Novel mutations of PKD genes in the Czech population with autosomal dominant polycystic kidney disease

The mutational screening of PKD genes was performed in a set of 56 unrelated Czech patients with ADPKD diagnosed by renal ultrasound in accordance with previously described criteria [26]. The study was approved by the Ethics Committee of General University Hospital in Prague on May 23, 2006. All patients provided written informed consent. According to the linkage analysis, the disease was linked to the PKD1 gene in 20 individuals. In 26 patients a relation of the disease to the PKD1 gene was not proved and 10 patients had an isolated occurrence of the disease within a family.

Linkage analysis was carried out using a panel of CA-repeat flanking markers. The combination and number of markers used for analysis were selected to achieve at least two informative markers, one on the 5′end and the second on the 3′end of PKD gene to rule out possible recombination. The markers used for PKD1 analysis were: D16S521, D16S3024, KG8, CW4, CW3, CW2, and for PKD2 analysis: D4S231, D4S1534, JSTG3, JV108, AICA1, D4S1563 and D4S414.

The genomic DNA was isolated from the peripheral blood lymphocytes using QIAGEN spin columns on a QIAcube (QIAGEN GmbH).

At first, the screening of the PKD1 gene was carried out in the selected set of 56 patients. For mutational analysis of the PKD1 gene, the technique of a long-range PCR (LR-PCR) followed by nested PCR was used [27]. Nested PCR amplicons were produced and analyzed by the method of High Resolution Melting (HRM); positive clones were reanalyzed and confirmed by sequencing in both directions. The protocol of the whole strategy (including primers and PCR reaction conditions) was previously described in detail by Reiterova et al. [16].

The DNA of patients where no definite pathogenic mutation was found within the PKD1 gene (29 samples) were subsequently analyzed via mutational screening of the PKD2 gene by HRM and via MLPA for large rearrangements of both PKD1 and PKD2 genes. This subset of 29 individuals included patients without any sequence changes found by mutational analysis of the PKD1 gene as well as patients with unpublished putative missense mutations, putative splice-site mutations and in-frame deletions/insertions detected in the PKD1 gene.

The coding region and intron-exon boundaries of the PKD2 gene were screened via 17 PCR fragments. The sequences of PCR primers (some primers were designed on the base of Tan et al.) and the conditions of PCR reactions are available upon request [28]. PCR fragments were produced and analyzed by HRM using the LightCycler® 480 (Roche Applied Science). The reaction mixture (total volume of 10 μl) contained 1 μl of diluted template DNA, 1 × LightCycler® 480 High Resolution Melting Master Mix (Roche Applied Science) and 5 pmol of primers. The LightCycler® 480 Gene Scanning Software (Roche Applied Science) analyses the melting profile of samples to identify changes in the shape of melting curves, which indicate both homozygous and heterozygous allelic variants in a sample (Figure 1) [29]. All PCR fragments showing aberrant melting curves by HRM were sequenced. The sequencing reaction was performed according to the manufacturer’s instructions using the BigDye® Terminator v1.1 Cycle Sequencing Kits (Applied Biosystems) and ran on an ABI PRISM® 310 Genetic Analyser (Applied Biosystems).

MLPA analysis (MRC-Holland) was carried out with SALSA MLPA KIT P351-B1/P352-B1 PKD1-PKD2, which together covers 37 from 46 exons of the PKD1 gene, its upstream regulatory sequences, entire PKD2 gene and three exons of the TSC2 gene adjacent to the PKD1 gene [30]. The MLPA reaction was performed with 50–100 ng of genomic DNA according to the manufacturer’s instructions. The result of MLPA was carried out by fragmentation analysis with the use of ABI PRISM® 310 Genetic Analyser (Applied Biosystems). To obtain the final results, the generated raw data were normalized according to the manufacturer’s instructions. At first intra-sample normalization was performed by dividing the peak area of each probe’s amplification product by the total area of only the reference probes in the probemix. Subsequently, the normalization between individual samples was done by dividing the intra-normalized probe ratio of each sample by the average intra-normalized probe ratio of reference samples (negative controls). In-house developed method was used for the entry of all these calculations.

All sequenced samples were compared to the following NCBI reference sequences: [GeneBank: NM_001009944.2] for the PKD1 gene and [GeneBank: NM_000297.3] for the PKD2 gene. The sequence changes found within this study were checked with the list of PKD sequence variants currently published in the Human Gene Mutation Database (HGMD) (http://​www.​hgmd.​cf.​ac.​uk) and the Autosomal Dominant Polycystic Kidney Disease Mutation Database (PKDB) (http://​pkdb.​mayo.​edu). In case of novel mutations, their pathogenic potential was assessed. Nonsense or frameshift variants leading to a premature STOP codon were considered as definitely pathogenic.

The pathogenic potential of missense variants was evaluated computationally by examining interspecies sequence variations using PolyPhen2 (http://​genetics.​bwh.​harvard.​edu/​pph2/​; [4, 31, 32]) and Sorting Intolerant from Tolerant (SIFT) (http://​sift.​bii.​a-star.​edu.​sg/​; [4, 32, 33]) software. The sequence variant was considered as likely pathogenic, if both scores assessed its pathogenity. When only one test predicted the pathogenic potential, the variation was reported as indeterminate. The pathogenic relevance of putative splice site mutations was evaluated in the same way using NetGene2 (http://​www.​cbs.​dtu.​dk/​services/​NetGene2/​; [34]) and Human Splicing Finder (HSF) (http://​www.​umd.​be/​HSF/​; [35]) software. Further criteria for classification of missense or splice site variant as likely pathogenic mutation were its segregation within the affected family and absence of additional, more probable mutation in a concrete patient (family).

Possible influence of sex and type of mutations on the age of renal failure was evaluated. For this purpose, Shapiro-Wilk normality test for each group was performed and subsequently, according to results of normality test, unpaired two-sample t-test was used to compare the groups. All computations were carried out using in-house developed scripts in R language.

One patient (275) from our study was found to have two truncating pathogenic mutations (one nonsense mutation and one frameshift mutation), both already published and both located within the exon 23 probably in cis regarding his mild phenotype of ADPKD without progression. Similar cases have been already described in the literature, but their existence is rare [11]. The patients 394, 395 and 427 are the representatives of the group with isolated occurrence of ADPKD within their families.

Via our mutational analysis of both PKD genes, 9 likely pathogenic missense mutations have been identified; all within the PKD1 gene. Five mutations have been determined as novel and two already published substitutions (p.Gly1999Ser and p.Glu2771Lys) were recurrent (both present in two different individuals). Eight substitutions segregated with the disease in affected families (Table 5) and were confirmed as pathogenic by both predictor programs (PolyPhen2 and SIFT), one substitution occurred de novo.

The probable splice-site mutations c.8792-2A > T (patient 482) and c.12004-15_12015del27 (patient 425) cover the consensus sequence of conserved splice-site and both predictor programs suggested their pathogenity. Although the probable splice-site mutation c.10821 + 4_10821 + 6del AGG (patient 413) is also situated within the conserved splice-site, its pathogenic impact is milder. According to NetGene2 and HSF, this mutation does not abolish the splice-site completely but decreases its efficiency significantly. Therefore this mutation was classified as pathogenic as well.

Within our set of 56 Czech individuals (families), four small in-frame deletions/insertions were detected; three of them were described as novel. The common molecular mechanisms of origin of such genomic events are non-homologous end joining (NHEJ) or microhomology-mediated replication-dependent recombination (MMRDR) [44]. All three novel small in-frame deletions/insertions characterized in the present study show high evolutionary conservation of involved amino acid residues; all three are the only probably pathogenic mutations identified in the patient and segregate with the disease in affected families (Table 5). Their pathogenic potential is therefore highly probable.

By the method of MLPA, one novel large deletion of the PKD1 gene was revealed in patient 105 with proven linkage of ADPKD to the PKD1 gene. The deletion was spread from the exon 3 to the first half of exon 15. The exact beginning of the deletion is arguable, because there is no probe for exon 2 in the MLPA kit used within our study. The deletion was identified in all available affected members of the family but was not characterized at the nucleotide level due to the sequence complexity of the PKD1 locus.

Four sequence changes found within our set of patients were categorized as indeterminate (Table 2). The possible splice-site mutations PKD1/c.11712 + 8C > A and PKD2/c.2019 + 9A > C segregate with the disease in affected families and are the only possible mutations found in patients 409 and 466, respectively. However, their location is out of the conventional range of the consensus splice-site sequence and both predictor programs (NetGene2 and HSF) evaluated them as non-pathogenic. Despite its segregation within affected family, also the substitution PKD1/p.Ala1201Val (patient 308) was determined as benign by both software programs used (PolyPhen2 and SIFT). Although the substitution PKD1/p.Ala3240Thr (patient 412) was predicted as possibly damaging by PolyPhen2, SIFT defined it as benign as well.

In the present study, only 5.3% of mutations were recurrent (2 different already published mutations, both found in two different unrelated families). Both these mutations occur within CpG dinucleotides (G > A transitions), which are mutational hotspots typical for recurrent sequence changes [45]. The substitution p.Glu2771Lys is even published for the sixth time (always from a different location). This fact strongly supports its recurrent origin instead of the existence of a common ancestor.

Such a low percentage of recurrent mutations is in contrast with studies by other groups. Up to 30% of recurrent mutations were mentioned regarding PKD genes in the study by Rossetti et al.[12]. Also, the latest comprehensive mutational analysis of PKD genes in 700 unrelated patients revealed 20.8% of recurrent mutations [36]. The difference can be caused by a smaller number of patients analyzed within our study (56) compared to extensive studies by Rossetti et al. (202) and Audrézet et al. (700).

According to the recent data from literature, ADPKD without family history is present in about 10% of adult patients [46]. The clinical characteristics of patients with and without family history are very similar but only about a half of patients without positive family history show extrarenal abnormalities. In our case, 18% of found likely pathogenic mutations fall into the group of patients with isolated occurrence of ADPKD (the so called de novo mutations).

Large DNA rearrangements account for a maximum of 4% of all PKD1 mutations and even fewer of all PKD2 mutations [18]. Finding of only one large deletion of the PKD1 gene within our set of patients (approximately 2.5%) is thus in accordance with the data from literature.

Sequence changes of all types were spread throughout the PKD1 gene (Table 1). The only evidence of a cluster of mutations (especially truncations) was at the end of exon 15. This region corresponds to the junction of the PKD repeats and the REJ domain of the resulted protein. The reference about this sort of clustering has been already mentioned in literature by Rossetti et al.[12]. No determination of mutational distribution in the PKD2 gene can be done as only one exonic mutation was found. Both PKD genes are characterized by a high level of allelic heterogeneity (a great number of polymorphisms), relatively high GC content and no mutation hotspots. The mutations are usually unique, highly variable and spread throughout the entire gene which fully corresponds with our observations [13]. However, the latest comprehensive study of PKD genes suggests the existence of theoretical hotspots for the three major types of mutations within the PKD1 gene [36]. Our results could not confirm this hypothesis because of the limited extent of the study that did not allow us such a specific computational analysis.

Generally, the ADPKD phenotype is very variable and the correlation between genotype and phenotype is usually not apparent, even within one family. One of the explanations could be a different genetic background and the effect of heritable modifying factors.

The influence of gender, mutation type and mutation location on phenotype of ADPKD is limited and not completely clear. A slight correlation between gender and median age at ESRD onset has been described in literature (56.1 years for men versus 59.5 years for women) [6]. Although the influence of gender has been demonstrated in case of the PKD2 gene too, the most recent report has not proved this dependency [4, 6]. Our data also suggested a slightly higher mean age of ESRD in ADPKD females than in males (53.8 ± 12.3 years versus 50.3 ± 11.4 years). The ages of ESRD in Czech patients compared with those described by Cornec-Le Gall et al. were slightly lower, but in concordance with observations made by Torra et al. (49.6 years for men versus 53.1 years for women) [47].

Another dependency examined within our study was the effect of distinctive type of mutation on the phenotype of ADPKD. We observed that the mean age of ESRD is slightly higher in case of Czech ADPKD patients with missense mutations and in-frame amino acids deletions (54.6 ± 14.4 years) in comparison with patients carrying protein truncating mutations (50.9 ± 11.1 years). This observation corresponds with a strong correlation between the mutation type and median age at ESRD onset that has been recently presented in literature (55.6 years in case of truncating mutations versus 67.9 years in case of mutations with no truncating effect) [6]. For further examination of this dependency a bigger data set would be necessary as only information from 11 patients with missense mutations and in-frame amino acids deletions was available.

Two PKD1 mutations were confirmed in a 24-year-old patient with normal renal function (patient 275 with two mutations within exon 23). The ultrasound of his parents was normal, without cysts. It seems probable that both mutations were de novo however DNA samples from parents were not available. Moreover, a family with large PKD1 deletion was described. The age of ESRD in two males (40 years and 53 years) was known in this family. Intrafamiliar variability has been repeatedly described in PKD families, and was noted in this family, as well. The clinical course in the male with ESRD at the age of 40 years was influenced by arterial hypertension. There were three more affected members in this family (35–50 years) with normal renal function or only mild renal insufficiency. The large PKD1 deletion does not lead to severe clinical course.

The more severe manifestation of the disease in patients with mutations in the 5′ region of the PKD1 gene has been described in comparison with patients carrying mutations in the 3′ region [11]. Nevertheless, a recent study with 741 patients has not shown the significant difference between the 5′ and 3′ end regions [6]. In the case of the PKD2 gene, no clear relationship between mutation location and ADPKD severity has been found, either [48]. Within our study, no effect of location of PKD1 mutations on the mean age of ESRD of Czech ADPKD patients has been observed.

Mutation analysis of highly polymorphic PKD genes is challenging because of the length of genes (especially PKD1 gene), multi-exon structures, high GC content, marked allelic heterogeneity and existence of highly homologous pseudogenes of the PKD1 gene. On average, as much as 10.1 different sequence changes per patient are reported in case of the PKD1 gene [12, 14]. Few complete screens of both genes have been already published with the use of different molecular techniques, but all described methods have still a lot of limitations, including time-demand, high price or insufficient sensitivity. The detection level of the present study is 71% (40 patients/families with likely pathogenic mutations in PKD genes from 56 patients in total) which is in accordance with the 75% efficiency reported by Bataille et al., where the same method of high resolution melting (HRM) was used [23]. The cases with no probable mutation found may correspond to allele dropout, different category of sequence changes that have not been analyzed within this study (such as deep intronic variants or missed mutations in promoter regions) or can be caused by incorrect original diagnosis. Hypomorphic variants or polygenic inheritance may also play an important role [15, 17]. By denaturing high-performance liquid chromatography (DHPLC), approximately 64% efficiency for definitely pathogenic mutations has been achieved [11]. A current publication concerning DHPLC presented the total detection rate of 52.3% [39]. The combination of DHPLC with direct sequencing has increased the level of detection up to 89% [12]. Similarly, 86% efficiency has been reported with Transgenomic’s SURVEYOR Nuclease and WAVE Nucleic Acid High Sensitivity Fragment Analysis System [28]. Using direct sequencing alone, different results have been achieved ranging from 64.5% of mutations detected by Hoefele et al., 78% of probable pathogenic mutations detected by Garcia-Gonzales et al. to the highest efficiency of 93.8% reported by Cornec-Le Gall et al.[6, 36, 49, 50]. In the most recent studies, the new high-throughput next generation sequencing (NGS) techniques have been tested for mutational analysis of PKD genes with the maximal detection rate of 63% and 69.4% [24, 25]. The NGS deep sequencing helps with identification of missed or atypical sequence variants, which are hardly detectable through the conventional Sanger sequencing (especially allele dropout, gene conversion or deep intronic variants). Promising is also the new LR-PCR Sequencing Method developed by Tan et al. with very high sensitivity, improved intronic coverage and marked saving of time and costs [51].

In comparison with other detection techniques, HRM analysis has good and comparable efficacy. Ninety-six or even 394 samples could be analyzed at once through the simple strategy of PCR followed automatically by HRM analysis. The most important benefits of this method are thus the savings of time and costs and general convenience of the working strategy, decreasing the probability of mistakes within the whole process.