A complete mutation screen of PKHD1 in autosomal-recessive polycystic kidney disease (ARPKD) pedigrees

Kidney International, Vol. 64 (2003), pp. 391–403

GENETIC DISORDERS–DEVELOPMENT

A complete mutation screen of PKHD1 in autosomal-recessive polycystic kidney disease (ARPKD) pedigrees SANDRO ROSSETTI, ROSER TORRA, ELIECER COTO, MARK CONSUGAR, VICKIE KUBLY, SERAFIN MA´LAGA, MERCEDES NAVARRO, MOUNIF EL-YOUSSEF, VICENTE E. TORRES, and PETER C. HARRIS Division of Nephrology and Division of Gastroenterology, Mayo Clinic, Rochester, Minnesota; Department of Nephrology, Fundacio´ Puigvert, Barcelona, Spain; Laboratorio de Gene´tica Molecular-Instituto de Investigatio´n Nefrolo´gica (IRSIN-FRIAT and Servico de Nefrologı´a Pedia´trica, Hospital Central de Asturias, Oviedo, Spain; and Hospital La Paz, Madrid, Spain

A complete mutation screen of PKHD1 in autosomal-recessive polycystic kidney disease (ARPKD) pedigrees. Background. Autosomal-recessive polycystic kidney disease (ARPKD) is an important neonatal nephropathy characterized by fusiform dilation of collecting ducts, congenital hepatic fibrosis, and in some cases Caroli’s disease. The ARPKD gene, PKHD1, has recently been identified. Herein we describe an effective method for PKHD1 mutation screening and the results from analysis of a novel ARPKD cohort. Methods. The coding region of PKHD1 was amplified as 79 fragments and analyzed for base pair changes by denaturing highperformance liquid chromatography (DHPLC). Forty-seven ARPKD and 14 pedigrees with congenital hepatic fibrosis and/or Caroli’s disease, were screened for PKHD1 mutations. Results. Thirty-three different mutations were detected on 57 alleles (51.1% ARPKD, 32.1% congenital hepatic fibrosis/ Caroli’s disease). In the 22 pedigrees where both mutations were identified, two were homozygous for 9689delA and the remainder were compound heterozygotes; a combination of truncating, missense and splicing changes. Patients with two truncating mutations all died in the perinatal period. Two frequent truncating mutations were identified: 9689delA (9 alleles) and 5896insA (8 alleles) plus some more common missense changes; haplotype analysis indicated most were ancestral mutations. Conclusion. DHPLC has been established as a rapid mutation screening method for ARPKD. The mutation detection rate was high in severely affected patients (85%), lower in those with moderate ARPKD (41.9%), and low, but significant, in adults with congenital hepatic fibrosis/Caroli’s disease (32.1%). The prospects for gene-based diagnostics are complicated by the large gene size, marked allelic heterogeneity, and clinical diversity of the ARPKD phenotype. Identification of some common mutations, especially in specific populations, will aid mutation screening.

Autosomal-recessive polycystic kidney disease (ARPKD, OMIM 263200) is a severe, inherited disorder of the kidney that typically presents in the neonatal period and has an incidence of ⬃1/20,000 [1]. The neonatal disease is characterized by bilateral renal involvement consisting of fusiform dilation of the collecting ducts, often leading to massive kidney enlargement [1, 2]. Approximately 30% of patients die in the perinatal period with pulmonary hypoplasia due to oligohydramnios [3, 4]. A less precisely defined number of patients present in later childhood or as adults and in these cases the major clinical complication is often liver disease [5]. The liver involvement consists of proliferation and dilatation of the interlobular ducts and portal fibrosis (congenital hepatic fibrosis) and, in some cases, focal dilatation of the larger intrahepatic bile ducts (Caroli’s disease) [6, 7]. Linkage analysis in ARPKD families mapped the gene, PKHD1, to chromosome region 6p21-cen [8], and subsequently by genetic and physical mapping to a 1cM interval [9–11]. These studies indicated that ARPKD is genetically homogeneous [12] and have facilitated prenatal diagnosis with closely linked flanking markers [13]. Recently, through the analysis of a newly described animal model of ARPKD, the PCK rat, and a positional cloning approach, the PKHD1 gene has been identified and characterized [14, 15]. PKHD1 is a large gene, extending over a genomic region of ⬃472 kb, with 66 coding exons and a transcript of 16,235 bp. The disease protein, fibrocystin, is predicted to contain 4074 amino acids with a molecular weight of ⬃447 kD. The conserved domains and predicted structure of fibrocystin suggest that it may have a role as a receptor in collecting duct and biliary differentiation [14]. In the two papers describing PKHD1, 29 different mutations and a total of 40 mutant alleles were identified [14, 15]. The majority of changes were missense (N ⫽ 17), but frameshifting insertions or deletions (N ⫽ 9)

Key words: ARPKD, PKHD1, mutations, congenital hepatic fibrosis. Received for publication December 24, 2002 and in revised form February 11, 2003, and March 13, 2003 Accepted for publication March 24, 2003

 2003 by the International Society of Nephrology

391

392

Rossetti et al: Mutation analysis in ARPKD

and nonsense mutations (N ⫽ 3) were also identified. Most mutations were unique but five were found on more than one allele, with two changes described four times. All cases in which both mutations were described (apart from one consanguineous family) were compound heterozygotes and of these only one case with two truncating mutations was described [15]. Typical patterns were a truncating and missense (N ⫽ 11) or two missense mutations (N ⫽ 5). In a significant proportion of cases no mutation, or just one mutant allele, was identified. Rapid and efficient mutation analysis of PKDH1 would allow gene-based diagnostics to be established and serve the significant demand for prenatal diagnosis in this neonatal disorder. At present, molecular diagnostics is possible by linkage analysis only in families in which DNA is available from an affected sib with a firm ARPKD diagnosis, and the family structure allows the disease haplotype to be defined [13]. Gene-based diagnostics could extend the number of informative families and also allow a definitive diagnosis of ARPKD in cases where the clinical data are equivocal. At present, confusion can occur between ARPKD and early-onset cases of ADPKD and other childhood causes of renal cystogenesis [13]. The significance of congenital hepatic fibrosis and/or Caroli’s disease, with minimal kidney disease, is also not always clear in an older child or adult. Additionally, mutation studies will help to determine the phenotypic range of ARPKD and show the extent to which the clinical variability is associated with specific genotypes. To allow efficient gene-based diagnostics, accurate, rapid and cost-effective methods of mutation screening are required. We describe here a mutation screen of the entire coding region of PKHD1 using denaturing highperformance liquid chromatography (DHPLC). This method has proved to be an efficient and effective method for detecting base pair mutations in large, multiexon genes (like PKHD1), with typical detection levels ⬎70% [16, 17]. This study shows the feasibility of mutation screening in ARPKD and the challenges to establishing routine gene-based diagnostics. During review of this manuscript, another paper describing mutation screening of PKHD1 was published [18], characterizing 34 novel mutations. Comparison with this paper will be made in the Results and Discussion sections. METHODS Details of the study cohort The study was approved by the appropriate Institutional Review Board or Ethics Committee and all participants gave informed consent. A family history and clinical information was collected from all pedigrees. Patients were diagnosed as ARPKD using established diagnostic criteria [13]. Imaging showed enlarged echogenic kid-

neys with poor corticomedullary differentiation in neonates and moderate renal enlargement with macroscopic cysts and/or medullary sponge kidney in older patients. Liver analysis in most cases demonstrated at least one of the following: radiologically, intrahepatic bile duct dilation or echogenic liver parenchyma; histologically, congenital hepatic fibrosis or ductal plate malformation; and clinically, portal hypertension. Parents were clinically unaffected with negative imaging studies. For analysis purposes, ARPKD was considered severe if the disease resulted in perinatal death, or moderate if the patient survived the perinatal period or was diagnosed later [18]. Patients with predominant liver disease were diagnosed as adults with congenital hepatic fibrosis and/or Caroli’s disease and minimal or no evidence of renal disease. The families come from Spain (designated OV, PRR or HEP), the United States (M), or United Kingdom (P). The proband and all family members wishing to participate gave a blood sample for DNA isolation. Amplification of the PKHD1 gene by polymerase chain reaction (PCR) Genomic DNA was isolated from a peripheral blood sample by standard methods. In a few perinatal cases, the QIAmp DNA Mini Kit (QIAGEN, Inc., Valencia, CA, USA) was used to extract genomic DNA from formalin-fixed paraffin-embedded kidney tissue blocks. Briefly, three 20 ␮m slices, obtained using a Biocut 2030 microtome (Leica Instruments, Bannockburn, IL, USA), were dissolved in xylene to release the tissue, followed by a 100% ethanol wash. The samples were digested with proteinase K at 56⬚C overnight and the DNA isolated using the standard QIAmp DNA Kit extraction protocol. All coding exons of the PKHD1 gene were amplified from genomic DNA as fragments of 150 to 370 bp. Primers were generally positioned in introns 25 to 30 bp from the exon boundary, in order to detect mutations of the canonic splice sites, but to minimize detection of nonpathogenic intronic changes. Polymerase chain reaction (PCR) was performed as previously reported [14, 17]. Briefly, 60 ng of genomic DNA was amplified in a mix containing 6 pmol of each primer, 200 ␮mol of each deoxynucleoside triphosphate (dNTP), 2.5 mmol/L MgCl2, 1 U of AmplitaqGold (Applied Biosystems, Foster City, CA, USA) in the supplied buffer, in a total volume of 25 ␮L. The amplification program consisted typically of an initial denaturation at 94⬚C for 2 minutes, followed by 35 cycles at 94⬚C for 30 seconds, 44 to 65⬚C for 30 seconds, and 72⬚C for 30 seconds, with a final extension at 72⬚C for 10 minutes. This protocol was used for all exons except 2 and 3, where a dimethyl sulfoxide (DMSO)based PCR buffer [19] (exon 2) and a hot start protocol with a high annealing temperature (exon 3) was employed due to persistent nonspecific amplification and primer

Rossetti et al: Mutation analysis in ARPKD

dimerization. PCR primers and conditions are summarized in Table 1. Heteroduplexes were generated by incubating the PCR product for 5 minutes at 95⬚C; cooling to 65⬚C, 0.1⬚C per second; 30 minutes at 65⬚C; and cooling to 37⬚C 0.1⬚C per second and 10 minutes at 37⬚C. This analysis assumed that patients were compound heterozygotes for mutations, as had previously largely been found [14, 15]. To screen for homozygous changes, an equal quantity of normal amplicon was added to the patient product before heteroduplex formation. Mutation analysis of the PKHD1 gene by DHPLC DHPLC was performed using the Wave system (Transgenomic, Inc., Omaha, NE, USA) following the protocol we described previously [14, 17]. Briefly, 300 to 600 ng of crude PCR product was injected into the chromatographic column (DNASep cartridge) (Transgenomic Inc.) and eluted through a 8.6-minute linear gradient of buffer A [5% triethylammonium acetate (TEAA)] and buffer B (5% TEAA and 25% acetonitrile). A 2% buffer B slope per minute was used. Melting profiles were analyzed using Wavemaker 4.0.32 software and each amplicon was run at the predicted melting temperature ⫾1 and 2⬚C (and additional temperatures when needed) to optimize conditions. Due to the initial lack of positive controls, a set of eight samples was used to test the analysis conditions. The optimal temperature was considered to be that immediately before a significant decrease in the retention time (before 3 minutes) and/or excessive broadening of the peak occurred, indicating excess denaturation. This was typically located in the range of 50% to 75% helical fraction. When a sequence change was found, that sample was used to refine the optimal analysis temperature; when the best resolution of the mutant amplicon occurred. Where more than one positive control was available, the most subtle change was chosen as an internal control. Samples showing an aberrant elution profile were typically reamplified and subjected to direct sequencing as previously described [17]. The DHPLC conditions and positive controls available for each amplicon are summarized in Table 1. Validation of mutations Missense mutations and subtle splicing mutations were validated through the analysis of 50 normal controls (100 normal chromosomes). This was performed using DHPLC and including the candidate mutation for comparison with the normal samples. Whenever DNA from other members of the pedigree was available, validation was also confirmed by family segregation analysis. This was performed by DHPLC, or alternatively by direct sequencing when more than one sequence change was present in the fragment being analyzed. Restriction assays were developed to facilitate the detection of the four most common mutations: 5895insA,

393

9689delA, T36M, and I222V. For the mutations T36M and I222V, the same restriction enzyme was used, HpyCH4 IV (New England Biolabs, Beverly, MA, USA). For T36M, the mutation abolishes a restriction site, so that the normal restriction pattern of exonic fragment 3 (54 ⫹ 52 ⫹ 52 bp) is changed to 106 ⫹ 52 bp; for I222V, the mutation creates a new restriction site and the exon 9 amplicon (which is not cut by the enzyme) generates fragments of 102 ⫹ 55 bp. A restriction-generating PCR (RG-PCR) approach was designed to detect 5896insA with a modified reverse primer (5⬘-ACTTCACACACCTTTAATG TGCACT-3⬘; bold indicates base modified) and the forward exon 36 oligonucleotide. The normal 206 bp fragment was resolved as 179 ⫹ 28 bp in the mutant when digested with Afl II (New England Biolabs). The restriction enzyme HpyCH4 III (New England Biolabs) was predicted to digest the mutant exon 58d amplicon as fragments of 119 ⫹ 188 bp. However, we found this digestion inconsistent and analyzed for homozygotes by DHPLC with normal fragment added. Restriction digestions were typically performed in a total volume of 20 ␮L, using 10 ␮L (⬃1 ␮g) PCR product and 10 to 20 units of restriction enzyme in the supplied buffer and incubated at 37⬚C for 2 hours. Bovine serum albumin (BSA) (1%) was added to Afl II. Restriction bands were visualized on 3% agarose gels after ethidium bromide staining. Mutation positions are described using the PKHD1 cDNA sequence, AY074797, with the A of the start codon designated as first nucleotide. The program SignalP 2.0 (http//www.cbs.dtn.dk/services/SignalP/) was used to analyze the consequence of the A17V mutation on cleavage of the protein. Haplotype analysis Families with recurrent/ancestral mutations were analyzed with the intragenic microsatellite markers D6S1344 (IVS15) and D6S1714 (IVS60) using primers described in Genbank and previously defined methods [20]. Briefly, the markers were amplified by PCR, the products resolved on a 10% polyacylamide gel, stained with ethidium bromide and visualized by ultraviolet transillumination. Marker segregation was analyzed to determine the haplotype associated with the specific mutations. The normal frequency of haplotypes was calculated from the normal chromosomes in the same families. RESULTS Development of a mutation screening strategy for PKHD1 The large size of the PKHD1 open reading frame (12,222 bp) and multiple exon structure indicated that a rapid semiautomated method for mutation screening was required. Previously, we had successfully developed a DHPLC method to screen the large, multiexon, ADPKD gene PKD1 [17] and therefore adapted these methods to

Size bp

269 158 260 189 178 180 186 157 152 157 233 170 247 193 347 188 185 228 204 289 245 250 269 208 229 363 207 253 348 164 312 365 345 350 368 349 171 239 288 231 222 293 278 232 288 250 152 195

Exonic fragment

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32a 32b 32c 32d 32e 32f 32g 33 34 35 36 37 38 39 40 41 42 43

50 54 65 54 52 47 47 47 44 49 49 49 52 50 51 52 49 53 49 57 57 56 48 55 54 48 49 46 54 50 50 56 52 50 50 50 50 52 51 49 49 51 50 49 51 50 43 50

Annealing temp ⬚C

PCR

AGGTTTCAGAACAGCAAAATAATCG TGGTTTGAATCTGACCTTCAAAACC TTTCACACTGTCCTGTGTCAATGAC TTGGGAATTCATGGTTTTTGATTC GAAAGGCTTGTGCCTCCTGTG CATTGAGTTTGAGCTAAGTCC GTTTATTGGGAGTATTGC TGAGTTGTCTGGTCATTC ACTCGTGCAGATTCCTGAG CAATCCCAGTTGATATTTTC TCCTGGTCTATATTTGGAAGC CCTACACACACACACATAC TTCCCCAATTTGGGAAGG TTGGTTACTCTTGCTTGACTC TGCATAGTATTGATCATG TTAGCACCATCATTTAGTCTTG AATTCCTGGCATTTTTTTC TATCTATGCCTGCCTTTC CACTAATAGAACTGAAGGAC TAACCGGAGAGGACTGCAAGTG TTTTCCACACAGCAAGTCTACCATC CACCCCAACCCAGACGTTAATAC TAGTGTCTGTGTTTTCTG TTCGGTTCCATGACAGAATTTACC CAGCTTGGGAGCACTTCACATATAC TGAAGTAATATCACTGAGAG CCTGTATGGTTGGTGATC CCCTTAAGTCAGTCCTAC GGGGTGACTGTGAATTTAATC ATCTCTCTCTGTCAGTTATTTCC TCTTAGTTCAGAATATCAG TGGGCTGGCAACAGGTTC ATGGGATTTGCTAATATG GACCACACCATTCTCTGC GCAATGTAACTTTTTTTAATGC AGCTCATCCGGTGCATTG GGGAGTACCACGTCAGAG GATCAAGAACTTGTACCTTTGTC TTCTTTCCTAATGGTGAC TAAGATTGATGACACCCC AACCAACCAACCCACCAAC TAAGCCTTATCCTCCCAG TCTGGACAACTTTTCCTC TGATGTCCTCAGTTCTATC ATGCTTTAGGTTCTCTGG AACAGAATCTCAGGAGCC AAGTGACATAAAATATACTC GATCCCCTGGATTTGTTG

Forward (5⬘-3⬘)

Reverse (5⬘-3⬘) TTCTCAAGGTAACCTATTGTGTTCTTA AAATGTGCACTTGGTAAAACCCC AAAATCCCTCATCCTGTCTGGTC ACATACCTTCCTCCAGCCTTAGAAC TGGCAAACAGATTCACAATTATTCC CATGCAGCATGTATGTAACTAG GTGGACGAACTTACAAGC GAGAAAGAAATGGATAAGAC CAAGATGAGAGAGATAGG ACAAGGGAAGGGGTACTTG CATCCCTCATGCCATACAGAC GTTTATTGAACAGCCCTG TTAGCAAAGGTGCTTTTG CTGGCAACAGAGAAAAGG AGCTCCATGGGACTGGAAAG AAAGACCACCCCCAGTTC CATTTTATAGAAAGAAAGAAGACC AATACCTACCCACCTGACCC TGACTGAATTCCCACCACGC TTTGAGGTAGGCATGTGACCGG CATTCTTAGGAGAAGGGACAGGTG TCCCAGGATGTTGTTCCCTTGG TCCAGGGCAGCAAATCCATG TGAAACTGGAGCTTGCACTTAGG TTAAGCCCATCTCAGAGCCAAG ACATACTGTGAGACCCTCC GAGAAAGAGATATGAAAGG TTTATAGGACCAATGCTC TGCTAGACCATCAAACAAATC AAATAGAATTGCTGGATAATTG TCATACATGAAGGTGAAG GCCATTATCCGAGGCATC GACAGGACTTGCCTCTTC TAAAAAACTGACAGGTAG TCCTATGTGATACCAAAG ATAACTCTTGAGGTGAAC TCCAGAAGTGAAAGGAGC TTAACCAAAGAATATCATTTCC TTTGTGGGGAAGTTCAGGG GCTGTTTGAATCAGTCTG TATTACCAACCTACAAAC ACTTCATTTCCTCTGATC TCTTCCATGTCAACTTAG TTGCTCATTAGACTTTCC TAGTGCCTTAAACATGGG TGGGGAGAATTCATTGTG GACAATTTTAAATACACTG TCAGTTCTGGTCTTCCTG

Primer sequences

53,54 57 60 60 58 55 56 55,57 53 54 51 56,58 59 58 61 58 58 61 60 61 59 59 58 60 56,58 52,57 59 53,56 58 56,58 54,58 61 60 60 58,59 59 58 60 60 58 61 61 59 58 59 56,58 53 57,58

Temp ⬚C 53 48 53 50 49 49 49 47 49 49 52 48 52 50 55 50 49 52 50 54 52 53 53 51 52 56 51 53 55 48 55 56 55 56 56 55 48 52 54 52 51 54 54 52 54 53 47 50

Initial % buffer B

DHPLC conditions

Continued

IVS1-47C/T T36M V/M65 383delC None IVS7⫹19T/C None I222V None None None None IVS14⫹23G/T 1185T/C None 1587T/C 1624del4 None None 2046A/C R/C760 IVS23⫹50C/T W/R852 None R/Q909 2853C/T None IVS28-2A→C 3537T/C None A/V1262 None L1407R None None S1664F IVS32⫹42del4 None S1833L Q1917R D1942G E1995G None 6383delT None None None I2331K

Positive control

Table 1. Details of polymerase chain reaction (PCR) amplicons and denaturing high-performance liquid chromatography (DHPLC) conditions used to analyze the PKHD1 gene

394 Rossetti et al: Mutation analysis in ARPKD

Size bp

198 190 199 213 324 257 277 147 207 215 224 153 231 228 283 292 272 307 265 227 350 350 350 218 270 202 213 263 230 283 292

Exonic fragment

44 45 46 47 48 49 50 51 52 53 54 55 56 57 58a 58b 58c 58d 59 60 61a 61b 61c 61d 62 63 64 65 66 67a 67b

47 45 47 46 51 48 50 47 48 48 44 46 48 52 48 50 48 50 45 50 48 48 48 48 53 53 49 47 54 57 57

Annealing temp ⬚C

PCR

TATCATACATGGGGTAAC GTTAGAAACATAAAAATTGG TCAGACCTTTGCTGTAAC GTCCAGTTTTCTTATTTTGC GTGCCATTGTGTAATAATCTTTCTG CAAATAATCTCTCAACCC GGGGTTCCTTACTAAATG CTTTGTATCACATGCAAG GGAAGTTATCACAATGGATTAG TTGTTTTTGTGACATATC CTCTCTTTCTTTTAATTTC CTATCCAACTGTTACTCC CACTGTTAGTATATCCAATG GTTTTTTTTTCCCACAACTC AGGAAAGTACCTGATGAC ATATTGTGTTTGGCACAG GAACTGCTTTGGTCTGAC AAAATTCCGTCAAAAAAG TGGCTGGTGGTTTATATG CATGAAATGAAAGAGTTGC TATCACTTGTTTTGCTTC AAGTCTGCTTCATGGATC AGTCTTAGAAAAAGGCTG CAACAGTAAGGAGCACTG GGATTGTGGAAAATTGCTACCATAG TCTGAATCCAACTTTTTCTTCCTCC TTCGCAGAAGACATGAAGACATTG TTATATTAGCATCTTATTAA GCTGATGGTCCCACTTACAACTG TGAAAACTAAATCCATTTCTTCCCC CCTGCAAGAGACTGGGAACTGG

Forward (5⬘-3⬘)

Reverse (5⬘-3⬘) AAGACAGCCAAAACATAG AACAACAACAATAACAAC AGCCTAAAACAACCACAC TCATCTGTTCTGTCTATTC CATCGGCAAGCTAAAAAG GCAGCATACCAACTAATG GCTCTCAAAACATTCATC TTCTGCCATACTAGACAC GATTCATCTCTTGGGTAG TCAACATGCTCGCAATCC CACAATACACACACATGC CCAAGAAAAAGCCCTAAG CATTCACTTACCTTAACC AGGCTCCAACTGGTAATGG GAACCACAAGGTTATTAG AGTCCACTTTCCTTATAG ATGACTGAATTCCTAAGC ATGGATGTATGAAATGGC GTACTTCATAAATATGGC ACCACAGGCATTGCATTC GAGGTACTTTTGTTCCCC GTTAGTTAGTCTTTCGAG GTAATTTGTTACTTGATAAG ATGGACCTAAAAAATCAG GGCTGAATGCTACATGCTACTTAGC GCTGCAAACATTTTCTGTGCAG CACAGAATAAAAGCACACTGT GACTTTTTTTTCAGAAATTTTC CCATCCACAGTGGGTCTCTCC ATCTGAGCAACTGCTCTTGGCC GAACATTCTGCCTTTCAGGCC

Primer sequences

Table 1. (Continued)

56,57 56 56 56 57 57 58,60 56 53,55 59 53 54 56,58 56,58 58 60 58 57 52 58 55,57 58 58 57 55,56 56,57 56,57 54 56,57 60 61

Temp ⬚C 50 50 50 51 55 53 54 49 51 51 51 49 52 52 54 54 53 54 53 52 56 56 56 51 53 50 51 53 52 54 54

Initial % buffer B

DHPLC conditions

None None None None 7587G/A 7764A/G R2671X P/S2720 None None R/C2840 T2869K None I2957T S3018F 9237G/A D/Y3139 9689delA None C3346R Q3392X I3553T 10856delA IVS61⫹9A/G R/W3739 11340T/C None W3871X Q/R3899 V/I3960 Q/R4048

Positive control

Rossetti et al: Mutation analysis in ARPKD

395

396

Rossetti et al: Mutation analysis in ARPKD

PKHD1. The 66 coding exons of PKHD1 were amplified from genomic DNA as 79 PCR amplicons, ranging from 150 to 370 bp (see Table 1 for details), a size we had previously found optimal for DHPLC analysis [17]. Most exons were amplified as a single fragment but multiple overlapping fragments were required for the larger exons 32, 58, 61, and 67 (see Table 1). To establish appropriate conditions for DHPLC, each fragment was analyzed using the Wavemaker software and idealized conditions determined empirically (see Methods section and Table 1 for details). Most amplicons were analyzed at a single temperature but, because of different distinct melting domains, 21 amplicons were analyzed at two different temperatures (see Table 1 for details). Fragments that generated an aberrant profile were sequenced to determine the DNA change and, if samples were available, segregation was analyzed in the family. Changes predicted to truncate the protein and missense and splicing changes that segregated with the disease (if analysis was possible) and not found in 100 normal controls, were considered to be mutations. An initial test of the screening system employed a small group of ARPKD patients as part of the study to identify the disease gene [14]; characterizing 18 different mutations throughout PKHD1. Mutation screening of a large novel cohort of ARPKD and congenital hepatic fibrosis/Caroli’s disease patients To establish a clearer view of the types of mutations associated with ARPKD, and the prospects for genebased diagnostics, a larger cohort of patients was analyzed. DNA samples were collected from Spanish, American, and British pedigrees. As the preliminary analysis [14] indicated that PKHD1 mutations were found in older patients with a primary diagnosis of congenital hepatic fibrosis and/or Caroli’s disease, as well as typical ARPKD, patients with a wide range of phenotypes were analyzed (see Methods section for details). A total cohort of 61 families was screened for mutations, 47 with a primary diagnosis of ARPKD and 14 congenital hepatic fibrosis and/or Caroli’s disease patients. In 56 cases, DNA from the proband was analyzed and in the remaining five, where the patient died in the perinatal period and no sample was available, DNA from the parents was screened. Examples of mutant DHPLC profiles and segregation analysis are shown in Figure 1. Details of the mutations identified in this cohort and the clinical characteristics of these patients are shown in Table 2. The positions of mutations are spread throughout the gene (Fig. 2) and the mutation details of the population are summarized in Table 3.

A total of 33 different mutations were characterized on 57 mutant alleles; both mutations were identified in 22 families and one mutation in a further 13 (see Table 3 for details). Ten mutations were found on more than one allele and two were particularly frequent, 9689delA (9 alleles) and 5895insA (8 alleles). Three missense changes, T2688K (5 alleles), T36M (4 alleles), and I222V (3 alleles), were also common. Eight different insertion or deletion mutations, predicted to cause a frame-shifting change, were identified, accounting for 26 of the mutant alleles, while 21 missense changes were found on 27 alleles. In addition, four potential intronic or exonic splicing mutations were characterized. One, IVS282A→C, is a clear splicing mutation but the remainder are less clearly pathogenic but may change or create splice sites: IVS33-9T→G, weakening the polypyrimidine tract; IVS43⫹4A→T, lowering the strength of the splice acceptor site; and 657C→T generating a potential cryptic splice site, AAG/G(T)GACT, close to the end of exon 9. Two of the missense mutations may also cause aberrant splicing. We have previously described how the conservative substitution, I222V, might cause cryptic splicing at the end of exon 9 [14]. The A17V mutation may also generate a cryptic splice site, TGG/G(T)AGGT, 4 bp 5⬘ to the normal IVS2 splice site, resulting in a frame-shifting change. All of these mechanisms have been documented as mutagenic in other genes [21, 22], but we have not proved that these changes disrupt splicing herein as we have been unable to amplify the PKHD1 transcript from readily available sources of patient RNA. A17V may also cause disease by disrupting cleavage of the protein. This change is predicted to move the site of cleavage of the signal peptide 4 residues C-terminal to the sequence LSL-HI. In the 22 pedigrees where both mutations were identified, 20 were compound heterozygotes. These consisted of one with two truncating changes; 10 truncating and missense; five, two missense; three missense and splicing; and one truncating and splicing (see Fig. 1 for examples and Table 2 for details). The probands in families PRR-1 and PRR-12 are predicted to be homozygous for 9689delA (from analysis of the parental alleles) but patient material was not available for confirmation. Restriction assays were developed to rapidly screen for the four most common changes, 5895insA, 9689delA, T36M, and I222V (see Methods section for details). Figure 1D shows how this assay was used to trace T36M in pedigree OV-7. To test whether other homozygous cases were missed by our screening method for heteroduplexes, the four common

䉴 Fig. 2. The open reading frame of PKHD1 showing the location of mutations described in this study and our previously reported changes [14]. Mutations detected at least twice in this study are in bold and those described in previous studies underlined [14, 15, 18]. Mutations are colorcoded: green, missense; red, insertion/deletion; purple, splicing; and light blue, nonsense. *Missense mutations that may lead to aberrant splicing.

Fig. 1. Examples of mutation analysis by denaturing high-performance liquid chromatography (DHPLC) and tracing alleles in autosomal-recessive polycystic kidney disease (ARPKD) families. Mutant heteroduplexes are visualized as an earlier elution peak or shoulder on the homoduplex peak. In each case the affected individuals are compound heterozygotes for mutations inherited from both parents, consistent with a recessive disease. Affected individuals are shown as filled shapes and carrier individuals, half filled. *Individuals with the mutation. (A ) Pedigree PRR-9 segregating the mutations 6383delT and I222V. Material was not available from the proband’s affected brother (9F; Table 2) for analysis. (B ) I222V and 383delC segregate in family PRR-15. (C ) The proband in PRR-17 is a compound heterozygote for two truncating mutations, 3761CC→G and 9689delA. (D ) The mutations T36M and IVS43 ⫹ 4A→T segregate in pedigree OV-7 with two affected children, 2537 (brother 2) and 2538 (brother 3). The T36M change is traced using a restriction digest with the enzyme HpyCH4 IV (see Methods section for details).

Patienta Gender

Presentationc

2655

2634

2433

2439

2441

2453

(2F)

(4F)

2644 (9F) 2651

2450

2455

2537 2538 2444 2581

PRR-17

PRR-7

OV-10

OV-16

OV-18

OV-33

PRR-2

PRR-4

PRR-9

PRR-15

OV-30

OV-35

OV-7

2689

2443

2437 2624 2440 2449 2816 (18F)

P728

OV-20

OV-14 PRR-3 OV-17 OV-29 PRR-5 PRR-18

OV-23

(12F)

PRR-12

M M M M F ?

?

M

M M M F

F

M

M M M

M

F

F

M

M

M

F

F

F

Birth 18 years 14 months Birth Birth, died PN Birth, died PN

In utero

Birth, died PN

Birth Birth 3 months 6 months

18 months

Birth

19 years Birth, died PN In utero

Birth, died PN

Birth, died PN

Birth

23 months

Birth

7 months

Birth

Birth, died PN

Birth, died PN

⫹⫹⫹cystic

Renald CHF

Livere

冤

Mutations

Table 2. Clinical phenotype and mutations Changef

冤

冤

9689delA 3229↓ 9689delA 3229↓ Potter’s ⫹⫹⫹cystic CHF 9689delA 3229↓ 9689delA 3229↓ Enlarged kidneys ⫹⫹⫹cystic CHF 9689delA 3229↓ 3761CC→G 1253↓ Enlarged kidneys ⫹⫹⫹cystic CHF 9689delA 3229↓ C3622Y 10865G→A Enlarged kidneys ⫹⫹⫹cystic CHF 9689delA 3229↓ A17V 50C→T* Echogenic kidneys ⫹⫹⫹cystic ? 5895insA 1965↓ C2688F 8063G→T I3468V 10402A→G Echogenic kidneys ⫹⫹cystic CHF 1529delG 509↓ 657C→T G219* T2869K 8606C→A Enlarged kidneys ⫹⫹⫹cystic CHF 3761CC→G 1253↓ I222V 664A→G* Potter’s ⫹⫹⫹cystic ? 5895insA 1965↓ C3346R 10036T→C Potter’s ⫹⫹⫹cystic CHF 383delC 127↓ Y1838C 5513A→G ESRD ⫹⫹cystic CHF 6383delT 2127↓ Potter’s ⫹⫹⫹cystic CHF I222V 664A→ G* Enlarged kidneys ⫹⫹⫹cystic ? 383delC 127↓ I222V 664A→ G* OH Enlarged, ⫹⫹⫹cystic CHF 10856delA 3618↓ IVS33-9T→G ? Splenomegaly ⫹cystic CHF IVS28-2A→C (1076↓) E3502V 10505A→T T2869K 8606C→A Enlarged kidneys ⫹⫹⫹cystic No T36M 107C→T Enlarged kidneys ⫹⫹⫹cystic No IVS43⫹4A→T (2332↓) Echogenic kidneys ⫹⫹⫹cystic CHF D1942G 5825A→G Echogenic kidneys ⫹⫹⫹cystic CHF T2869K 8606C→T P739L 2216C→T Enlarged kidneys, OH ⫹⫹⫹cystic CHF T36M 107C→T P805L 2414C→T I3177T 9530T→C Echogenic kidneys Enlarged, ⫹⫹⫹cystic CD, CHF I757L 2269A→C I3177T 9530T→C Echogenic kidneys ⫹⫹⫹cystic ? 1529delG 509↓ ESRD ⫹⫹cystic CHF 5895insA 1965↓ Enlarged kidneys ⫹⫹⫹cystic CHF 5895insA 1965↓ OH CHF 5895insA 1965↓ Potter’s ⫹⫹⫹cystic CHF 5895insA 1965↓ Enlarged kidneys ⫹⫹⫹cystic ? 5895insA 1965↓

Autosomal-recessive polycystic kidney disease PRR-1 (1F) ? Birth, died PN Potter’s

Pedigree

Age at diagnosisb 58 58 58 58 58 32 58 61 58 2 36 50 61 17 9 55 32 9 36 60 5 34 39 9 5 9 61 IVS33 IVS28 61 55 3 IVS43 36 55 22 3 24 58 22 58 17 36 36 36 36 36 D V P T P V I V

E V T

I

I

Y

C

I

V

C M

A

C

P M M M P P P M P M M P M M P P M P M M P M P P M P M M P M M P M P M P

M P M P (P) M S S M (P)

X X X X X X X X X X X X X X

X

X

X

X

X X

X

X X

X

X

Exon/ Normal IVS Mouseg Segregationh screeni

Continued

N N N N N [15] N N N N [14, 15, 18] N N N N N [15] [14, 15, 18] [14, 15, 18] N N N N [14, 15, 18] N [14, 15, 18] N N N N N [14, 15, 18] N N N N [14, 15, 18] [18] N N N N [14, 15, 18] [14, 15, 18] [14, 15, 18] [14, 15, 18] [14, 15, 18]

Refj

398 Rossetti et al: Mutation analysis in ARPKD

R1046

R1051

2659 2661

M83

M84

HEP-1

M F

F

F

F

Adult Adult

18 years

38 years

36 years

Cholangitis

Cholangitis

Normal Normal

1 cyst

MSK, 1 cyst

Normal

⫹cystic

Splenomegaly CD, CHF

18 years

Renald ⫹⫹⫹cystic ⫹⫹⫹cystic ⫹⫹⫹cystic Echogenic Enlarged, ⫹cystic ⫹⫹⫹cystic

Presentationc

Birth Echogenic kidneys In utero, died PN Enlarged kidneys, OH 2 years Hematuria Birth Enlarged kidneys 21 months Echogenic kidneys Birth Enlarged kidneys

Age at diagnosisb 9689delA S1867N E3529Q T36M P1389T T36M

Mutations

CD, CHF 9689delA T2869K CD 10364delC I3468V CD, CHF V1741M S1833L CD, CHF T2869K I2957T CD 5895insA CD

CHF CHF CHF CHF No CHF

Livere

3229↓ 8606C→A 3454↓ 10402A→G 5221G→A 5498C→T 8606C→A 8870T→C 1965↓

3229↓ 5600G→A 10585G→C 107C→T 4165C→A 107C→T

Changef

58 55 61 61 32 34 55 57 36

58 34 61 3 32 3

M V S V I

V

S E T P T

NP NP NP NP NP NP M P M

P P NP NP M M

X X X X X

X

X X X X X

Exon/ Normal IVS Mouseg Segregationh screeni

N N N N [14] N N [14, 15, 18] [14, 15, 18]

N N N [14, 15, 18] N [14, 15, 18]

Ref j

b

( ), DNA from the patient is unavailable and the parents were screened for mutations PN, died in perinatal period c Potter’s, Potter’s phenotype, consisting of pulmonary hypoplasia, characteristic facies and skeletal abnormalities [32]; OH, oligohydramnios d Cystic ⫹⫹⫹, multiple dilated collecting ducts fill kidney; cystic ⫹⫹, moderately cystic, cystic ⫹; some cysts; MSK, medullary sponge kidney e Liver phenotype; CD, Caroli’s disease; CHF, congenital hepatic fibrosis f Nucleotide change or ↓, frameshift after indicated amino acid; *, possible cryptic splicing change; ( ), predicted consequence of aberrant splicing g Corresponding residue in murine Pkhd1 h P, paternal; M, maternal; NP, samples not available for segregation analysis; S, segregation indicates the mutations are on separate alleles but the parental origin cannot be determined; ( ), parental origin inferred i Screen of 100 normal chromosomes; X, negative j Ref, reference; N, newly described

a

2675

2427 F R1101 M R1050 F 2633 F 2436 F 2447 M hepatic fibrosis 2666 F

OV-1 M94 M96 PRR-6 OV-13 OV-27 Congenital HEP-3

HEP-13

Patienta Gender

Pedigree

Table 2. (Continued)

Rossetti et al: Mutation analysis in ARPKD

399

400

Rossetti et al: Mutation analysis in ARPKD

Table 3. Details of mutations in autosomal-recessive polycystic kidney disease (ARPKD) and congenital hepatic fibrosis (CHF) patient populations Clinical phenotype ARPKD Population and mutation details

Severe

Pedigreesa Both mutations detectedb Single mutation detectedc Mutant allelesd Mutant pedigreesd,e Different mutations Ancestral mutationsa,f

Moderate

10 7 3 17 10

(20) (7) (3) (85) (100) 12 8 (14)

37 11 9 31 20 17 10

All

(74) (11) (7) (41.9) (54.1)

47 18 12 48 30 29 12

(22)

CHF

(94) (18) (10) (51.1) (63.8)

14 4 1 9 5

(28) (1) (1) (32.1) (35.7) 8 6 (7)

(36)

Total 61 22 13 57 35

(122) (19) (11) (46.7) (57.4) 33 12 (41)

a

( ), Disease alleles b ( ), Segregation demonstrated c ( ), Parental origin known d ( ), % of total e At least one mutation detected f Detected at least twice in this study or previously described [14, 15, 18]

Table 4. Haplotypes associated with ancestral mutations Common haplotype marker allele sizes bp Mutation T36M 383delC I222V 1529delG 3761CC→G 5895insA 9689delA

Exon

D6S1344a

D6S1714a

Total mutant alleles

3 5 9 16 32 36 58

180 168 172 168 168 172 176

123 127 127 127 127 129 127

6 2 4 2 2 9 9

Mutant alleles with common haplotype

Normalb frequency of haplotype %

4c 2 4 2 2 5 5 ⫹ 4d

0 22.5 5 22.5 22.5 5 25

a

D6S1344, IVS15; D6S1714, IVS60 In 40 normal chromosomes c All six have 180 bp D6S1344 allele d Segregation not proven in four b

mutations were analyzed by restriction assays, or by DHPLC analysis after adding normal DNA. In addition, the entire gene was screened by DHPLC with normal DNA added in six mutation-negative patients, but neither of these methods identified any further homozygous mutations. To test whether the recurrent mutations had a common ancestral origin, the haplotype associated with the mutation was determined with two intragenic markers, D6S1344 and D6S1714 (see Table 4). In all cases (except T2869K), a common haplotype was found associated with the majority of mutant alleles, including all alleles for four mutations, indicating that they are most likely ancestral changes. As well as the putative pathogenic changes we detected a total of 34 polymorphic changes, 24 of which are exonic (see Table 5 for details). These were defined as polymorphisms because they were detected in the normal population or did not segregate appropriately with the disease phenotype. Several of the changes involve nonconservative amino acid substitutions. DISCUSSION We have described a rapid and effective means to screen for mutations in the PKHD1 gene. Although di-

rect sequencing may be the gold standard for mutation detection in small genes [23], for large multiexon genes, such as PKHD1, a semiautomated screening method is required for cost-effective analysis [16, 18, 24]. A total of 33 different mutations were identified, 26 of which are newly described changes and both disease mutations were characterized in 22 cases. This analysis gave a much clearer view of the type and pattern of mutations associated with ARPKD, revealed some common changes, and provided a hint of genotype/phenotype associations. The overall disease allele detection rate of 57 of 122 (46.7%) is low compared to other large multiexon genes that have been analyzed using similar methodologies, where detection rates of ⬃70% have typically been described [16, 17, 24, 25] and lower than the rate of 61% described in the recent screen of PKHD1 [18]. However, if we subdivide our population into severe and moderate ARPKD, as described by Bergmann et al [18] and separate the congenital hepatic fibrosis/Caroli’s disease patient group (see Methods section for details), the detection levels in the two studies are similar. We identified mutations on 85% and 41.9% of alleles in the severe and moderate groups, respectively, (compared to 77% and 40% [18]). The detection level in the congenital hepatic

Rossetti et al: Mutation analysis in ARPKD Table 5. Details of polymorphisms found in the study population

Designation IVS1-47C/T IVS1-30insA 214C/T 234C/T IVS7⫹19T/C IVS14⫹23G/T 1185T/C 1587T/C 2046A/C 2196C/T R/C760 IVS22⫹13T/G IVS23⫹50C/T IVS23⫹53A/G N/S830 3537T/C 3756G/C A/V1262 4920A/G L/F1709 IVS32⫹42del4 L/V1870 7587G/A 7764A/G IVS53-32C/G 9237G/A D/Y3139 S/R3505 10521C/T IVS61⫹9A/G 11340T/C Q/R3899 V/I3960 Q/R4048

Nucleotide change/ amino acid position

L72 D78 D395 N529 P682 V732 2278C/T

2489A/G N1179 L1252 3785C/T V1640 5125C/T 5608T/G G2529 L2588 A3079 9415G/T 10515C/T H3507 P3780 11196A/G 11878G/A 12143A/G

Exon/IVS

Allele frequency

Reference

IVS1 IVS1 4 4 IVS7 IVS14 15 17 21 22 22 IVS22 IVS23 IVS23 24 30 32 32 32 32 IVS32 35 48 49 IVS53 58 58 61 61 IVS61 63 66 67 67

40/128 2/238 19/128 8/128 32/128 1/238 2/128 8/128 21/128 1/238 37/128 16/128 53/128 14/128 10/128 2/238 2/238 9/238 3/238 1/338 5/128 4/128 33/128 30/128 33/128 33/128 2/128 5/128 5/128 5/128 4/128 37/128 3/238 42/128

Novel Novel Novel [14] Novel Novel Novel [14] Novel Novel [14] Novel Novel Novel [18] Novel [14] [14] [14] Novel Novel Novel [14] [14] Novel [14] [14] [14] [14] Novel [14] [14] [14] [14]

fibrosis/Caroli’s disease group was 32.1% and no comparable group was present in the other study. Therefore, the detection rates for comparable populations were slightly higher in this study, but the number of patients in the severe ARPKD group, in which PKHD1 mutations were most frequently detected, was smaller. The different levels of mutation detection in the various phenotype groups (Table 3) [18] indicate that defining the clinical phenotype is critical. In the severe group, where the ARPKD diagnosis is clear, the detection level of 17 of 20 alleles, with at least one mutation identified in all patients, compares favorably with other large multiexon genes. The level may be enhanced in this group because many mutations appear to be exonic truncating changes [18] (see below) and therefore readily detected by the exonic screening method employed. The much lower detection level in the moderate ARPKD group suggest that this may in part be due to the inadvertent inclusion of other childhood forms of cystic disease (such as early onset, de novo cases of ADPKD) [2, 26]. This suggestion is supported by the finding of no mutation in 17 of 37 (45.9%) moderate ARPKD pedigrees. An

401

alternative explanation, although not supported by linkage studies of ARPKD [12, 13], is that there is significant genetic heterogeneity. Mutation analysis of a recently identified PKHD1 homolog, PKHDL1, did not indicate that it was associated with ARPKD [27]. Careful clinical reassessment, linkage studies and mutation screening of ADPKD and known nephronophithisis genes are now required in the pedigrees in which no PKHD1 mutation was identified. The inclusion of 14 pedigrees where the major disease manifestation is congenital hepatic fibrosis and/or Caroli’s disease, with minimal renal involvement, significantly influenced the overall detection rate. We previously found [14], and confirmed in this study with the identification of mutations in five such families (see Tables 2 and 3), that these predominant liver phenotypes can be associated with PKHD1 mutation. However, the rate of detection of mutant alleles (9 of 28) (32.1%), and patients with at least one mutation (5 of 14) (35.7%), is lower than in the more typical ARPKD population (see Table 3). So, although PKHD1 can be associated with the later onset, predominant congenital hepatic fibrosis/Caroli’s disease phenotype, many of the patients in this population are not due to mutation at this locus. It will now be important to look at this population in light of the mutation data to more precisely define the PKHD1 associated phenotype. Other factors that may account for missed mutations is the complexity of PKHD1 and, in particular, that multiple splice forms may be generated from this locus [14, 15, 28]. Consequently, additional exons may be present that we have not screened, but which form parts of critical PKHD1 splice variants. Comparative analysis to murine Pkhd1 [29] and a greater understanding of the physiologic significance of the different splice forms will help to identify other potential exons. The structure of PKHD1, particularly the large introns, may also explain some of the missed mutations. Changes in the introns, distant from the splice donor and acceptor sites, which were not analyzed in this study, may lead to aberrant splicing [22] and be particularly important in the moderate ARPKD group. Unfortunately, PKHD1 is not widely expressed in material generally available from patients, (e.g., white blood cells) and so it has not been possible to screen for these cryptic splice event by analysis of RNA. Gross DNA deletions may also be a significant form of mutation, although we did not detect such changes in our previous study by Southern blotting [14]. A problem that has been highlighted here is defining what is a mutation in this disorder. The group of truncating and typical splicing mutations are clearly disease associated. Missense changes that are frequently found in ARPKD, segregate appropriately, change highly conserved residues, and were not detected in the normal population (such as T36M, I222V, and C3346R) can also be fairly safely considered mutations [14, 15]. However,

402

Rossetti et al: Mutation analysis in ARPKD

it is much less straightforward to categorize some of the remaining missense changes and potential splice changes situated distant from the canonic sequences. This was evident in this study where in five patients three potential mutations were identified and, using the mutation criteria we defined of testing segregation and absence in the normal population, we were unable to exclude one as a polymorphism. One potential mutation, T2869K, illustrates the dilemma. It was present in several of these pedigrees and, although this residue is not strongly conserved (being valine in the mouse), it is a relatively nonconservative change. Furthermore, it was not found in our screen of normal chromosomes and present in several ARPKD pedigrees. Another example of the uncertainty is the change (D/Y3139) that we have defined as a polymorphism (because of its presence in the normal population) but was previously defined as a mutation [15]. Of course, presence in the normal population is not sufficient to exclude a change as a disease-associated mutation in a recessive disorder, although given the estimated carrier frequency (1 of 70 [13]) and the large number of different mutations, the prevalence of any individual mutation would be expected to be low in the normal population. As data become available from more studies, enrichment of a change in the ARPKD population and segregation consistent with pathogenicity may be the best evidence that it is disease associated. However, whether a change is a mutation or a polymorphism may be something of a gray area in ARPKD, dependent on what combination of alleles are found in an individual or even the combination of changes on an allele. For instance, if one allele is a hypomorphic missense mutation, coinheritance of an inactivating change may be sufficient to cause disease. Whereas, in combination with another hypomorphic allele, that same mutation may not cause disease, or the disease may be milder. A similar situation with the R229Q mutation/polymorphism in NPHS2 has recently been described in late-onset focal segmental glomerulosclerosis [30]. The description of further mutations and phenotype/genotype studies will be required to resolve this uncertainty (see below). An encouraging finding from a diagnostic viewpoint is the detection of more common changes. The mutations 5895insA and 9689delA were found on eight and nine alleles, respectively, and 5895insA was previously described in six other cases [14, 15, 18]. Haplotype analysis (Table 5) indicates that both are probably ancestral rather than recurrent changes. Interestingly, a single haplotype was likely found on all 9689delA alleles, whereas the 5895insA haplotype was divergent on four of nine alleles, probably due to recombination and marker mutations, and indicating a more ancient origin. This is consistent with finding 9689delA only in the Spanish population, while 5895insA has been seen in different studies and different populations [14, 15, 18]. Bergmann et al

[18] also described two ancestral mutations from the Finnish population and a high level of the missense mutation T36M. In the latter case, they suggested that it may have arisen multiple times, as it is a cytosine to thymine change at a CpG dinucleotide (sites of known enhanced mutation [31]) and is associated with many different haplotypes. However, our analysis of six T36M alleles identified a common haplotype on four and a shared rare allele in all six cases (Table 5) with D6S1344 that lies close to this mutation (⬃25 kb). This data indicate a common origin for many T36M alleles, but that the mutation is relatively ancient and recombination has disrupted haplotypes generated with more distant markers. A number of other mutations have been found several times and in more than one study, I222V, 3761CC→G, and I2957T, indicating that they may be relatively widespread ancestral mutations [14, 15, 18]. Detection of one of these characteristic mutations in a patient with an ARPKDlike phenotype (even without finding the other mutation) is now strong evidence that the patient has ARPKD. The initial studies suggested that the combination of mutations in ARPKD was unusual for a recessive disorder with only one homozygous case described (in a consanguineous family), most were compound heterozygotes, and only one pedigree had two truncating mutations [14, 15]. The recent more extensive mutation screen has described 19 homozygous cases, all from known consanguineous families or Finnish families homozygous for the common Finnish mutation R496X, or T36M [18]. However, many different mutations (39) was also found in that study and the present cohort largely reflects that pattern (with 33 different mutations) and just two homozygous cases, for the mutation 9689delA. The two homozygous cases were not known to be from consanguineous families and may reflect the relative high frequency of the 9689delA mutation in Spain. The compound heterozygotes were a combination of truncating and missense (N ⫽ 10), two missense (N ⫽ 5), and other splicing combinations (N ⫽ 4). The variety of mutation combinations and clinical variability of the disorder suggests that phenotype/genotype correlation may be found and that some missense mutations may by hypomorphic alleles. Indeed, the three cases with two truncating mutations in this study, and the 12 other similar cases that have now been published [15, 18], have the severe form of the disease with death in the perinatal period (Table 2). However, it is worth noting that four other patients with the severe phenotype in this study, and 17 other published cases [18] (where both mutations have been identified) have a missense change on one or both alleles, similar to the compound heterozygotes in the moderate ARPKD group. Therefore, it will be important to collect more families where both mutations are known and accurately define the phenotype before genotype/phenotype questions can be clearly answered. The wide range of

Rossetti et al: Mutation analysis in ARPKD

different mutations also means that they will need to be grouped by type and/or position to obtain statistically significant data. This process will have to be done carefully as not all missense mutations are likely to have the same effect on the protein (some may also cause aberrant splicing) (see Table 2) and the position of the mutation may be critical, with the variety of splice forms another important variable. Although many mutations have now been identified in PKHD1 the prospects for gene-based diagnostics still appear difficult. In particular, the relative low level of mutation detection in moderate ARPKD patients and clearly defining a PKHD1 mutation are problematic. Undoubtedly, the identification of more common mutations, especially in particular populations, will aid molecular diagnostics in those locations. As further mutations are defined, and the identity of disease associated changes and polymorphisms can be more clearly established, the prospects for gene-based diagnostics will improve. ACKNOWLEDGMENTS We wish to thank the patients and their families for taking part in the study and Dr. Hiesch, Dr. Germain, Dr. Temple, and Dr. Alvarez for referring patients and providing clinical information. The study was supported by NIDDK grant DK58816, the PKD Foundation (S.R. is a PKDF fellow) and the Mayo Foundation. Reprint requests to Peter C. Harris, Ph.D., Division of Nephrology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905. E-mail: [email protected]

REFERENCES 1. Zerres K, Rudnik-Scho¨neborn S, Steinkamm C, et al: Autosomal recessive polycystic kidney disease. J Mol Med 76:303–309, 1998 2. Guay-Woodford LM: Autosomal recessive polycystic kidney disease: Clinical and genetic profiles, in Polycystic Kidney Disease, edited by Watson ML, Torres VE, New York, Oxford University Press, 1996, pp 237–266 3. Roy S, Dillon MJ, Trompeter RS, et al: Autosomal recessive polycystic kidney disease: Long-term outcome of neonatal survivors. Pediatr Nephrol 11:302–306, 1997 4. Kaplan BS, Fay J, Shah V, et al: Autosomal recessive polycystic kidney disease. Pediatr Nephrol 3:43–49, 1989 5. Blyth H, Ockenden BG: Polycystic disease of kidneys and liver presenting in childhood. J Med Genet 8:257–284, 1971 6. Jorgensen MJ: The ductal plate malformation: A study of the intrahepatic bile duct lesion in infantile polycystic disease and congenital hepatic fibrosis. Acta Pathol Microbiol Scand (Suppl): 1–87, 1977 7. Desmet VJ: Congenital diseases of intrahepatic bile ducts: Variations on the theme “ductal plate malformation.” Hepatology 16: 1069–1083, 1992 8. Zerres K, Mu¨cher G, Bachner L, et al: Mapping of the gene for autosomal recessive polycystic kidney disease (ARPKD) to chromosome 6p21-cen. Nat Genet 7:429–432, 1994 9. Lens XM, Onuchic LF, Wu G, et al: An integrated genetic and physical map of the autosomal recessive polycystic kidney disease region. Genomics 41:463–466, 1997 10. Mu¨cher G, Becker J, Knapp M, et al: Fine mapping of the autosomal recessive polycystic kidney disease locus (PKHD1) and the genes MUT, RDS, CSNK2b, and GSTA1 at 6p21.2-p12. Genomics 48:40–45, 1998 11. Park JH, Dixit MP, Onuchic LF, et al: A 1-Mb BAC/PAC-based physical map of the autosomal recessive polycystic kidney disease

12.

13.

14. 15.

16. 17. 18. 19. 20. 21. 22. 23. 24.

25.

26. 27.

28.

29.

30.

31.

32.

403

gene (PKHD1) region on chromosome 6. Genomics 57:249–255, 1999 Guay-Woodford LM, Muecher G, Hopkins SD, et al: The severe perinatal form of autosomal recessive polycystic kidney disease maps to chromosome 6p21.1-p12: Implications for genetic counseling. Am J Hum Genet 56:1101–1107, 1995 Zerres K, Mu¨cher G, Becker J, et al: Prenatal diagnosis of autosomal recessive polycystic kidney disease (ARPKD): Molecular genetics, clinical experience, and fetal morphology. Am J Med Genet 76:137–144, 1998 Ward CJ, Hogan MC, Rossetti S, et al: The gene mutated in autosomal recessive polycystic kidney disease encodes a large, receptor-like protein. Nat Genet 30:259–269, 2002 Onuchic LF, Furu L, Nagasawa Y, et al: PKHD1, the polycystic kidney and hepatic disease 1 gene, encodes a novel large protein containing multiple immunoglobulin-like plexin-transcription-factor domains and parallel beta-helix 1 repeats. Am J Hum Genet 70:1305–1317, 2002 Xiao W, Oefner PJ: Denaturing high-performance liquid chromatography: A review. Hum Mutat 17:439–474, 2001 Rossetti S, Chauveau D, Walker D, et al: A complete mutation screen of the ADPKD genes by DHPLC. Kidney Int 61:1588–1599, 2002 Bergmann C, Senderek J, Sedlacek B, et al: Spectrum of mutations in the gene for autosomal recessive polycystic kidney disease (ARPKD/PKHD1). J Am Soc Nephrol 14:76–89, 2003 Dode´ C, Rochette J, Krishnamoorthy R: Locus assignment of human a globin mutations by selective amplification and direct sequencing. Br J Haemat 76:275–281, 1990 Harris PC, Thomas S, Ratcliffe PJ, et al: Rapid genetic analysis of families with polycystic kidney disease by means of a microsatellite marker. Lancet 338:1484–1487, 1991 Nakai K, Sakamoto H: Construction of a novel database containing aberrant splicing mutations of mammalian genes. Gene 141:171– 177, 1994 King K, Flinter FA, Nihalani V, et al: Unusual deep intronic mutations in the COL4A5 gene cause X linked Alport syndrome. Hum Genet 111:548–554, 2002 Klein B, Weirich G, Brauch H: DHPLC-based germline mutation screening in the analysis of the VHL tumor suppressor gene: Usefulness and limitations. Hum Genet 108:376–384, 2001 Choy YS, Dabora SL, Hall F, et al: Superiority of denaturing high performance liquid chromatography over single-stranded conformation and conformation-sensitive gel electrophoresis for mutation detection in TSC2. Ann Hum Genet 63:383–391, 1999 Le Marechal C, Audrezet MP, Quere I, et al: Complete and rapid scanning of the cystic fibrosis transmembrane conductance regulator (CFTR) gene by denaturing high-performance liquid chromatography (D-HPLC): Major implications for genetic counselling. Hum Genet 108:290–298, 2001 Zerres K, Rudnik-Scho¨neborn S, Deget F, et al: Childhood onset autosomal dominant polycystic kidney disease in sibs: Clinical picture and recurrence risk. J Med Genet 30:583–588, 1993 Hogan MC, Griffin MD, Rossetti S, et al: PKHDL1, a homolog of the autosomal recessive polycystic kidney disease gene, encodes a receptor with inducibile T lymphocyte expression. Hum Mol Genet 12:685–698, 2003 Xiong H, Chen Y, Yi Y, et al: A novel gene encoding a TIG multiple domain protein is a positional candidate for autosomal recessive polycystic kidney disease. Genomics 80:96–104, 2002 Nagasawa Y, Matthiesen S, Onuchic LF, et al: Identification and characterization of Pkhd1, the mouse orthologue of the human ARPKD gene. J Am Soc Nephrol 13:2246–2258, 2002 Tsukaguchi H, Sudhakar A, Le TC, et al: NPHS2 mutations in late-onset focal segmental glomerulosclerosis: R229Q is a common disease-associated allele. J Clin Invest 110:1659–1666, 2002 Cooper DN, Krawczak M, Antonorakis SE: The nature and mechanisms of human gene mutation, in Metabolic and Molecular Bases of Inherited Disease, 7th ed, edited by Scriver C, Beaudet AL, Sly WS, Valle D, New York, McGraw-Hill, 1995, pp 259–291 Potter EL: Facial characteristics of infants with bilateral renal agenesis. Am J Obstet Gynecol 51:885–888, 1964