 38.6%. RNA analysis of a novel PKD2 mutation (c.595_595 + 14delGGTAAGAGCGCGCGA) suggested monoallelic expression of the wild-type allele. Furthermore, patients with PKD1-truncating mutations reached end-stage renal disease (ESRD) earlier than patients with non-truncating mutations (47 ± 3.522 years vs. 59 ± 11.687 years, P = 0.016). The mutation screening of PKD genes in Chinese ADPKD patients will enrich our mutation database and significantly contribute to improve genetic counselling for ADPKD patients.

Scientific RepoRts | 5:17468 | DOI: 10.1038/srep17468 ADPKD is a heterogeneous monogenic disorder resulting from mutations in two genes: PKD1 and PKD2. The PKD1 gene (OMIM 601313), located in chromosome region 16p13.3, consists of 46 exons with an open reading frame of 12,912 bp that encodes the 4,303-amino acid peptide polycystin-1 (PC1) 4 . Exons 1-33 of PKD1 are duplicated approximately six times at the homologous genes (HGs), which has made the genetic analysis of PKD1 challenging 5 . PKD2 (MIM 173910, chromosome region 4q21-22) is a single-copy gene including 15 exons with a 2,907-bp coding sequence and is predicted to encode a 968-amino acid peptide called polycystin-2 (PC2) 6 . PC1 and PC2 act as flow-dependent mechanosensors in renal primary cilium that regulate the differentiated state of tubular epithelial cells 7 . Clinical data show that mutations in PKD1 and PKD2 account for 85% and 15% of all cases of ADPKD, respectively 8 . Compared to PKD2 mutations (where the average age of ESRD onset is 74 years), PKD1 mutations (where the average age of ESRD onset is 54.3 years) are associated with the more serious form of the disease 9 .
Diagnosis of ADPKD is performed mainly by renal ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI) but cannot exclude the disease in at-risk individuals until the age of 40 years, especially in families with mutations in PKD2 10 . Molecular diagnostics play a significant role in confirming a definite diagnosis, especially in young renal donors, patients with a negative family history, individuals with early onset ADPKD or atypical symptoms and for subjects with affected relatives 11 . There is no hotspot mutation for PKD1 or PKD2, indicating that mutations are usually unique to a single family and are highly variable 12 . The structural complexity of PKD1 and the high allelic heterogeneity of PKD genes make clinical molecular diagnostics difficult 13 . Although a few studies on novel mutations in the PKD1 and PKD2 genes in Chinese patients have been carried out with different methods, including PCR-single-strand conformation polymorphism (PCR-SSCP) 14 , denaturing high-performance liquid chromatography (DHPLC) 15 and next-generation sequencing (NGS) 16 , direct sequencing is undoubtedly the gold standard for accurately identifying the majority of PKD mutations 17 . Multiplex ligation-dependent probe amplification (MLPA) was developed to detect large genomic rearrangements in PKD genes that cannot be detected by sequencing 18 .
Analysis of the pathogenicity of variants of an uncertain significance plays an important role in the molecular diagnosis of ADPKD because of the high level of genetic variation found in the PKD1 gene 19  In the present study, we performed long-range PCR (LR-PCR) followed by nested PCR and MLPA of PKD1 and PKD2 in 49 Chinese patients with a definite diagnosis of ADPKD. A group of novel mutations in PKD1 and PKD2 is described in this paper. All mutation data detected will contribute to better diagnostics and genetic counselling in a clinical setting.

Results
We performed complete mutational analysis by direct sequencing and MLPA analysis of PKD1 and PKD2 in 49 unrelated patients with the diagnosis of ADPKD obtained by ultrasound. Thirty-two definitely pathogenic variants and 12 likely pathogenic variants (42 variants in PKD1 and 2 in PKD2) were found in 44 patients (Table 1) variant was considered to be nonpathogenic. Figure 1 shows the distribution of definitely pathogenic and likely pathogenic variants found in the PKD1 gene. We further confirmed that there are no variant hotspots in PKD1 gene.

Definite pathogenic mutations.
Definite pathogenic mutations were found in 32 of the families and included 10 nonsense mutations, 17 frameshift mutations, 2 splicing mutations, and 3 large rearrangements. These disease-causing mutations are shown in Table 2. The percentage of our ADPKD patients without a family history was 12.2% (6/49). We analysed PKD1 and PKD2 in the parents of six probands with no family history of ADPKD and found that in five instances, the pathogenic mutations occurred de novo in the probands. No pathogenic mutation of PKD1 or PKD2 was found in the other patient without a family history ADPKD.
RT-PCR analysis of PKD2 mRNA in patient G0904. A novel splicing mutation in PKD2 (c.595_595 + 14delGGTAAGAGCGCGCGA) was found in patient G0904 and was predicted to affect the splice site of the gene. The predicted absence of PKD2 exon 2 would produce a premature termination codon (PTC) downstream in exon 3 and therefore could not escape nonsense-mediated mRNA decay (NMD) 20,21 . To evaluate the influence of the c.595_595 + 14del, RT-PCR analysis of total RNA extracted from peripheral blood mononuclear cells (PBMCs) from patient G0904 and a healthy control was performed using the primers PKD2-E1F/PKD2-E2R and PKD2-E1F/PKD2-E4R. It is worth noting that we could not detect aberrant PKD2 transcripts in patient G0904 (Fig. S1A). However,  Table 3 for details). heterozygosity of rs2728118 (c.420G > A) in exon 1 of PKD2 was only found in the genomic DNA (GA) and not in the cDNA of the patient (AA) (Fig. S1C). rs2728118 was inherited from the patient's father, and the c.595_595 + 14del mutation was inherited from the patient's mother (Fig. S1B). rs2728118 is located upstream of the c.595_595 + 14del mutation and is distributed in trans on the chromosome. Further quantitative PCR (qPCR) analysis indicated that the patient PKD2 mRNA level was reduced to 55.97 ± 1.78% (P = 5.44 × 10 −5 ) compared to the normalized control mRNA level (Fig. S1D). These results suggested monoallelic expression of the wild-type allele, inheritance form the patient's father and nonsense-mediated mRNA decay (NMD) of the aberrantly spliced PKD2 transcripts.
Likely pathogenic mutations. A total of 22 unclassified variants and 12 previously reported polymorphisms (Table S1) were detected in our patients. We evaluated the pathogenic potential of the unclassified variants using web-based prediction programs. The results of the evaluation are presented in Table 3. Twelve likely pathogenic mutations (11 PKD1 mutations and 1 PKD2 missense mutation) were identified in 12 patients. Among these mutations, one small novel in-frame mutation in PKD1  (c.8157-8159delCAC) showed high evolutionary conservation, and two novel splicing variants in PKD1 (c.10220 + 2T > C and c.10617-1delG) were predicted to affect the splice site and were found to segregate with the disease in affected families. Therefore, these three mutations in PKD1 are highly likely to be pathogenic mutations. Two novel substitutions PKD1 (p.R3046C and p.Y3819N) coexisted in patient G0241 and were predicted in silico to be likely pathogenic and likely polymorphic, respectively. Amino acid multi-alignments demonstrated that the positions of the mutations are highly conserved across species for both p.R3046C and p.Y3819N (Fig. 2). Therefore, we speculate that p.R3046C and p.Y3819N are hypomorphic alleles. Segregation of the two variants with the disease was not validated due to the lack of blood samples from the families.
Large deletion mutation. To identify large deletions or duplication mutations that cannot be detected by Sanger sequencing, we performed a copy number analysis of the probands without pathogenic mutations in the PKD genes (Fig. S2). Using MLPA, three large deletions of the PKD1 gene were found in patients G0677, G0018 and G1800. Deletion of PKD1 exon 1 (relative peak ratio 0.52) was found in patient G0677, a 45-year-old man, and a similar result was found in his affected family members. Deletion of PKD1 exon 21 (relative peak ratio 0.51) was identified in patient G1800 and his affected  Table 3. Evaluation of the pathogenic potential of variants. Abbreviations: ID, identification; NA, not analysed; MG, mutation group; PKDB, PKD mutation database; B: highly likely pathogenic; C: likely pathogenic; I: indeterminate; N: neutral; D: likely polymorphic; E: polymorphic. a The mutation has been previously classified to be "highly likely pathogenic". b The mutation has been previously classified to be "indeterminate". c Polycystin orthologue alignment (human, mouse, rat, chicken, X. tropicalis, fugu, dog, opossum and tetraodon).
mother. Patient G1800 is 26 years old and was found to have bilateral renal cysts at approximately 18 years of age. The renal function of patient G1800 is well controlled except for slight hypertension. DNA sequence analysis of exons 1 and 21 demonstrated the absence of single-base mutations under the oligonucleotide probe. Patient G0018, who is 20 years old and without a family history of ADPKD, showed a deletion for probes 1 to 30. Large deletion mutations segregated with the disease in all of the affected family members tested, although q-PCR confirmation was not performed at the nucleotide level due to the sequence complexity of the PKD1 locus.
We therefore performed a Kaplan-Meier survival curve analysis to investigate whether the type of PKD1 mutation (non-truncating mutations including missense and small in-frame mutations vs. truncating mutations) influenced the age of ESRD onset. As shown in Fig. 3, the age of ESRD onset in patients with PKD1-truncating mutations (n = 23) was earlier than that of the patients with non-truncating PKD1 mutations (n = 9) (log-rank test, P = 0.016). The median age of ESRD onset in the patients with PKD1-truncating mutations and patients with non-truncating mutations was 47 years (95% CI, 47 ± 3.522 years) and 59 years (95% CI, 59 ± 11.687 years), respectively.  analysis of PKD1 and PKD2 is needed for Chinese patients with ADPKD who require a genetic diagnosis. The 27 novel mutations we found in the Chinese population will enrich the PKD mutation database and significantly contribute to the genetic counselling of ADPKD patients. Patients with PKD2 mutations typically develop ESRD two decades later than those with mutations in PKD1 9 . Therefore, it is of prognostic value to determine the location of the mutation in an affected family    (PKD1 or PKD2). The mutational distribution in the PKD2 gene could not be determined because only two exonic mutations were found. The mutation detection rate of PKD2 (4.1%) in our study was much lower than the average percentage (15%) 8 . This difference may be due to the low number of patients analysed in our study compared to previous studies by Rossetti et al. Previous studies have shown that the mutation type and location in PKD1 may influence renal survival 23 . We detected a total of 44 pathogenic variations in the PKD1 gene; of these, 9 (20.9%) are located in exon 15, which is consistent with the findings recorded in the PKDB (277/1,272). This region corresponds to the junction of the PKD repeats and the REJ domain of the resultant protein. In our study, 78.6% of the mutations in PKD1 were predicted to truncate the protein, including frameshift mutations, nonsense mutations, splicing mutations, and large deletions. The high frequency of these mutations is in concordance with recent results from Cornec-Le Gall et al., who showed that approximately two thirds of PKD1 mutation-positive pedigrees carry truncating mutations 24 . Cornec-Le Gall et al. reported that carriers of a PKD1-truncating mutation have a significantly earlier age of ESRD onset than patients with a non-truncating mutation (55 years vs. 67 years). Our data support the view that a more severe phenotype can be expected in patients with a PKD1-truncating mutation. The exception was patient G0599, who carried the known missense mutation p.C155Y but also had received a right renal transplant at 26 years old.

Discussion
Two novel PKD1 mutations (p.R3046C and p.Y3819N) were found in a female patient with ADPKD, and it is probable that both mutations are hypomorphic alleles. However, DNA samples from the parents of this patient were not available. Moreover, the likely pathogenic mutations identified in our study should be confirmed in future studies including additional ADPKD families.
Large genomic rearrangements account for approximately 6.8% of the pathogenic mutations in the PKD1 gene and an even smaller percentage of PKD2 gene mutations 18 . In our study, we identified three large deletions: one involving deletion of exon 21, one involving deletion of exons 1 to 30, and a large deletion of exon 1 25 . Our data are consistent with previously reported data.
The high level of allelic heterogeneity in both the PKD1 and PKD2 genes and the prevalence of private mutations in ADPKD patients imply that there is a high frequency of de novo mutations in this disease. Indeed, approximately 10% of adult ADPKD patients do not have a family history of the disease 26 . The percentage of our ADPKD patients without a family history of disease was 12.2%, which is consistent with previously reported results. Five different pathogenic mutations occurred de novo in probands without a family history of ADPKD.
Pathogenic mutations were not found in 10.2% of our 49 unrelated patients, a result in accordance with those reported in a previous CRISP study (10.9%) 8 . This finding may be because mutations occur within deep intronic regions, as well as in promoters and other distantly located regulatory regions not covered by the current exon-based sequencing method. Alternatively, some of the missense mutations that are classified as nonpathogenic mutations by the prediction system may represent hypomorphic alleles. Such variants alone may result in only mild cystic disease, but two such variants in trans may cause disease 27,28 . Additionally, we only screened PKD1 and PKD2 mutations and therefore cannot exclude the existence of other mutations that might contribute to the cystic phenotype, such as those in HNF1b, PRKCSH, SEC63 or PKHD1 29 . Furthermore, mosaicism may influence the genotype and phenotype of ADPKD; however, this condition is usually not detected in screenings, and a significant proportion of de novo mosaic mutations may be missed 30,31 . The likely pathogenic mutations identified in our study need to be confirmed in future studies with additional ADPKD families.

Conclusions
In our study, 27 novel pathogenic mutations in the PKD genes were detected in 49 Chinese individuals. A novel splicing mutation in PKD2 (c.595_595 + 14delGGTAAGAGCGCGCGA) was confirmed to be definitely pathogenic. Patients carrying PKD1 mutations, especially those with truncating mutations, could have a more rapidly progressive disease than those with non-truncating mutations. Our study will enrich the mutation database of PKD and significantly contribute to the genetic counselling of ADPKD patients.

Methods
Patients. A total of 49 unrelated patients were enrolled from the Women's Hospital School of Medicine Zhejiang University from October 2010 to December 2013. Patients were definitively diagnosed with ADPKD based on the criteria recommended by Ravine D et al. 32 . All patients provided written informed consent, and their family and medical histories were recorded. The general clinical data of the probands are summarized in Table 5. Peripheral blood samples were collected from all probands and their family members when possible. The study was performed with the approval of the Ethics Committee of the Women's Hospital School of Medicine Zhejiang University. The study was conducted in adherence to the Declaration of Helsinki.
Mutation analysis of PKD1 and PKD2. Genomic DNA was extracted from peripheral blood samples using the QIAGEN spin columns on a QIACube (QIAGEN GmbH) according to the manufacturer's instructions. The mutational screening of PKD1 and PKD2 via Sanger sequencing was carried out in the 49 probands. LR-PCR followed by nested PCR was adopted for the mutational analysis of the PKD1 gene 5   RT-PCR analysis of PKD2 mRNA in PBMCs. Total RNA from PBMCs was extracted and reverse transcribed from patient G0904 and a healthy control. RT-PCR was performed with primers PKD2-E1F/ PKD2-E2R and PKD2-E1F/PKD2-E4R. qPCR was performed with primers PKD2-E13F/PKD2-E14R, and the results were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH), according to the method described previously 36 . RT-PCR and qPCR primers are listed in Table S3.

Data analysis and sequence variation classification. Mutations of PKD genes were analysed using
Mutation Surveyor ® software. Nucleotide changes were nominated according to the NCBI reference sequences of PKD1 (NM_001009944.2) and PKD2 (NM_000297.3). The HGMD (http://www.hgmd.cf.ac. uk), the Exome Sequencing Project (http://evs.gs.washington.edu/EVS) and PKDB (http://pkdb.mayo. edu) were checked for previously reported sequence changes. Novel mutations in this study were assessed for their pathogenic potential. Nonsense or frameshift variants resulting in a PTC were identified to be definitely pathogenic. The pathogenicity of missense variants was computationally evaluated with the SIFT, PolyPhen2 and AlignGVGD prediction programs by analysing interspecies sequence variations [37][38][39] . The NetGene2 (http://www.cbs.dtu.dk/services/NetGene2/) 40 and Human Splicing Finder (HSF) (http:// www.umd.be/HSF/) 41 software were used to evaluate the splice site mutations. All variations analysed by these web-based software programs were finally sorted into four categories: likely pathogenic, indeterminate, likely polymorphic and polymorphic. Only gene variations that were predicted to be damaging by SIFT, PolyPhen-2 and AlignGVGD were considered to be "likely pathogenic", as long as no other definite mutation was found in the same patient. If a definite mutation coexisted with a damaging missense mutation in the same patient, the missense mutation was considered to be "indeterminate". Similarly, only the variations that were scored as benign by all the software programs were considered to be "polymorphic". Otherwise, the mutations were classified as "likely polymorphic". Furthermore, pedigree co-segregation analysis of the potential pathogenic mutations in PKD genes was examined in all available members of the probands' families (including healthy individuals). Likely pathogenic missense or splice site variants would segregate within the affected family.  Statistical analysis. Cumulative renal survival curves were generated using the Kaplan-Meier method and compared using the log-rank test. P < 0.05 was considered to be statistically significant. All analyses were performed using SPSS 17.0.