●

vchTtflruwlrhfcD©

Id

AnasdgdsgsAbpiphtm

gPiae

A

Haplotype Analysis Improves Molecular Diagnostics of AutosomalRecessive Polycystic Kidney Disease

Mark B. Consugar, MS, Sarah A. Anderson, BS, Sandro Rossetti, MD, V. Shane Pankratz, PhD,Christopher J. Ward, MBChB, PhD, Roser Torra, MD, Eliecer Coto, PhD, Monif El-Youssef, MD,

Sibel Kantarci, PhD, Boris Utsch, MD, Friedhelm Hildebrandt, MD, William E. Sweeney, PhD,Ellis D. Avner, MD, Vicente E. Torres, MD, PhD, Julie M. Cunningham, PhD,

and Peter C. Harris, PhD

Background: Autosomal recessive polycystic kidney disease (ARPKD) is characterized by wide phenotypicariability, ranging from in utero detection with enlarged, echogenic kidneys to an adult presentation withongenital hepatic fibrosis. The ARPKD gene, PKHD1, covers about 470 kb of DNA (67 exons), and mutation studiesave found marked allelic heterogeneity with a high level of novel missense changes and neutral polymorphisms.o improve the prospects for molecular diagnostics and to study the origin of some relatively common mutations,he authors have developed a strategy for improved ARPKD haplotyping. Methods: A protocol of multiplex PCR anduorescence genotyping in a single capillary has been developed to assay 7 highly informative simple sequenceepeat (SSR) markers that are intragenic or closely flanking PKHD1. Results: Examples in which haplotype analysis,sed in combination with mutation screening, improved the utility of molecular diagnostics, especially in families inhich just a single PKHD1 mutation has been identified, are illustrated. The new markers also allow screening for

arger DNA deletions, detecting unknown consanguinity and exploring the disease mechanism. Analysis of 8ecurring mutations has shown likely common haplotypes for each, and the divergence from the ancestralaplotype, by recombination, can be used to trace the history of the mutation. The common mutation, T36M, wasound to have a single European origin, about 1,225 years ago. Conclusion: Improved haplotype analysis of ARPKDomplements mutation-based diagnostics and helps trace the history of common PKHD1 mutations. Am J Kidneyis 45:77–87.2004 by the National Kidney Foundation, Inc.

NDEX WORDS: Autosomal recessive polycystic kidney disease (ARPKD); PKHD1; haplotype analysis; molecular

iagnostics; prenatal diagnosis; ancestral mutations.

p1cbb

tGMldtrURW

b

F

mS

UTOSOMAL RECESSIVE polycystic kid-ney disease (ARPKD) is an important re-

al cause of morbidity and mortality in neonatesnd infants (incidence 1 in 20,0001-3). In its mostevere form, the disease is detected in utero oruring the perinatal period by the presence ofreatly enlarged kidneys and results in neonataleath in 20% to 30% of patients.4-6 The progno-is for patients surviving the neonatal period isood, although 20% to 45% progress to end-tage renal disease (ESRD) within 15 years.6-9

nother pathognomonic feature of ARPKD isiliary dysgenesis, resulting in congenital he-atic fibrosis and, in some cases, dilatation of thentrahepatic bile ducts (Caroli’s disease). A pro-ortion of patients first present later in child-ood, or even as adults, with complications ofhe liver disease and less marked renal involve-ent.9-13

ARPKD appears, from linkage studies, to beenetically homogeneous with the disease gene,KHD1 (localized to chromosome region 6p12),

dentified in 2002.13-16 PKHD1 is a large gene,bout 470 kb, containing 67 exons (additional

xons may be employed in alternatively spliced

merican Journal of Kidney Diseases, Vol 45, No 1 (January), 200

roducts) with a longest open reading frame of2,222 bp.13,14 The large ARPKD protein, fibro-ystin, in common with other PKD proteins, haseen localized recently to primary cilia and theasal body, although its function is unknown.17-20

From the Division of Nephrology, Department of Labora-ory Medicine, Division of Biostatistics, and Division ofastroenterology, Mayo Clinic College of Medicine, Rochester,N; Department of Nephrology, Fundació Puigvert, Barce-

ona, Spain; Laboratorio de Genética Molecular-Institutoe Investigatión Nefrológica (IRSIN-FRIAT), Hospital Cen-ral de Asturias, Oviedo, Spain; Cold Spring Harbor Labo-atory, Cold Spring Harbor, NY; Department of Pediatrics,niversity of Michigan, Ann Arbor, MI; and Children’sesearch Institute, Medical College of Wisconsin, Milwaukee,I.Received July 2, 2004; accepted in revised form Septem-

er 9, 2004.Supported by NIDDK grant DK58816 and the PKD

oundation.Address reprint requests to Peter C. Harris, PhD, Depart-

ent of Nephrology, Mayo Clinic Rochester, 200 First StreetW, Rochester, MN 55905. E-mail: harris.peter@mayo.edu© 2004 by the National Kidney Foundation, Inc.0272-6386/04/4501-0008$30.00/0

doi:10.1053/j.ajkd.2004.09.009

5: pp 77–87 77

pbsi3mblsnmucosc1crr

w8wwhlpiii

hdfucitfh

gdit ination n at t

CONSUGAR ET AL78

The identification of PKHD1 has allowed theattern of mutation associated with ARPKD toe characterized. Four large mutation screeningtudies (and the original gene descriptions) havedentified a total of 139 different mutations on09 nonconsanguineous alleles.1,12-14,21-24 Someutations are predicted to truncate the protein,

ut the majority are missense changes.1,21 Pre-iminary genotype/phenotype correlations havehown a universally severe outcome (death in theeonatal period) associated with 2 truncatingutations.12,22,23 The majority of mutations are

nique to a single family, and most patients areompound heterozygotes, but there is evidencef a few more common changes, especially inpecific geographic populations.12,22 The mostommon mutation, T36M, accounts for about7% of all mutant alleles1; however, there areonflicting data on whether it is an ancestral orecurrent mutation.12,21,22,24 Mutation detection

Fig 1. Map of the ARPKD gene, PKHD1, shows (top)ene with the physical locations of markers on chromescribed in the Marshfield (Marsh) and Genethon (Gen

s an enlarged image of PKHD1 indicating the intron–exhis study with their Mb locations indicated. Recombumber of informative meioses for the interval are show

ates in the 4 studies were highest in patients f

ith the most severe form of the disease (up to5%). In moderately affected children and adultsith predominant liver disease, the detection rateas more modest (about 40%12,22), althoughigher rates have been found in a recent unpub-ished study.21 The variable detection rate mayartly reflect the inclusion of non-ARPKD casesn some screened populations and probably alsondicates that a significant number of mutationsn this dispersed gene remain undetected.

Microsatellite markers used to localize PKHD1ave been employed in linkage-based moleculariagnostics of ARPKD with significant demandor antenatal diagnosis.16 This analysis has beentilized successfully in many cases, althoughrossovers between the (sometimes distant) flank-ng markers has complicated diagnostic predic-ions in some instances (Fig 1).16 Furthermore,or the linkage-based approach, it is necessary toave a certain diagnosis of ARPKD and DNA

sitions of existing microsatellite markers flanking thee 6 shown in megabases (Mb) and genetic distancesps and Mücher study34 in centiMorgans. At the bottomucture and positions of microsatellite markers used inn (Rec) events detected in this study relative to thehe bottom.

the poosome) maon str

rom an affected family member. However, in

mcntAt

hptmm

abI(mum plotyp

HAPLOTYPE ANALYSIS OF ARPKD 79

any cases, material is unavailable from de-eased neonates or previously terminated preg-ancies. The identification of PKHD1 has raisedhe prospects for mutation-based diagnostics forRPKD,24,25 but this has been complicated by

Fig 2. Examples of infantile polycystic kidney disssociated haplotypes are outlined. (A) Haplotype anaetween the markers MBC-1 and MBC-2. (B) The mutat

I-2. The maternally inherited mutation has not been ideblue) and traced to just the affected sib (II-1). (C) Fautation (G470D) is linked to PKHD1 by haplotype

nknown etiology. (D) A phenotypically ARPKD-like fautation because all affected siblings have different ha

he large gene size, marked allelic heterogeneity, t

igh level of missense changes (and neutralolymorphisms), and relatively low mutation de-ection rate in some populations.12,22,23 In only ainority of families have 2 clearly pathogenicutations been identified.1,21 To aid the diagnos-

pedigrees showing the PKHD1 haplotypes. ARPKD-hows a recombination in the T36M-bearing haplotype95insA is inherited from the father by siblings II-1 and, but a disease-associated haplotype can be deducedith atypical presentation of ARPKD with 1 detected

is. The mother (shaded circle) has hepatic cysts of43) is shown by linkage not to be caused by PKHD1

es.

easelysis sion 58ntifiedmily w

analysmily (M

ic process, we have identified new microsatellite

mdfs

D

RgAolaopo

msrapS5cfiaok

dupweikAcsgch

shhpcb

IM

Nur(SpaDmdDDtcrd(

S5(T(maP17rMwtt

5DMMDD3

rom 3=

CONSUGAR ET AL80

arkers within and closely flanking PKHD1 andeveloped a system to rapidly haplotype ARPKDamilies. These markers have been used also totudy the history of common ARPKD mutations.

METHODS

etails of the Study CohortThe study was approved by the relevant Institutional

eview Boards or Ethics Committees, and all participantsave informed consent. Clinical details of the majority ofRPKD families in the study have been described previ-usly.12,13 The clinical and other molecular details of fami-ies with newly identified mutations (T36M and 5895insA)re described in a manuscript in preparation, and the originsf the pedigrees are indicated. In this study, the severehenotype is defined as death in the neonatal period, and allther cases are designated as moderate disease.Patient II-4 of a US family, M97 (Fig 2C), presented at 9onths with an enlarged liver and kidney. Ultrasound analy-

is showed increased echogeneity of these organs, andesults of a liver biopsy showed ductal plate malformationnd congenital hepatic fibrosis typical of ARPKD. Theatient also had portal hypertension and an enlarged spleen.ubsequently, analysis of her sisters (II-1 and II-3) at 10 andyears both showed multiple bilateral small parenchymal

ysts and renal enlargement but normal livers (no histologyndings were available). The fourth sister (II-2 at 8 years)nd the father had normal ultrasound scans, but examinationf the mother found multiple hepatic cysts and normalidneys.A second US family, M43 (Fig 2D), had possible ARPKD

iagnosed in 3 affected siblings. The parents had negativeltrasound examinations at 30 and 34 years of age. Theroband (II-2) was diagnosed in utero (32 weeks’ gestation)ith enlarged kidneys. At 11 years, she had markedly

nlarged, hyperechogenic kidneys, with 3 to 4 visible cystsn each, and a normal liver. Her older sister (II-1) had cysticidneys diagnosed at 26 months after the diagnosis in II-2.t 13 years, she had increased echogenicity and 3 to 4 small

ysts in each kidney and a few small liver cysts. The thirdister, II-3, was diagnosed in utero. She had enlarged echo-enic kidneys, with visible cysts in both kidneys, and tinyysts in the liver at 6 years. All the affected siblings were

Table 1. Details of PK

Name Repeat*Position

(kb)† Location Forwa

=AR (GT)15 �68 5= TTGTCAGC6S1344 (AGAT)8 29 IVS 15 TGGTTGTTBC-1 (GT)18 196 IVS 43 TTCAGTTTTBC-2 (TC)15(AC)19 283 IVS 52 TGAGGTGA6S243 (AC)24 388 IVS 60 TCCTATGA6S1714 (GT)18 395 IVS 60 AGCACCAA=AR (AC)14 �34 3= TTCTCTTAG

*Repeat length of most common allele.†Distance from transcriptional start site; 3=AR, distance f

ypertensive from birth (or when diagnosed) but none had m

plenic enlargement. It is not known if any of the childrenave congenital hepatic fibrosis, because no liver biopsiesave been performed. An unusual feature of this family isersistent mild to moderate hypoglycemia in all affectedhildren. One other similar case of unknown etiology haseen described26 (Müller, unpublished data).

dentification, Amplification, and Analysis of theicrosatellite MarkersThe genomic sequence of PKHD1 obtained from contig

T_007592.13 (www.ncbi.nlm.nih.gov) was scanned forninterrupted di-, tri-, and tetranucleotide simple sequenceepeats (SSRs) of n � 9 using the RepeatMasker programftp.genome.washington.edu/cgi-bin/RepeatMasker). OnceSRs were identified, polymerase chain reaction (PCR)rimers were designed (using the MacVector program) tomplify as fragments of 110 to 260 bp (Tables 1 and 2).6S1960 and D6S466 were used as distant 5= flankingarkers in the haplotype studies (Figs 1 and 3). Previously

escribed primers were used for D6S1344, D6S1714,6S1960, and D6S466, but new primers were designed for6S243 (Table 1).27 The physical and genetic locations of

he various markers, indicated in megabase (Mb) pairs orentiMorgans (cM) on Fig 1, are taken from the MapViewepresentation of the human genome and include geneticistances from the Genethon and Marshfield genetic mapswww.ncbi.nlm.nih.gov/mapview/maps.cgi).

Genomic DNA was isolated from peripheral blood and theSRs amplified by PCR using standard methods. Briefly,0 ng of DNA was amplified using 5 pmoles of each primerApplied Biosystems, Foster City, CA; Integrated DNAechnologies, Coralville, IA), 5 pmoles of each dNTPInvitrogen Life Technologies, Carlsbad, CA), 2.5 to 4.5mol/L MgCl2, 1 U of HotStar Taq (Qiagen, Valencia, CA),

nd the supplied PCR buffer in a total volume of 25 �L. TheCR amplification program typically consisted of: 94° C for20s and 35 cycles of: 94° C, 60s; 56 to 58° C, 60s; and2°C, 60s, followed by 72°C for 10 minutes. All PCReactions were performed on an MJ Research (Waltham,

A) PTC-255 Tetrad thermocycler. One primer was taggedith a fluorophor (FAM-6, HEX, ROX or NED; Table 1) and

he second GTTT tailed to help prevent nontemplate nucleo-ide additions to the product.28

A system was also developed to amplify all markers as 2

icrosatellite Markers

r (5=3 3=) Reverse Primer (5=3 3=) Label

TAGTCAC GCACCAACCTAATAGTCTCTTC HEXTCTGAAC CTGGTCTGCCTTTTTAACAT 6-FAMCCATAC TTAAAGACCCTATTCCTTCC HEXAGAAGAGC AAAGCCAGTTTCCTGACAC NED

ACCTAGAATTG GCGAGAACTATTATGCTGGG ROXACAGAAC TGTATCCACTGCCATCACTT ROXCTAGGATAC CAAATGGAACATGAAAAAAG 6-FAM

end of transcript.

HD1 M

rd Prime

TTCCGCCTTCCTCCCGAGTG

GATGTATGACTTCTT

ultiplex PCR reactions consisting of 5=AR, MBC-1 and

3Ms52mM2D0I�Tsa

F

33Sc

soesma1fusaasd(

A

teb

5

D

M

M

HAPLOTYPE ANALYSIS OF ARPKD 81

=AR (Multiplex 1), and D6S1344, D6S243, D6S1714, andBC-2 (Multiplex 2), using the AccuPrime multiplex PCR

ystem (Invitrogen Life Technologies). For Multiplex 1,0 ng of DNA was amplified using 5 pmoles of each primer,.5 �L of 10X AccuPrime PCR Buffer II (containing 15mol/L MgCl2, 2 mmol/L dNTPs), 1.5 �L 50 mmol/LgCl2, and 0.5 �L AccuPrime Taq in 25�L. For Multiplex

, 50 ng of DNA was amplified using 5 pmoles of the6S1344 and D6S1714 primers, 15 pmoles for D6S243, and.5 pmoles for MBC-2, 2.5 �L 10X AccuPrime PCR BufferI (containing 15 mmol/L MgCl2 and 2 mmol/L dNTPs), 0.5L 50 mmol/L MgCl2, and 0.5 �LAccuPrime Taq, in 25 �L.he PCR amplification program for both multiplexes con-isted of 94°C, 120s, and 35 cycles of 94°C, 30s; 57°C, 30s;nd 68°C, 45s, followed by 68°C for 10 minutes.

luorescence GenotypingAll samples were assayed on an Applied BioSystems

100 Genetic Analyzer and visualized using the Genotyper.7 software (Applied Biosystems, Foster City, CA). The 7SR amplicons were designed so that the PCR reactions

Table 2. Details of Allele Fr

Marker Allele Length (bp) Frequency (%) Het (%

=AR 76.0200 8.7202 25.4204 23.9206 17.4208 15.9210 5.8212 0.7214 2.2

6S1344 71.2164 41.8168 20.9172 29.1176 6.7184 1.5

BC-1 40.8117 0.8121 75.0123 3.0125 14.4127 6.8

BC-2 80.9138 9.2140 19.2162 10.0164 30.0166 26.2168 4.6172 0.8

Abbreviation: Het, observed heterozygosity.

ould be pooled and resolved in a single capillary with p

eparation achieved by size and/or the emission wavelengthf the 4 different fluorophors (Table 1). The detected size ofach allele was corrected to the true size (determined byequencing the product from one or more individuals ho-ozygous for a common allele). The D6S1344 allele associ-

ted with the T36M mutation was found to be 176 bp (not80 bp, as previously described12) on resequencing. Allelerequencies were calculated from analysis of 121 to 138nrelated normal chromosomes and heterozygosities ob-erved in 115 to 125 unrelated normal or ARPKD individu-ls. The disease-related haplotype was determined by manualnalysis of families. Common ARPKD mutations were as-ayed in families as previously described by restrictionigest, denaturing high-performance liquid chromatographyDHPLC) analysis, and sequencing.12

ging the T36M MutationAn estimate of the age of the T36M mutation was ob-

ained by assuming that all cases are derived from a singlevent on a specific haplotype that diverged owing to recom-ination and marker mutation. The rate of recombination is

cies of the PKHD1 Markers

arker Allele Length (bp) Frequency (%) Het (%)

S243 87.9233 2.5235 5.7237 6.6239 0.8241 16.4243 2.5245 10.6247 17.2249 13.9251 13.9253 3.3255 4.9

S1714 62.1113 1.4119 1.4121 0.7123 2.1125 15.7127 59.3129 12.1131 5.7133 1.4

AR 52.9130 10.4132 50.0134 39.6

equen

) M

D6

D6

3=

roportional to the distance from the mutation and, consider-

Trtassi

CONSUGAR ET AL82

Fig 3. Diagram of haplotypes associated with the ancestral mutations 9689delA (A), I222V (B), 5895insA (C), and36M (D). The first row shows the allele sizes of the common ancestral haplotype, and the colored area indicates theegion of this haplotype that is shared in the different pedigrees. The geographic origin of the families is shown onhe right. Asterisks show cases in which the phase could not be determined, but one allele size matched thencestral haplotype. The ^ symbol indicates a homozygous case, but that the phase was unknown. Two allele sizeseparated by a slash indicate cases in which the allele size associated with the mutation is either one of thosehown (phase unknown) and different from the ancestral haplotype. The T36M haplotype in P777 [II1] shows a novel

ntrafamilial haplotype due to a recombination event (Fig 2A).

ia(l(1eT51ptmcoalwtTta

DA

llIaed(7aflwgFbgMgo6f22slmaf

gswtbsndi(tnilit

tictbmptldlihTtas

ltfcPampfcrlfua

HAPLOTYPE ANALYSIS OF ARPKD 83

ng the known crossover rate in the interval between D6S1344nd D6S1714 and the physical distance between the markersFig 1), we can estimate genetic distances (cM) assuming ainear relationship: 5=AR-(0.32)-T36M-(0.06)-D6S1344-0.54)-MBC-1-(0.28)-MBC-2-(0.32)-D6S243-(0.03)-D6S-714-(0.35)-3=AR. From the haplotype data (Fig 3D) we canstimate the number of crossovers that have occurred in the36M haplotype with the various markers: 0, D6S1344; 3,=AR; 3, MBC-1; 7, MBC-2; 10, D6S243 and D6S1714; and1, 3=AR. We applied a version of the Poisson branchingrocess model,29 recognizing that the location of the muta-ion was known. A first-order approximation, similar to theethod used by Xiong and Guo,30,31 was applied to obtain

losed-form estimates that allowed for marker mutationsccurring at a rate of 3.3 � 10�4 per marker per generationnd the available genetic distance data for multiple markeroci. We accounted for phase-ambiguous haplotypes byeighting the data by the probability of the proposed haplo-

ype, conditional on the genotype-consistent haplotypes.he conditional probabilities were obtaining using the haplo-

ype frequencies estimated by an implementation of the EMlgorithm.32

RESULTS

evelopment and Characterization of theRPKD Markers

Three highly informative, existing microsatel-ite markers are intragenic to PKHD1; 2 in thearge IVS60 (D6S1714 and D6S243) and 1 inVS15 (D6S1344; also called D6S1956; Fig 1nd Tables 1 and 2). The existing flanking mark-rs commonly used for linkage analysis in thisisorder are relatively distant from the geneFig 1; about 400 kb to about 6 Mb 3= and about00 kb to about 2.4 Mb 5=16,33,34). These result inrecombination rate between the most distant

anking markers of about 8% (�12% in women)ith a much lower but detectable level of intra-enic crossovers (about 1%, about 2% in women;ig 1) in this large gene. To identify new markersoth within and closely flanking the gene, theenomic sequence was screened for SSRs (seeethods for details). Four new markers, 2 intra-

enic (IVS43; MBC-1 and IVS52; MBC-2), andnes closely flanking each end of the gene (about8 kb 5=; 5=AR and about 34 kb 3=; 3=AR) wereound and characterized (Fig 1 and Tables 1 and). The markers were designed to be amplified as

multiplex PCR reactions and resolved in aingle capillary of an ABI 3100 Genetic Ana-yzer (Table 1 and see Methods for details). Thearker set was typed in up to 80 normal individu-

ls (including samples from a large reference

amily) and 181 samples from 59 ARPKD pedi- c

rees, most of which had previously beencreened for mutation.12,19 In all families inhich samples were available, a specific haplo-

ype associated with each PKHD1 mutation coulde determined. Analysis of 200 informative meio-es showed that the markers were stable with noew alleles detected and no significant linkageisequilibrium between markers. Linkage stud-es in phase-known meiosis found 2 crossoversboth maternal) in a total of 138 phase-informa-ive meioses (Figs 1 and 2A). Although theumber of meioses studied here is small, they dondicate that intragenic recombination occurs at aow, but significant, level between the new flank-ng markers (about 1% to 2%), consistent withhe published linkage data.

The utility of this improved haplotype analysiso aid mutation-based molecular diagnostics isllustrated in Fig 2B. In this case, a relativelyommon PKHD1 mutation, 5895insA, was de-ected in the patient by DHPLC12 and shown toe inherited from the father, but the secondutation remained unidentified. The finding of 1

athogenic mutation, even without the identifica-ion of the second (in a patient with an ARPKD-ike phenotype) is very strong evidence that theisease is PKHD1 associated. Mutation and hap-otype analysis showed that 1 sibling (II-2) inher-ted the mutation but not the maternal diseaseaplotype and therefore is an unaffected carrier.he third sibling inherited neither affected haplo-

ype. In the setting of prenatal diagnostics, thisnalysis could be used for reliable antenatalcreening when just 1 mutation is identified.

A second family, M97 (Fig 2C), shows howinkage can aid in the diagnosis of ARPKD whenhe disease presentation is atypical (see Methodsor details) and just a single novel missensehange is detected by mutation screening ofKHD1. In this case, the nonconservative aminocid substitution G470D (a position conserved inouse and the related fibrocystin-L protein and

art of a recently defined protein domain35) wasound in the mother and segregated with renalystic disease in the siblings. The paternal de-ived mutation was not identified, but microsatel-ite analysis showed that a single haplotype wasound in the 3 affected relatives (and not thenaffected case), consistent with PKHD1 link-ge (Fig 2C). Diagnosis in this family is compli-

ated by the mild cystic diseases, limited to the

khmtoharefisPw

HM

ftplDoscwwcmi4dhwlmmtgD

smwsbflm

smm1e(l

Aiate1lmcicficictA(ptbiat

tdsphalcahattwwhf

CONSUGAR ET AL84

idney, in II-1 and II-3, and that the mother hadepatic cysts. Nevertheless, the combination ofutation and haplotype analysis indicates that

his is most likely an ARPKD family. The etiol-gy of the disease in the mother is unknown (aseterozygous mutations are not usually associ-ted with a phenotype in ARPKD) and mayepresent a case of isolated polycystic liver dis-ase. The linkage-based approach is also usefulor identifying unlinked families. One such fam-ly (M43; Fig 2D) with an ARPKD-like diagno-is (see Methods for details) was unlinked toKHD1 and, as expected, no PKHD1 mutationsere detected.

aplotypes Associated With Recurringutations

Mutation analysis of ARPKD families hasound a number of ancestral or recurrent muta-ions, either associated with specific geographicopulations or more widely distributed.1,21 Pre-iminary haplotype analysis with the markers6S1344 and D6S1714 suggested an ancestralrigin for many of these,12,21 but the new markeret has allowed a more complete analysis. Threehanges, 383delC, 1529delG, and 3761CC3 G,ere observed in 2 families and, in each case,ere associated with a specific haplotype. In the

ase of one other mutation (I3177T), the com-on haplotype was variable at just the 5= flank-

ng marker (5=AR). More data were available onother mutations, and these were analyzed in

etail (Fig 3). In each case, a likely commonaplotype could be deduced, but for some thereas considerable divergence in different fami-

ies. The pattern of alleles associated with eachutation, with markers closest to the mutationost consistently matching the common haplo-

ype, suggested a common origin with diver-ence owing to recombination (see Fig 3 andiscussion for detailed analysis).

DISCUSSION

We have described a group of PKHD1 micro-atellite markers for improved haplotyping. Thesearkers will increase the number of families inhich molecular diagnostics for ARPKD is pos-

ible and improve its reliability. The linkage-ased analysis, with 7 markers spanning andanking the gene, is simply achieved with 2

ultiplexed PCR reactions and analyzed in a f

ingle capillary. The linkage data complementsutation analysis, and the combined linkage/utation approach will aid diagnostics when justmutation has been detected (Fig 2B and C),

xclude some unlinked (misdiagnosed) familiesFig 2D), help detect consanguinity, and identifyarge deletion mutations (see below for details).

A mutation-based screening approach toRPKD diagnostics has the advantage in that it

s possible even if material from a previouslyffected fetus or neonate is not available and/orhe diagnosis of ARPKD is not certain.25 How-ver, the complexity of PKHD1 means that onlymutation often is detected, especially in fami-

ies with less severe disease.12,22,23 Furthermore,any putative mutations are novel missense

hanges with unproven pathogenicity.1,21 If fam-ly members are available, linkage analysis canomplement the mutation-based approach by de-ning the disease-associated haplotypes. Theombined linkage/mutation screening approachs particularly useful in families in which 1haracteristic PKHD1 mutation has been de-ected (and therefore is very likely bona fideRPKD), but the second remains to be found

see Fig 2B for example). The highly polymor-hic nature of the markers means that the haplo-ype is informative in practically all families. Allut 1 parent (from a total of 80 analyzed) werenformative for at least 2 markers, with an aver-ge of 4.9 markers polymorphic in each haplo-ype.

Linkage-based analysis generally requires ma-erial from an affected sibling to identify theisease-associated haplotypes and a firm diagno-is of ARPKD.16 However, the combined ap-roach often may be informative if 1 mutationas been detected even without DNA from anffected subject, if samples from unaffected sib-ings are available. Linkage-based analysis alonean be used for diagnostics if material from anffected sibling is available, even if no mutationas been detected.16 This analysis is highly reli-ble if there are no recombinations in the haplo-ype and the disease is caused by PKHD1 muta-ion. However, in mutation screens of familiesith an ARPKD diagnosis, no PKHD1 mutationas detected in 34.4% of families, with an evenigher percentage (42.7%) in less severely af-ected cases.12,22,23 Although in a proportion of

amilies this is probably caused by both PKHD1

mcsntdoiwlhd(cthtccbfalb

sdigmaswtospapfm

oPwiotgld

tfiS

fdfSc2faflamihtabbcnasw

lcaaoamceidEoT(c(pCck1s

HAPLOTYPE ANALYSIS OF ARPKD 85

utations being missed, it is likely that some areaused by other disorders that clinically appearimilar to ARPKD, such as nephronophthisis, deovo cases of early-onset ADPKD, glomerulocys-ic disease, or other hepatorenal fibrocystic syn-romes, and could possibly reflect genetic heter-geneity of ARPKD.2,3,16,36 In the familyllustrated here (Fig 2D), an ARPKD diagnosisas proposed because of the cystic kidney and

iver disease in 3 siblings and lack of a familyistory (see Methods for details). Although theisease is not the classical severe form of ARPKDwith evidence of hypogylcemia) and there is noonfirmed congenital hepatic fibrosis, the presen-ation is not dissimilar to other cases in which weave found PKHD1 mutations,12,13 and a diagnos-ic service is likely to regularly receive suchases. Therefore, linkage-based analysis aloneould be considered risky if no mutation haseen identified, especially if the disease in theamily is not entirely typical of ARPKD. Linkagenalysis can, however, help identify such fami-ies and exclude these from PKHD1 mutation-ased diagnostics.In addition to haplotype analysis, the evenly

paced markers will be useful for detecting largeeletions, seen as hemizygosity in linkage stud-es.12 One possible large deletion case was sug-ested in this study by a single allele size atany adjacent intragenic markers (including rare

lleles), but parental DNA to test if this repre-ented hemizygosity or rare homozygous allelesas unavailable. The common SNPs spread

hroughout the gene will also aid in the detectionf larger deleted regions.21 In addition, the markeret and family linkage studies will highlightossible cases of unknown consanguinity, seens homozygosity for all PKHD1 markers in theatient, and indicate that special care is requiredor mutation analysis if heteroduplex detectingethods, such as DHPLC, are employed.37

The detailed haplotype analysis has shed lightn the likely origin of some of the more commonKHD1 mutations. Many of these are associatedith a single haplotype that diverges rarely,

ndicating a common and probably fairly recentrigin. An example of such a change is 9689delAhat was found on 9 alleles, including in homozy-osity in 2 individuals,12 with the common hap-otype differing only in 1 case, resulting in a

ifferent 5=AR allele (Fig 3A). The hypothesis c

hat this is a recent mutation is supported by thending that all but 1 of 12 described cases are ofpanish or Portuguese origins.12,21,24

A number of recurring mutations have beenound in several studies in different European-erived populations. Analysis of I222V in 4amilies, 3 from Spain and 1 from the Unitedtates,12,13 showed a common haplotype in 2ases and a third that differed at just D6S243 (bybp, suggesting a marker mutation; Fig 3B). The

ourth case only shared the 5=AR and D6S1344lleles with the common haplotype (the markersanking the mutation), and the alleles of D6S1714nd 3=AR. Because the allele sizes of the 3=arkers are common ones (Fig 3B and Table 2),

t is likely that the divergence from the commonaplotype is caused by a single crossover be-ween D6S1344 and MBC-1. In a similar way,nalysis of 5895insA indicates a common originut that the haplotype is broken up in many casesy recombination at the 5= or 3= end, and in 2ases at both ends of the haplotype (Fig 3C). Theumber of likely recombination events indicatesrelatively ancient origin for this mutation, con-

istent with its detection in several differentestern European–derived populations.12-14,22,24

The mutation that is most interesting to ana-yze from an historical viewpoint is T36M, be-ause it is very common (about 17%12-14,22-24)nd has been suggested to be an ancestral12 or,lternatively, a recurrent change.22 The recurrentrigin is suggested because the change is a C3 Tt a CpG dinucleotide, known warm spots forutation in the human genome38, and lack of a

ommon haplotype with distant flanking mark-rs.22 Recently, an unpublished analysis withntragenic markers suggested that at least 18ifferent haplotypes are associated with T36M inuropean populations.21 These data appear atdds with ours in which 14 of 15 phase-known36M haplotypes carried a rare allele size

176 bp) at the closest marker, D6S1344. The 1ase in which this allele size could be excludedP803: Fig 3D and Table 2) is most easily ex-lained by a slippage mutation at the marker.alculations using the binomial distribution indi-ate the probability of observing 14 of 15 (phase-nown) T36M haplotypes associated with the76-bp, D6S1344 allele size, if they are dueolely to recurrent changes, is 5.1 �10�16. All

ases share a region of the common haplotype,

a5h(la(iT(

tepFstttgfiTmtomnaorddpc

Atyro

MHWs

m8

kTO

pOW

Ar

ro2

mc1

rc

d

sN

n2

Gi

tk

me2

ac

CONSUGAR ET AL86

lthough there is evidence of recombination of=AR in 3 cases and of various parts of the 3=aplotype in 11 cases (Fig 3D). In 1 familyP777), a recombination event of the T36M hap-otype has occurred within the family (Figs 2And 3D). In the one T36M homozygous caseP807), the haplotypes varied 3= to MBC-1,ndicating chance inheritance of 2 divergent36M haplotypes, rather than consanguinity

Fig 3D).It is possible to estimate the age of this muta-

ion by analyzing the number of recombinationvents with flanking markers and considering theossibility of marker mutation (see Methods andigs 1 and 3D for details). This approach isimilar to studies to age the CFTR �508 muta-ion by analysis of marker mutations,39,40 al-hough it does not rely directly on coalescentheory. From the data available, an origin of 49enerations ago was calculated with a 95% con-dence interval of 17 to 112 generations (Fig 4).his can be converted into a time of approxi-ately 1,225 years if we assume a generation

ime of 25 years. The nearly universal Europeanrigin of the 74 families described with thisutation and lack of this change in a limited

umber of studied families (33) of non-Europeanncestry suggests that the mutation probablyccurred once in northern Europe.12-14,22-24 In aecent study of 37 Turkish families, T36M wasetected just once (1.4% of alleles; unpublishedata). Therefore, it appears that it is the Europeanopulation that is most at risk for ARPKD be-

Fig 4. Log likelihood map shows the most likelyge of the T36M mutation (49 generations) and theonfidence limits (17 to 112 generations).

ause of this change, and routine prescreening of p

RPKD samples from this population by restric-ion digest12 may be productive. Haplotype anal-sis of further T36M cases may allow a moreefined dating and geographic localization of theriginal mutation.

ACKNOWLEDGMENTThe authors thank Eric Bergstralh for statistical help, Dragdalena Adeva for analysis of clinical notes, Markeckathorne and Jason Lilly for technical advice, and Drsilson, Thompson, Goodship, and McEwen for patient

amples.

REFERENCES1. Harris PC, Rossetti S: Molecular genetics of autoso-al recessive polycystic kidney disease. Mol Genet Metab

1:75-85, 20042. Guay-Woodford LM: Autosomal recessive polycystic

idney disease: Clinical and genetic profiles, in Watson ML,orres VE (eds): Polycystic Kidney Disease. New York, NY,xford University Press, 1996, pp 237-2663. MacRae Dell KM, Avner ED: Autosomal recessive

olycystic kidney disease, in GeneReviews; Genetic Diseasenline Reviews at GeneTests-GeneClinics (University ofashington, Seattle, 2003; www.genclinics.org)4. Kaplan BS, Fay J, Shah V, Dillon MJ, Barratt TM:

utosomal recessive polycystic kidney disease. Pediatr Neph-ol 3:43-49, 1989

5. Kaariainen H, Koskimies O, Norio R: Dominant andecessive polycystic kidney disease in children: Evaluationf clinical features and laboratory data. Pediatr Nephrol:296-302, 19886. Roy S, Dillon MJ, Trompeter RS, Barratt TM: Autoso-al recessive polycystic kidney disease: Long-term out-

ome of neonatal survivors. Pediatr Nephrol 11:302-306,9977. Capisonda R, Phan V, Traubuci J, et al: Autosomal

ecessive polycystic kidney disease: Outcomes from a single-enter experience. Pediatr Nephrol 18:119-126, 2003

8. Cole BR, Conley SB, Stapleton FB: Polycystic kidneyisease in the first year of life. J Pediatr 111:693-699, 19879. Guay-Woodford LM, Desmond RA: Autosomal reces-

ive polycystic kidney disease: The clinical experience inorth America. Pediatrics 111:1072-1080, 200310. Blyth H, Ockenden BG: Polycystic disease of kid-

eys and liver presenting in childhood. J Med Genet 8:257-84, 197111. Fonck C, Chauveau D, Gagnadoux MF, Pirson Y,

runfeld JP: Autosomal recessive polycystic kidney diseasen adulthood. Nephrol Dial Transplant 16:1648-1652, 2001

12. Rossetti S, Torra R, Coto E, et al: A complete muta-ion screen of PKHD1 in autosomal recessive polycysticidney pedigrees. Kidney Int 64:391-403, 200313. Ward CJ, Hogan MC, Rossetti S, et al: The geneutated in autosomal recessive polycystic kidney disease

ncodes a large, receptor-like protein. Nat Genet 30:1-11,00214. Onuchic LF, Furu L, Nagasawa Y, et al: PKHD1, the

olycystic kidney and hepatic disease 1 gene, encodes a

npr

g(1

s(m

ct

alb

se2

eki2

m(

ok8

ppN

msM

Npg

p

wC

ge

oan

fA

o5

oH

tt

ct

t(a

ma3

ad

l2

tBlM

op

h

HAPLOTYPE ANALYSIS OF ARPKD 87

ovel large protein containing multiple immunoglobulin-likelexin-transcription-factor domains and parallel beta-helix 1epeats. Am J Hum Genet 70:1305-1317, 2002

15. Zerres K, Mücher G, Bachner L, et al: Mapping of theene for autosomal recessive polycystic kidney diseaseARPKD) to chromosome 6p21-cen. Nat Genet 7:429-432,99416. Zerres K, Mücher G, Becker J, et al: Prenatal diagno-

is of autosomal recessive polycystic kidney diseaseARPKD): Molecular genetics, clinical experience, and fetalorphology. Am J Med Genet 76:137-144, 199817. Masyuk TV, Huang BQ, Ward CJ, et al: Defects in

holangiocyte fibrocystin expression and ciliary structure inhe PCK rat. Gastroenterol 125:1303-1310, 2003

18. Wang S, Luo Y, Wilson PD, Witman GB, Zhou J: Theutosomal recessive polycystic kidney disease protein isocalized to primary cilia, with concentration in the Basalody area. J Am Soc Nephrol 15:592-602, 200419. Ward CJ, Yuan D, Masyuk TV, et al: Cellular and

ubcellular localization of the ARPKD protein; fibrocystin isxpressed on primary cilia. Hum Mol Genet 12:2703-2710,00320. Zhang MZ, Mai W, Li C, et al: PKHD1 protein

ncoded by the gene for autosomal recessive polycysticidney disease associates with basal bodies and primary cilian renal epithelial cells. Proc Natl Acad Sci U S A 101:2311-316, 200421. Bergmann C, Senderek J, Küpper F, et al: PKHD1utations in autosomal recessive polycystic kidney disease

ARPKD). Hum Mutation 23:453-463, 200422. Bergmann C, Senderek J, Sedlacek B, et al: Spectrum

f mutations in the gene for autosomal recessive polycysticidney disease (ARPKD/PKHD1). J Am Soc Nephrol 14:76-9, 200323. Furu L, Onuchic LF, Gharavi AG, et al: Milder

resentation of recessive polycystic kidney disease requiresresence of amino acid substitution mutations. J Am Socephrol 14:2004-2014, 200324. Bergmann C, Senderek J, Schneider F, et al: PKHD1utations in families requesting prenatal diagnosis for auto-

omal recessive polycystic kidney disease (ARPKD). Humutat 23:487-495, 200425. Zerres K, Senderek J, Rudnik-Schoneborn S, et al:

ew options for prenatal diagnosis in autosomal recessiveolycystic kidney disease by mutation analysis of the PKHD1ene. Clin Genet 66:53-57, 200426. Müller D, Zimmering M, Roehr CC: Should nifedi-

ine be used to counter low blood sugar levels in children e

ith persistent hyperinsulinaemic hypoglycaemia? Arch Dishild 89:83-85, 200427. Dib C, Faure S, Fizames C, et al: A comprehensive

enetic map of the human genome based on 5,264 microsat-llites. Nature 380:152-154, 1996

28. Brownstein MJ, Carpten JD, Smith JR: Modulationf non-templated nucleotide addition by Taq DNA polymer-se: Primer modifications that facilitate genotyping. BioTech-iques 20:1004-1006, 1008-1010, 199629. Kaplan NL, Hill WG, Weir BS: Likelihood methods

or locating disease genes in nonequilibrium populations.m J Hum Genet 56:18-32, 199530. Hill WG, Weir BS: Maximum-likelihood estimation

f gene location by linkage disequilibrium. Am J Hum Genet4:705-714, 199431. Xiong M, Guo SW: Fine-scale genetic mapping based

n linkage disequilibrium: Theory and applications. Am Jum Genet 60:1513-1531, 199732. Excoffier L, Slatkin M: Maximum-likelihood estima-

ion of molecular haplotype frequencies in a diploid popula-ion. Mol Biol Evol 12:921-927, 1995

33. Alvarez V, Malaga S, Navarro M, et al: Analysis ofhromosome 6p in Spanish families with recessive polycys-ic kidney disease. Pediatr Nephrol 14:205-207, 2000

34. Mücher G, Becker J, Knapp M, et al: Fine mapping ofhe autosomal recessive polycystic kidney disease locusPKHD1) and the genes MUT, RDS, CSNK2b, and GSTA1t 6p21.2-p12. Genomics 48:40-45, 1998

35. Rigden DJ, Mello LV, Galperin MY: The PA14 do-ain, a conserved all-� domain in bacterial toxins, enzymes,

dhesins and signaling molecules. Trends Biochem Sci 29:35-339, 200436. Johnson CA, Gissen P, Sergi C: Molecular pathology

nd genetics of congenital hepatorenal fibrocystic syn-romes. J Med Genet 40:311-319, 200337. Xiao W, Oefner PJ: Denaturing high-performance

iquid chromatography: A review. Hum Mutat 17:439-474,00138. Cooper DN, Krawczak M, Antonorakis SE: The na-

ure and mechanisms of human gene mutation, in Scriver C,eaudet AL, Sly WS, Valle D (eds): Metabolic and Molecu-

ar Bases of Inherited Disease (ed 7). New York, NY,cGraw-Hill, 1995, pp 259-29139. Morral N, Bertranpetit J, Estivill X, et al: The origin

f the major cystic fibrosis mutation (D F508) in Europeanopulations. Nat Genet 7:169-175, 199440. Morral N, Nunes V, Casals T, et al: Microsatellite

aplotypes for cystic fibrosis: Mutation frameworks and

volutionary tracers. Hum Mol Genet 2:1015-1022, 1993