|Home | About | Journals | Submit | Contact Us | Français|
Large DNA rearrangements account for about 8% of disease mutations and are more common in duplicated genomic regions, where they are difficult to detect. Autosomal dominant polycystic kidney disease (ADPKD) is caused by mutations in either PKD1 or PKD2. PKD1 is located in an intrachromosomally duplicated region. A tuberous sclerosis gene, TSC2, lies immediately adjacent to PKD1 and large deletions can result in the PKD1/TSC2 contiguous gene deletion syndrome. To rapidly identify large rearrangements, a multiplex ligation-dependent probe amplification assay was developed employing base-pair differences between PKD1 and the six pseudogenes to generate PKD1-specific probes. All changes in a set of 25 previously defined deletions in PKD1, PKD2 and PKD1/TSC2 were detected by this assay and we also found 14 new mutations at these loci. About 4% of the ADPKD patients in the CRISP study were found to have gross rearrangements, and these accounted for about a third of base-pair mutation negative families. Sensitivity of the assay showed that about 40% of PKD1/TSC contiguous gene deletion syndrome families contained mosaic cases. Characterization of a family found to be mosaic for a PKD1 deletion is discussed here to illustrate family risk and donor selection considerations. Our assay improves detection levels and the reliability of molecular testing of patients with ADPKD.
Autosomal dominant polycystic kidney disease (ADPKD) is a common cause of end-stage renal disease (ESRD) due to progressive cyst development and enlargement.1 ADPKD is genetically heterogeneous with PKD1 (46 exons; 16p13.3) accounting for ~85% of cases and PKD2 (15 exons; 4q21) ~15%.2–5 Immediately adjacent to PKD1 lies TSC2, the most common gene for tuberous sclerosis (TSC), a dominantly inherited disorder characterized by the development of benign hamartomas in multiple organs and often resulting in mental retardation/behavioral problems.6 The majority of PKD1 (~40 kb, encoding exons 1–33) lies in a genomic region that is intrachromosomally, segmentally duplicated with six variously rearranged PKD1-like pseudogenes (PKD1P1-P6) located in 16p13.1.2,7 As PKD1P1-P6 have 99% sequence similarity to PKD1, protocols utilizing the rare mismatches with PKD1 have been required to specifically amplify PKD1 for mutation analysis.8
Base-pair screening strategies have identified mutations in ~87% of ADPKD patients, including the well-characterized Consortium for Radiological Imaging Studies of PKD (CRISP) population.5 The majority of mutations are predicted to truncate the protein, but a quarter are missense.
Analysis by field inversion gel electrophoresis revealed that 2–3% of PKD1 mutations are large deletions that remove one to multiple exons,2,5,9 a lower level than the 8% estimated for mutations genome wide.10 One PKD2 deletion and several cases with additional phenotypes removing larger regions have been described.11,12 Large PKD1 deletions that also disrupt the adjacent TSC2 gene result in a contiguous gene syndrome (CGS) with a distinctive phenotype of early onset PKD plus various manifestations of TSC.13–20 Approximately 50 PKD1/TSC2 deletion patients have been described with an average age of 20 years at ESRD.18 However, some PKD1/ TSC2-CGS patients have milder renal disease and one factor associated with a variable phenotype is mosaicism (only some cells harbor the deletion as the mutation occurred early in development), noted in 10 cases.15,18
The difficulty of screening for large rearrangements by gel and fluorescence in situ hybridization methods may underestimate the level of ADPKD deletion mutations. Multiplex ligation-dependent probe amplification (MLPA) is a recently developed, highly quantitative method that allows detection of deletions or duplications of exons and has been used to detect large rearrangements in a range of human diseases, including TSC.15,21,22 MLPA utilizes two tagged primers that lie immediately adjacent in the genome that can only serve as a PCR template once they are ligated. This method has the advantage of speed and high-throughput compared to conventional deletion/duplication detection methods. We describe here an assay utilizing FlexMAP beads and the Luminex instrument to screen the ADPKD genes and PKD1/TSC2-CGS cases and characterize the large rearrangements detected.
MLPA probes were designed to assay for gross deletions/ duplications to PKD1, PKD2, and PKD1/TSC2, including flanking regions. Nine probes were located in the single copy exons of PKD1 (exons 34–46); one in each of the PKD2 exons, plus two flanking probes (total region ~120 kb); six in TSC2; and ten in genes flanking PKD1/TSC2, six up to 150 kb centromeric to PKD1, and four up to 80 kb telomeric to TSC2 (total region 320 kb; see Figure 1 and Supplementary Table S1 for details). In the duplicated region of PKD1 (exons 1–33), 12 exonic probes plus one to IVS2 were designed. To obtain locus specificity, the 3′ end of the short MLPA probe was positioned at a mismatch with all the PKD1P1-P6 pseudogenes (see Supplementary Table S1). Two probes within the single copy area 1–1.5 kb 5′ to PKD1 exon 1 were also developed. In all assays, four probes to autosomal genes elsewhere in the genome and an X-chromosomal marker, DMD, were employed as controls; males showed an ~50% decrease in signal associated with the presence of just one X chromosome (Figure 2).
To test the specificity of the probes, especially those in the duplicated part of PKD1, DNA from a total of 18 PKD1/ TSC2-CGS families18 were analyzed by MLPA (See Table 1 and Figure 1 for details). Cases 8, 11, 12, 14, and 16 are illustrated (Figure 2) and, as expected, all (except 16) showed deletion of both PKD1 and TSC2 with one (12) also deleted for SLC9A3R2, 5′ to TSC2. Several others (8, 11, and 14) removed one or more genes 5′ to PKD1. Previous analysis of case 16 showed that it removed just 84 bp at the 3′ end of the TSC2-coding region, but did not delete the exon 41 MLPA probe.18 All 18 cases showed complete correspondence between the MLPA-defined-deleted region and the previous gel and fluorescence in situ hybridization-based analysis (except in sample 20; see below).
To further test the assay, six ADPKD cases with known intragenic PKD1 deletions were analyzed. All deletions were confirmed with at least one MLPA probe showing the expected reduction in signal in the appropriate area (Figure 3 and Figure 4 and Table 1). For PKD2, DNA from a patient with a large deletion of 4q21.1→21.3, including PKD2,12 showed the expected signal reduction for all probes (Figure 5).
Previous studies showed that a significant proportion of PKD1/TSC2-CGS cases were somatic mosaics;18 a particular challenge to detection by MLPA. Previously defined mosaic cases (19, 20, and 22) showed a reduction of copy number in the deleted area significantly less than the 50% associated with deletion in every cell (see Figure 6, ‘Materials and Methods’, and Table 1 for details). Interestingly, the deletion in case 20 was found to extend further centromeric than previously defined (Figure 1 and Figure 6),18 whereas two other previously described non-mosaic deletions (4 and 6) were found to be mosaics (Table 1). Two previously defined familial cases where a mosaic parent transmitted the deletion to their offspring were also analyzed. In family 7, the mosaic deletion in the mother (M7) was detected, but the low level of deleted cells in F13 was not clearly visible from the MLPA analysis.
Further analysis of the previously characterized 9.2 kb deletion in family 3011575 confirmed the deletion in a sibling (R1502; Figure 3b), but surprisingly, the deletion was not detected in the father (F301157) by MLPA despite him having clear ADPKD with ESRD at 55 years. Amplification across the breakpoint identified a 1.7 kb deletion-specific fragment in the proband and at a much lower level in the father (Table 1). Southern analysis with a 3′ PKD1 probe showed the breakpoint product present at a low level in the father. These results indicate that F301157 is a somatic mosaic with ~15% of cells harboring the deletion in leukocyte DNA. This is consistent with F301157 having a de novo change, with no known PKD in his parents or his seven siblings.
To determine the frequency of large DNA rearrangements in typical ADPKD, the CRISP cohort was analyzed. After sequence analysis, no mutations were defined in 26 families, whereas a further 53 had nondefinite mutations (mainly missense).5 All 79 cases were screened by MPLA and 4 new cases from the mutation negative group revealed large rearrangements (four large deletions had also previously been defined; Table 1 and Figure 3 and Figure 45). A 21 kb deletion detected in 493328 was fully characterized by amplification across the breakpoint and sequencing (Table 1 and Figure 3a and Figure 4). A deletion of 48–87 kb in 407132 removed exons 1–21 of PKD1, plus the flanking gene, RAB26. In family 100009, two probes immediately 5′ to exon 1 were deleted in three affected family members (Figure 7a). To confirm that the mutation disrupted the gene, field inversion gel electrophoresis showed a ~2.5 kb deletion of an exon 1 containing genomic fragment (Figure 7a) that was not detected with an exon 1 probe (SM3). Case 259940 showed increased signal with the exon 18 MLPA probe and long range (LR)-PCR revealed a slightly larger product characterized as a 246 bp duplication of IVS17 and including 119 bp of exon 18 (Table 1 and Figure 7b). The total frequency of large rearrangements in the CRISP population was therefore 8 of 202 (4%).
Screening further base-pair mutation-negative ADPKD samples revealed three additional deletions. In M143, a 1.5–7 kb deletion was found in two affected sisters (Figure 7c). In M29, a deletion removing PKD1 exons 1–15 and extending 5′ of the gene to include at least E4F1 (a region of 120–175kb) was defined. The deletion was confirmed in three affected family members by MLPA (Figure 7d) and other methods (see Table 1 and Figure 1 and Figure 4 for details).
P467 had childhood seizures, mild mental retardation and an atrial rhabdomyoma. Cystic kidneys were not diagnosed until 31 years, but he had ESRD following a left nephrectomy at 33 years. Original gel-based screening did not detect a deletion but MLPA showed a large deletion of >240 kb, removing the entire PKD1 and TSC2 genes (Figure 2). In retrospect, the negative gel results were due to deletion of all tested probes.
P484 had facial angiofibromas, hypopigmented macules, and a hepatic angiomyolipoma with PKD detected by ultrasound at 12 years. At 30 years, she had a left nephrectomy because of serious bleeding, and a serum creatinine = 1.3 mg/100 ml. MLPA revealed a deletion of 100–120 kb (Figure 2).
P455 presented at 2 years with seizures and behavioral problems and had massively cystic kidneys and hypertension. MLPA revealed a deletion removing PKD1 exons 23–46 and 3′ TSC2 to exon 30 (Figure 2).
P973 was diagnosed with PKD and TSC at 5 years, and at 11 years the kidneys were significantly enlarged (13.5 and 14.5 cm), but with normal function. MLPA indicated a deletion removing PKD1 exons 34–46 and at least TSC2 exon 41 (Figure 2).
P974 developed multiple cortical tubers, seizures, and enlarged kidneys at less than 1 year. Her present serum creatinine = 0.76 mg/100ml at 14 years. MLPA showed a mosaic deletion of PKD1 exons 25–46 and all of TSC2. Gel analysis confirmed a mosaic deletion of ~60 kb (Figure 6a and b).
M214 was diagnosed at 6 months with enlarged kidneys, which were decompressed. TSC was diagnosed at 13 years with facial angiofibromas and by 23 years renal angiomyolipoma developed that required embolization. Her present serum creatinine = 1.5 mg/100 ml at 43 years. MLPA detected a mosaic deletion of 40–55 kb (Figure 6a).
P444 had hyperechogenic, enlarged kidneys detected in utero and TSC diagnosed by the detection of cerebral hamartomas. At 12 years, she is hypertensive with enlarged kidneys (17 and 15.5 cm) but normal function (serum creatinine = 0.68 mg/100 ml). MLPA detected an 8–14 kb mosaic deletion (Figure 6a).
We have developed an MLPA assay to rapidly screen for large rearrangements causing ADPKD and the PKD1/TSC2-CGS, improving diagnostics for these disorders. Samples can readily be assayed in a 96-well plate format with results available within 2 days, in contrast to labor intensive gel or fluorescence in situ hybridization analysis. The screen covers genomic regions around PKD2 (120 kb) and PKD1/TSC2 (320 kb), with a probe density of one per 1.4 kb in the compact PKD1 gene (except IVS1), and one per exon in PKD2. This probe density is likely to detect the vast majority of large deletion mutations; smaller changes to PKD1 are generally evident as altered LR-PCR products.5 The utility of the assay was shown by detection and defining 25 previously detected deletions, plus characterizing six new PKD1, one PKD2, and seven PKD1/TSC2 mutations. During the preparation of this paper a similar MLPA scheme for PKD1 was described, but no intragenic deletions were detected.23
A single-mismatch between PKD1 and the six PKD1-pseudogenes was sufficient for locus specificity when the mismatch was placed adjacent to the ligation site at the 3′ end of the short probe. This mismatch presumably destabilized annealing of the short probe so that ligation could not occur. The MLPA system employed here has distinct advantages over other described methods. In particular, the Luminex Flex-Map bead system allows a higher level of multiplexing than capillary electrophoresis systems; 88 different probes can be utilized in one assay.
In all tested cases where more than one adjacent MLPA probe had reduced signal, LR-PCR or gel methods identified a deletion. However, in three cases signal reduction for a single probe was associated with a rare base-pair change: PKD1:8681_8688del9 (2894delANS), exon 23; PKD1:11186A → C (H3729P), exon 39 and TSC2:4767G → A (P1589P), exon 36. The 8681del9 deletion was previously described as a rare polymorphic deletion removing a duplication that has occurred since PKD1 and the PKD1P1-P6 diverged.5 This difference was used to generate the exon 23-specific probe (Supplementary Table S1). The substitutions 11186A → C and the 4767G → A lay at the −2 position of the PKD1:exon 39 short probe and the −1 position of the TSC2:exon 36 short probe, respectively. These cases illustrate that rare single nucleotide polymorphisms (SNPs) can disrupt ligation and amplification and, therefore, signal difference associated with just one MLPA probe needs to be tested by LR-PCR and sequencing.
In common with other CGS cases, all the newly described cases had TSC with significant renal cystic disease.15,18 Although in our original report we stressed the infantile presentation in non-mosaic cases,18 others have highlighted cases where PKD was only evident much later.19 The four non-mosaic cases had moderately severe cystic disease (resulting in ESRD at 33 years in P467), but the diagnosis was not always made during infancy; at 2, 5, 12 or 31 years. Interestingly, in the three mosaic cases, PKD was always diagnosed before 1 year, showing the severity of cystic disease and penetrance of the mutation (at least in leukocyte DNA) does not always strongly correlate.
The MLPA assay will be useful in all TSC patients with significant renal cystic disease to differentiate CGS cases from those with TSC and PKD inherited as separate gene mutations.24,25 Approximately 60% of TSC cases are de novo; 80% due to TSC2 mutation,26 and practically all described CGS cases are de novo or inherited from a mosaic case.18 Previously, mosaicism has been found to be commonly associated with large TSC deletions,15,18 a finding echoed here in the CGS. Mosaic cases were readily detected by the MLPA assay (except where the deleted allele was present in o20% cells) and represented three of seven new cases; 40% of all PKD1/TSC2 deletion families had a mosaic case (Table 1 and Figure 1). Identifying mosaicism is important for determining risk in siblings of the patient.
This study determined that the level of large rearrangements causing typical ADPKD is ~4%, a little higher than estimates from gel-based studies,9 and included the first duplication. These numbers somewhat underestimate total PKD1 deletions as those that also disrupt TSC2 have the CGS phenotype. One PKD2 single exon deletion was found, indicating that these mutations are also relatively rare. Although the level of large PKD1 or PKD2 rearrangements is relatively low, 31% of sequence determined mutation negative cases had large deletions, indicating that MLPA would be a useful addition to any clinical, mutation-based test. It is worth noting that the four new large rearrangements found in CRISP were from the mutation negative group, not those with likely missense changes,5 increasing the probability that the missense changes are truly pathogenic. This assay will also prove useful for screening for somatic loss of heterozygosity in cyst lining cells.27
All previously described large deletions in typical ADPKD were intragenic;2,5,9 however, in this study two deletions extended centromeric to PKD1. One (407132) removed at least the closest flanking gene, RAB26. In the second (M29), the deletion extended 100–150 kb centromeric from PKD1 and disrupted up to 10 genes (Figure 1). This is a family with a sibship of 15, 7 of whom are known to have ADPKD (Figure 7c); however, no unusual clinical features were noted. This is significant because the M29 deletion extends further centromeric than any in the PKD1/TSC2-CGS cases. Previously, a CGS case with a ~280 kb deletion, probably disrupting up to ABCA3, had craniofacial and limb abnormalities (acrofacial dysostosis), which is not characteristic of TSC.28 This phenotype may be due to disruption of ABCA3 or other genes (RNPS1, DCI, or DNASE1L2) deleted in the published case but not disrupted in M29. Recessive mutations to ABCA3 are associated with severe neonatal surfactant deficiency,29 but there is no reported heterozygous phenotype. None of the other implicated genes have been associated with a disease phenotype. Fine mapping of both mutations by modifying the MLPA assay may help to understand acrofacial dysostosis possibly associated with this region.
The renal disease severity of large PKD1 deletion cases as a group (or families individually) was unremarkable. For instance, the average age at ESRD was 55.7 years, similar to the PKD1 average. Family 100009, with exon 1 deleted, had three cases with large renal volumes (>2,800 ml), in the top 5% of the age-corrected CRISP volumes.30 However, two other family members had average (or even below average) volumes, probably indicating that genetic modifiers rather than allelic effects have the greatest influence on the severity of renal disease.31
Three of the twelve PKD1 large rearrangement families were molecularly proven to have de novo mutations. One consequence of a significant level of de novo mutations, by analogy with TSC and the PKD1/TSC2-CGS, is that a significant level of mosaic cases may be expected. One such case was detected in F301157. Recently, the first case of mosaicism in ADPKD was described.32 Diagnosing mosaicism is important as it may have prognostic implications; although F301157 developed ESRD at 55 years even though the deleted allele was rare in leukocyte DNA (but presumably more common in the kidney). Furthermore, mosaicism is important to consider when determining the risk of transmission to children and the probability that siblings will be affected. In particular, it can be a confounding factor when screening for a living renal donor in offspring from the mosaic case.32
The study was approved by relevant Institutional Review Boards and Ethics Committees, and participants gave informed consent. Clinical and family records were reviewed to collect a full clinical history.
DNA was isolated from blood samples with the Puregene DNA Purification System (Gentra Systems; Minneapolis, MN, USA). Details of new and previously published cases are summarized in Table 1. Field inversion gel electrophoresis and pulsed field gel electrophoresis employed liquid and agarose block DNA, respectively, as previously described.2,6,18 Details of probes and microsatellites employed, as illustrated in Figure 1, and a comprehensive restriction map of the area, have previously been described.2,6,18,33 Methods for LR-PCR have been described8 and primer sequences are available on request.
Two adjacent gene-specific oligo probes were designed within regions of target genes without known SNPs (Supplementary Table S1).34,35 The short probe had 24–30bp of complementary sequence 5′ appended with a universal forward primer (5′-acattttgct gccggtca-3′); and the long probe had 40–50 bp of complementary sequence 5′-phosphorylated and 3′ appended with complementary Luminex (Luminex Corporation; Austin, TX, USA) FlexMAP bead sequence and a universal reverse primer (5′-gtcctttgtcgatactgg-3′). All probes were synthesized by Integrated DNA Technologies (Coralville, IA, USA); short and long probe scale/purification was 100 nmol/HPLC and 250 nmol/PAGE, respectively.
Due to the presence of six PKD1-pseudogenes (PKD1P1-P6), PKD1 versus PKD1P1-P6 mismatches were identified to design PKD1 exons 1–33-specific probes. MacVector (MacVector Inc.; Cary, NC, USA) was used to establish ClustalW sequence alignments. Localization of a single base-pair mismatch at the ligation site (the extreme 3′ end) of the short probe, and clustering of any other available mismatches to the remainder of the short probe, resulted in PKD1-specific MLPA probes (Supplementary Table S1 for probe details and positions of mismatches).
For hybridization, a mix containing 2.5 fmol/µl of each probe was prepared in 1 × Tris-EDTA (TE; Fisher Scientific; Fair Lawn, NJ, USA). Genomic DNA (400 ng) was denatured at 95 °C for 5min, cooled to 25 °C and a solution of 1.5 µl MLPA probe mix and 1.5 µl of Ligase-65 MLPA Buffer (MRC Holland; Amsterdam, the Netherlands) added and hybridized at 60 °C for 16–24 h.
The ligation reaction mix contained: 25 µl of water; 3 µl of Ligase-65 Buffer A; 3 µl of Ligase-65 Buffer B; and 1 µl of Ligase-65 (MRC Holland). The ligation solution (32 µl) was added to the DNA reaction tube, incubated at 54 °C for 15min., followed by a 95 °C enzyme denaturation, 5 min, and 4 °C cooling, 5 min.
PCR amplification of the ligated probes employed the mix: 27.5 µl water; 1.5 µl of 50 mM MgCl2; 4 µl dNTPs (2.5 mM each; Invitrogen Corporation, Carlsbad, CA, USA); 0.25 µl each F-/R-universal primer (20 µm; Integrated DNA Technologies); and 1 µl Platinum Taq Polymerase (Invitrogen). The MLPA ligation product (10 µl) was added to 40 µl of the PCR mix and amplification employed the following program: 95 °C, 5 min: 23 cycles of 95 °C, 30 s; 60 °C, 30 s; 72 °C, 30 s; 72 °C, 20 min; and 4 °C, 5 min.
The Luminex FlexMAP technology employs flow cytometry and color-coded microspheres to rapidly identify up to 88 different samples (in this case MLPA probes) in one assay. A Luminex FlexMAP bead mix corresponding to the MLPA probes to be analyzed was prepared by mixing the FlexMAP beads (Luminex; part no. L100-Uxxx-01, xxx = bead no.: 001–099; 2000 beads per DNA sample) with 40 µl per DNA sample of 1 × tetramethylammonium chloride (Sigma-Aldrich; St Louis, MO, USA).
The PCR reaction mixture (10 µl) was added to 40 µl of the FlexMAP bead mix and hybridized at 37 °C for 1 h. A 250 µl tetramethylammonium chloride solution containing 2 µl of Streptavidin, R-Phycoerythrin (Invitrogen) was prepared; 25 µl of this solution was aliquoted, along with 50 µl of the FlexMAP bead hybridization mixture, to a flow cytometer 96-well analysis plate (Luminex). The samples were hybridized at 37 °C for 1 h in the dark and analyzed using a Luminex-100 flow cytometer.
The resulting data were deconvoluted according to relative copy number versus both the external control genes as well as a wild-type normal control female sample using the GeneMarker software (Softgenetics, LLC; State College, PA, USA). The ‘normal’ signal copy number threshold was set at 2.0 ± 0.4. An X-linked control gene (versus wild-type normal control male sample), as well as multiple gene-specific (PKD1, PKD2, and PKD1/TSC2) characterized deletion samples, was utilized as internal controls to confirm the specificity of the MLPA probes. MLPA probe gene copy numbers were exported to Excel (Microsoft Corporation; Redmond, WA, USA) and reported as copy number vs gene probe histograms.
The means and standard deviations (s.d.) of probe variability were determined for the normal and deleted regions and limits set at 3 × s.d.. of the moving average using the method of quality control charts.36 Deletions with a mean value for the deleted region falling outside of these 3 × s.d. limits (P = 0.01) were considered to be mosaics and the percentage of cells with the deleted allele is shown in Table 1.
The patients, their families, and clinicians, including Dr A Britland, Keighley, UK, are thanked for taking part in the study. Members of CRISP who collected radiological and clinical information: WM Bennett, CM Meyers, LK Moen, JP Miller, JE Bost, BF King, LH Wetzel, DA Baumgarten, PJ Kenney, J Mahan, B Stafford, L Stevens, K Cornwell, D Watkins, S Langley and P Trull are also thanked. The study was supported by NIDDK grant DK58816 and CRISP support from NIDDK Cooperative Agreements: DK56943, DK56956, DK56957, and DK56961. WC Wong was supported as a Mayo Clinic Summer Undergraduate Research Fellow (SURF).
All the authors declared no competing financial interests.