Cell, Vol. 87, 979–987, December 13, 1996, Copyright 1996 by Cell Press
The Molecular Basis of Focal Cyst Formation
in Human Autosomal Dominant
Polycystic Kidney Disease Type I
Feng Qian,* Terry J. Watnick,* Luiz F. Onuchic,
and Gregory G. Germino
Department of Medicine, Division of Nephrology
The Johns Hopkins University School of Medicine
720 Rutland Street
Baltimore, Maryland 21205
de Almeida et al., 1995). Although all types of ADPKD
present with an identical profile of extrarenal manifesta-
tions (including liver cysts and aneurysms), PKD1 is the
most severe, with a lower median survival and a higher
risk of progressing to end-stage renal disease (Ravine
et al., 1992).
account for interfamilial variation, they cannot explain
why related individuals who presumably share a com-
mon mutation have different clinical presentations. Se-
vere childhood cases of ADPKD born into families that
exhibit the classic adult presentation of the disease are
dramatic examples of this phenomenon(Blyth and Ock-
enden, 1971; Kaariainen, 1987; Fick et al., 1993; Zerres
et al., 1993). Fick et al. (1994) argued that these families
are examples of anticipation and proposed that an un-
stable trinucleotide repeat might be responsible for
Since these initial clinical observations were made,
the genes for PKD1 and PKD2 have been identified.
PKD1 encodes a 14 kb mRNA that is derived from 46
exons that extend over ?50 kb of genomic DNA (The
European Polycystic Kidney Disease Consortium, 1994;
The American PKD1 Consortium, 1995; Hughes et al.,
1995; The International Polycystic Kidney Disease Con-
sortium, 1995). An unusual feature of the gene is that
?70% of its length is replicated in at least three copies
clustered together on 16p13.1. The other loci, which
also are transcribed, have a sequence nearly identical
to that of PKD1 in their shared segments. The PKD1
gene product, polycystin, is 4302 amino acids in length
and is likely to be an integral membrane glycoprotein
that regulates cell–cell or cell–matrix interactions. PKD2
encodes a 5.4 kb mRNA within ?68 kb of genomic DNA
(Mochizuki et al., 1996). The predicted protein of 968
amino acids has significant homology to the voltage-
activated Ca2?channel ?1E as well as to a portion of
polycystin. It isthought toencode anintegral membrane
protein that may function as an ion channel or pore
whose activity is regulated by polycystin. Neither gene
has any trinucleotide repeats within its mRNA that are
>5 unitsin length, norare anyfound within thecomplete
gene sequence of PKD1.
Mutation analysis of PKD1 has been greatly hindered
by high sequence similarity to its replicated loci. None-
theless, the mutations reported to date for both PKD1
and PKD2 have been stable nucleotide substitutions,
deletions, or insertions and do not explain why family
members with the same germline mutation exhibit dra-
matic phenotypic variability (Peral et al., 1995, 1996a,
1996b;Mochizuki et al.,1996).In fact, Peraletal. (1996b)
recently reported a set of fraternal twins with the same
germline nonsense mutation in PKD1 yet with remark-
ably different phenotypes.One child had a severe infan-
tile-onset form of the disease, whereas the other had
no cysts at the age of 5 years. The authors suggested
that the difference in clinical presentation was most
likely due to the effect of a small number of genetic
Although breedingexperiments inmicewith recessive
is a common disease and an important cause of renal
failure. Itis characterizedby considerable intrafamilial
phenotypic variation and focal cyst formation. To elu-
cidate the molecular basis for these observations, we
have developedanovelmethodfor isolating renal cys-
tic epithelia fromsingle cysts andhave used it to show
that individual renal cysts in ADPKD are monoclonal.
Loss ofheterozygositywas discoveredwithin asubset
of cysts for two closely linked polymorphic markers
located within the PKD1 gene. Genetic analysis re-
vealed that it was the normal haplotype that was lost.
This study provides a molecular explanation for the
focal nature of cyst formation and a probable mecha-
nism whereby mutations cause disease. The high rate
at which “second hits” must occur to account for the
large number of cysts observed suggests that unique
structuralfeatures of thePKD1 genemaybe responsi-
ble for its mutability.
Autosomal dominant polycystic kidney disease (ADPKD)
is one of the most common inherited diseases of hu-
mans; itisestimated toaffect 1 in1000 ofthe population
(Gabow, 1993). Renal cysts are the major clinical fea-
tures of the disease and appear to increase in size and
number throughout the lifetime of an individual. This
process results in renal failure in approximately half of
affected individuals by age 50 years. Although the renal
lesion is the most prominent feature, ADPKD is a sys-
temic disorder with a variety of other manifestations
including liver cysts, cerebral aneurysms, and a variety
of cardiac valvular abnormalities.
ADPKD exhibits considerable variability with respect
to both its renal and extrarenal manifestations. Genetic
heterogeneity is likely to account for at least some of
thesedifferences.Linkage studieshave determinedthat
there are at least three forms of ADPKD. PKD1, which is
most common and accounts for 85%–95% of all cases,
maps to chromosome 16p13.3 (Reeders et al., 1985).
The second type, PKD2, which affects most of the re-
maining families, maps to chromosome 4q13–23 (Kim-
berling et al., 1993; Peters et al., 1993). A small number
of families have another form, which has not yet been
mapped (Bogdanova et al., 1995; Daoust et al., 1995;
*These authors contributed equally to this work and are listed in
Figure 1. Clonality Assay for Single Cysts
Derived from Kidney Donors JHU93 and
(A) Cystic epithelia were isolated from 11 in-
tact cysts (1–11) from donor JHU93. Malege-
nomic DNA (?) was added to an aliquot of
each cyst-derived DNA sample as well as
DNA isolated from the donor’s nucleated
blood cells. One half of each mix was di-
gested with HpaII and then used as template
for PCR using theandrogen receptor primers
AR1 and AR2. The amplification products are
presented in lanes labeled ?.The other undi-
gested halfof eachmix was usedas template
for a control amplification done inparallel us-
ing identical reaction conditions. The results
are presented in lanes labeled ?. The sym-
bols to the left of the figure identify the posi-
tion of the AR alleles of the donor (?) and
male control. Three bands are present in the
? lanes of all samples except for that which
had only male DNA as template (?) and cyst
9, which yielded no products. All samples derived from cysts yielded a single band after digestion if the reaction was complete (cysts 2, 4,
5, 6, 8, and 11). In contrast, DNA from blood of the donor yielded two bands despite complete digestion with HpaII prior to PCR. The clonal
status of samples 1, 3, and 7 were deemed uninterpretable because of incomplete digestion.
(B) Sixteen intact cysts (C1–16) from donor JHU273 were analyzed using a protocol similar to that described above. Each mix of male- and
cyst-derived DNAwas digested with HhaI ratherthan HpaII prior to amplification usingAR1 and AR2.Both the finalrinse (w)and fluidcontaining
the cyst-lining epithelial cells (c) were evaluated for each sample. For C1 and C2, the original cyst fluid (f) was similarly examined. A dilution
series of DNA prepared from nucleated blood cells of JHU273 was analyzed using an identical procedure (right). These data suggest that a
minimum of 2.5 ng of cyst DNA was used as template for each amplification. Assuming that the average cell contains ?10 pg of DNA and
one tenth of the total yield of cyst DNA was used for each reaction, each sample was estimated to contain between 2500 and 500,000 cells.
cystic diseasesupporta rolefor theeffect ofotherloci in
disease expression, genetic backgroundprobably does
not explainall aspectsof phenotypicvariability inhuman
ADPKD (Iakoubova et al., 1995). Even within a single
kidney, fewer than 1% of nephrons actually develop
cysts, suggesting that cystogenesis is a focal process.
Histopathologic studies of ADPKD kidneys have con-
firmed that cyst formation begins with the localized out-
growth of a tubule in any nephron segment, including
segments with different embryologic origins (Evan and
McAteer, 1992;Scha ¨fer etal.,1994).These observations
suggest that cyst formation is a two-step process and
that an inherited mutation at one of the ADPKD gene
loci is necessary but insufficient for cystogenesis.
We have hypothesized that the focal nature of cyst
formation in ADPKD probably holds clues to an under-
standing of the pathogenesis of this disorder. To study
this process, we have developed a novel method for
isolatingepithelial cellsfromsingle renal cyststhat mini-
mizes contamination by other cells and have used it to
show that renal cysts in ADPKD are monoclonal. We
have demonstrated loss of heterozygosity (LOH) within
individual cystsfortwocloselylinkedpolymorphic mark-
ers located within the PKD1 gene. Genetic analysis has
confirmed that it is the normal haplotype that is lost in
somatic tissues. This study provides a molecular expla-
nation for thefocalnatureof cyst formationand a proba-
ble mechanism whereby mutations cause disease.
al., 1987).Recently, a polymerasechain reaction (PCR)–
based clonality assay has been developed that can be
used for small quantities of template (Gilliland et al.,
1991). This method exploits methylation-sensitive re-
striction sites (HpaII and HhaI) close to the highly poly-
gene. The methylation status of these sites correlates
with X-inactivation (Allen et al., 1992). Willman et al.
(1994) used this approach to show the monoclonal na-
tureofhistiocystosis X.We havemodified thistechnique
to determine the clonality of single cysts in ADPKD
Initial studies using DNA prepared from large speci-
mens (>5 g) containing multiple cysts failed to detect
monoclonality (data not shown). We then attempted to
dissection. Again, most samples were polyclonal, but
histologic analysis revealed that the microdissected
tiple nonepithelial cellular elements (data not shown).
To circumvent these difficulties we adapted a method
previouslyused to study toad bladderepithelia (Handler
et al., 1979) and applied it to the isolation of cyst-lining
cells. In this method, the cyst remains intact during the
entire procedure. EDTA injected into the cyst lumen is
basement membrane, which serves as a natural barrier
to contaminating cells.
Results of the clonality assay of cysts isolated from
two individuals with ADPKD are presented in Figure 1
prepared from individual cysts of donor JHU93 (Figure
1A), two failed to yield sufficient product for analysis
(samples 9 and 10). The results of three other samples
Renal Cysts in ADPKD Are Monoclonal
X-chromosome inactivation has been commonly used
to demonstrate the clonal nature of tumors (Fearon et
Molecular Basis of Focal Cyst Formation in PKD1
were deemed uninterpretable because of incomplete
digestion (samples 1, 3, and 7). The remainder (samples
2, 4, 5, 6, 8, and 11) yielded a single product after diges-
tion.Samplespreparedfrom16individual cystsof donor
JHU273 were similarly analyzed (Figure 1B). The clonal
status of three samples could not be determined be-
cause their final rinse solutions also yielded PCR prod-
ucts (C3 and C15) or because no PCR product was
detected (C5). Of the remaining 13 cysts, 11 (85%)
yielded a single PCR product after HhaI digestion. Only
2 of the13cysts (C6 and C14)yielded two products,and
in each instance the bands were of unequal intensity,
suggesting a clonal bias.
A total of 76 renal cysts derived from 8 affected fe-
males were analyzed for clonality, and 62 cysts (82%)
were found to be monoclonal (Table 1). It is very likely
that theactualpercentageof monoclonalcysts is>82%.
In some samples, rinse solutions were not tested for
luminalcontamination (Table1). Lateranalyses revealed
that the rinse solution of approximately one half of the
apparently polyclonal cysts yielded PCR products, sug-
gesting probable contamination (Figure 1). The small
number of cysts that are truly polyclonal may have re-
sulted from the fusion of two or more formerly neigh-
boring monoclonal cysts.
Figure 2. Position of Intragenic PKD1 Polymorphisms Used to
The PKD1 gene contains 46 exons and is bisected by a polypyrimi-
dine tract of ?2.5 kb (hatched box). The replicated portion of the
gene begins with the first exon and ends in intron 34 (stippled bar).
The microsatellite KG8 lies within the 3?UTR of the gene. EJ1 is a
polymorphic locus located in exons 45–46. The primers used to
generate these markers and their relative positions are indicated.
The PCR product that contains KG8 and EJ1 is approximately 1.6
kb in length and is generated using FQF28 and KG8R8.
band for KG8. Figure 3A is a representative example of
data using these markers. Two cysts from this kidney
(C8 and C14) were found to have LOH. A total of 46
cysts from four donors were evaluated with KG8, and
17% werefoundtobehemizygousfor thismarker (Table
1). At least two cyst preparations per donor had loss of
a KG8 allele, and in each case it was the same allele.
Allelic Loss of a PKD1 Intragenic Microsatellite, KG8
It is possible that cysts may develop from a cluster
of cells sharing a common X-chromosome inactivation
status rather than from clonal expansion of a single
progenitor cell.Toprove that renal cysts aretruly mono-
we tested for LOH for PKD1 in individual cysts using
the microsatelliteKG8 (Snarey et al., 1994). This marker
lieswithin the3?untranslatedregion(3?UTR) ofthePKD1
mRNA (Figure 2). Two alleles were equally amplified in
most samples. In a subset, however, only a single band
was amplified. We included primers for the androgen
receptor as an internal control. Two bands correspond-
ing to theandrogen receptor alleles were equally ampli-
fied in all samples, including those that had a single
Loss of a Second Closely Linked Marker, EJ1
We sought to confirm the KG8 results using a second
marker that was either intragenic or immediately proxi-
mal to PKD1, but the closest known highly polymorphic
marker was at least 70 kb away. In the course of per-
forming mutation analyses, we discovered a polymor-
phic locus, EJ1, in exons 45–46 that is located approxi-
mately 1.6 kb proximal to KG8 (Figure 2). It has at least
three allelic variants (A1–A3) that can be detected using
heteroduplex analysis. We used this marker to test for
Table 1. Summary of Clonality Assay and LOH for Single Renal Cysts from ADPKD Patients
Number of Cysts
JHU246 13 1292
aNumber of cysts from which androgen receptor-specific PCR products could be amplified after HhaI or HpaII digestion.
bSamples for which the final rinses were tested for contamination.
ND, not determined; NI, not informative, nt, length in nucleotides.
Figure 4. Haplotype Analysis of Donor JHU273
A 1.6 kb PCR fragment containing KG8 and EJ1 was amplifed from
blood and then cloned (Figure 2). Individual clones were assayed
fortheirrespectiveKG8 (A)andEJ1 alleles(B).Cloneswereassigned
an EJ1 allele based on mixing studies with unlabeled A1 (B) and A2
(data not shown). All clones contained either KG8 allele 110 bp and
EJ1 allele A1 (clones 1 and3) or KG8 allele 106 bp and EJ1 A2 allele
(clones 2, 4, and 5). nt, length in nucleotides.
that cyst samples isolated from the same donor had
loss of an identical allele (Figure 3C). A total of 30 cysts
from three donors were evaluated with this marker, and
7 were found to have either a deletion or novel somatic
mutation involving this locus (Table 1).
Loss of the Unaffected Haplotype
The concordance between the KG8 and EJ1 data
strongly suggested that the lost alleles belong to the
same haplotype. We confirmed this hypothesis using a
method that exploited thephysical proximity of the loci.
Agenomic fragment containingEJ1and KG8was ampli-
fied from the peripheral blood of each donor showing
LOH for both markers using primers FQF28 and KG8R8
(Figure 2). The products were cloned and the haplotype
of each was determined. A representative example of
this analysis for JHU273 is shown in Figure 4. For each
donor, analysis of cloned genomic segments confirmed
that the lost KG8 and EJ1 alleles indeed belonged to
the same haplotye (Table 1). No haplotype common to
all donors was found to be lost in the samples.
We next sought to determine whether it was the nor-
mal or affected haplotype that was lost in cystic tissue.
Linkage analysis using KG8 and 3?HVR was performed
on theonly twofamilies available for testing. 3?HVR was
linked to the disease in these families, with maximum
lod scores of1.34 (pedigree1,Figure 5A) and 2.18(pedi-
gree 2, Figure 5B) at a recombination fraction of 0.0. A
positive lod score (1.93) was also obtained with KG8 in
pedigree 2. While linkage of ADPKD to KG8 was only
weakly positive in pedigree 1 (maximum lod score of
0.32), two-point analysis of 3?HVR and KG8 favored the
haplotype shown in Figure 5A (maximum lod score of
0.99 at a recombination fraction of 0.0). These results
strongly sugest that it was the chromosome 16 haplo-
type not associated with disease that was lost in cystic
tissue (Figure 5).
Figure 3. Loss of Heterozygosity for PKD1 Markers in Cysts from
Kidney Donor JHU273
(A) DNA isolated from cyst samples was used as template for PCR
using KG8 andtheandrogenreceptor primersAR1 andAR2 (without
prior HhaI digestion). Sample numbers in this figure correspond to
those usedin Figure1.The numbers ontheright indicate thelengths
of the KG8 and androgen receptor alleles. In two cysts (C8 and
C14), only one KG8 allele, 110 bp long, was amplified (top), whereas
two bands were equally amplified by AR1 and AR2 (bottom). nt,
length in nucleotides.
(B) Heteroduplexanalysis ofEJ1 amplified fromtheblood ofJHU273
(lane 1) shows that the patient is informative at this locus. Sample
numbers in this figure correspond to those used in Figure 1. The
majority of renal cysts assayed have the same pattern as lane 1.
C8 and C14, however, demonstrate LOH at this locus since hetero-
duplex formation is no longer present. C12 shows a novel hetero-
(C) Mixing studies were performed to determine which EJ1 allele
remained incyst samples withapparent LOH(C8 andC14)or anovel
heteroduplex pattern (C12). Samples with patternsrepresentative of
blood (C2 andC15) wereincluded as controls.Unlabeled EJ1 alleles
A1, A2,and A3weremixed separately withradiolabeled EJ1that had
been amplified from each cyst and then analyzed for heteroduplex
formation. The addition of A1 did not alter the preexisting hetero-
duplex patterns of any of the cysts, whereas the addition of A3
resulted in a novel pattern in cysts without LOH. Although mixing
with A2 did not change the patterns for C2 and C15, it restored the
control heteroduplex pattern in C8 and C14 and yielded a unique
band in C12. These results suggest that A1 is present in all cysts,
whereas A2 has been lost from C8 and C14 and replacedby a novel
allele in C12. A3 is missing from all cysts because this allele is not
present in JHU273.
A Novel Somatic Mutation
One cyst sample was discovered to have discordant
resultswith KG8andEJ1.This sample(C12)didnothave
loss of a KG8 allele but did have a unique heteroduplex
pattern with EJ1 (Figure 3). Mixing studies determined
that the A2 allele of EJ1, which was lost in the other two
LOH in cysts from donors known to be polymorphic at
this locus. There was complete concordance between
the studies using KG8 and EJ1, with one exception (dis-
cussed below) (Figure 3B). All samples that lacked a
KG8 allele also lacked one for EJ1. Mixing studies (see
Experimental Procedures) were used to demonstrate
Molecular Basis of Focal Cyst Formation in PKD1
Figure 5. Linkage Analysis of Two Polymorphic Markers Linked to
PKD1 in the Families of Donors JHU273 (A) and JHU246 (B)
PKD1 is linked to 3?HVR and KG8 with maximum lod scores of 1.34
and 0.32, respectively, at a recombination fraction of 0.0 for both
markers in pedigree 1 (A). The posterior probability of linkage of the
disease inthis family to PKD1 is>95%. Two-point analysis of 3?HVR
and KG8 favors the haplotype shown in (A). In pedigree 2 (B), the
maximum lod scores for 3?HVR and KG8 are 2.18 and 1.93, respec-
tively, at a recombination fraction of 0.0. The genotypes D,110 and
A,110 segregate with ADPKD in pedigree 1 (A) and pedigree 2 (B),
respectively. In each kidney donor (arrows), it is the KG8 allele that
doesnot segregatewithdiseasewhichis lostincystsdemonstrating
LOH. Affected genotypes are indicated by closed bars; deduced
genotypes are shown in brackets. The pedigrees as shown do not
includeall family members.Trianglesare usedto protecttheidentity
Figure 6. A 2 bp Deletion in C12 (?C12) Creates a Novel MwoI Site
(A) EJ1 was amplified from C12, and clones of two alleles (A1 and
A4) were identified. The sequence of the novel allele, A4, is shown
along with the corresponding sequence of A1. A4 contains a 2 bp
deletion (underlined inA1) that creates a new MwoIsite (arrow).The
deletion was confirmed byanalysis of twoclones sequencedinboth
(B) (Top) The MwoI restriction map of EJ1 A1 (M). An internal PCR
product of 172 bp was used to test for the presence of the novel
MwoI site (M*). (Bottom) Comparison of the MwoI maps of this
subfragment of A1 and A4. Cleavage at the new site is predicted
to result in two fragments of nearly identical length in A4.
(C) To confirm that the DC12 deletion was a somatic mutation, the
172 bp subfragment was amplified from the following templates:
DNA isolated from peripheral blood of JHU273 (lanes 1 and 2) and
cloned alleles A1 (lanes 3 and 4) and A4 (lanes 5 and 6) as well as
from the original EJ1 PCR product (lanes 7 and 8). Undigested
products (lanes 1, 3, 5, and 7) and products digested with MwoI
(lanes 2, 4, 6, and 8) were analyzed on a 4% Nusieve gel. As pre-
dicted, two indistinguishable bands of 80 and 81 bp are present in
lanes 6 and 8 and absent in lanes 2 and 4. A 161 bp fragment is
also observed in the original EJ1 PCR product (lane 8). The 11 bp
restriction fragment common to all could not be detected. The re-
sults were confirmed using an independently derived C12 EJ1 PCR
product (data not shown).
cysts (C8 and C14) from this donor, had been replaced
by a novel allele (A4) in C12. We cloned both the A1
and A4 alleles of the EJ1 locus from this sample and
determined the sequence of the novel allele (Figure 6A).
Comparison ofthesequence ofA4to thatof A1revealed
two differences. The first is a 2 bp deletion (?C12) at
positions 12,694–12,695 of the cDNA sequence (HUMP-
KD1A, GenBank accession number L33243). The ?C12
deletion is predicted to cause a reading frameshift re-
sultingin a truncatedprotein. This mutationalso creates
a new MwoI endonuclease restriction site not present
in the sequence of either the A1 or A2 alleles of EJ1
(Figure 6B). This site was used to confirm the presence
of the deletion in the original C12 DNA sample (Figure
6C). The donor’s blood DNA lacks the restriction site,
proving that the mutation was acquired and not present
in the donor’s germline. The second difference in the
A4 sequence is a C-to-T transition at position 12,617,
which hasno effect on the protein sequence. This is the
same sequence variant associated with the A2 allele,
suggesting that the ?C12 deletion occurred on this
examined. We cannot determine whether the two cysts
with LOH for KG8 had identical mutations or were inde-
pendent events since they had identical X-chromosome
inactivation patterns. Cysts from other kidneys, how-
ever, that had a similarpattern of LOHfor PKD1 markers
had different X chromosomes inactivated, suggesting
that independent events had resulted in LOH (data not
shown). Formal proof of our hypothesis must await the
development of methods that can be used for mutation
analysis of the full length of the PKD1 gene.
The very high rate of somatic mutation predicted by
our model is surprising since no known dynamic ele-
ments have been identified within the genomic se-
quence of PKD1 and since the adult kidney is thought
tohave a relatively low mitotic index. However, we have
previously reported an extremely unusual 2.5 kb poly-
pyrimidine tractwithin intron 21that maybe responsible
for the gene’s increased rate of mutation (The American
PKD1 Consortium, 1995). Similar but much shorter ele-
ments present within other genes have been shown to
undergo triple-helix formation bothin vitro (Young et al.,
1991) and in vivo (Rao et al., 1988). Wang et al. (1996)
recently have shown that triplex formation induces
mutagenesis in a mammalian cell culture system and
demonstrated a requirement for excision and transcrip-
tion-coupled repair in this process. The authors hypoth-
esized that formation of the triple helix causes a stall
in transcription that triggers gratuitous and potentially
error-prone repair. They proposed that naturally oc-
curring triple helices may similarly trigger repair and
mutagenesis and thus constitute endogenous sources
of genetic instability.
We postulate that the polypyrimidine tract within
PKD1 may cause ongoing errors in its transcription-
coupled repair that result ina high frequency of somatic
mutation. This model can explain the multiplicity of sec-
ond hits as well as the apparent development of new
cysts throughout the lifetime of an individual. Likewise,
the genetic instability possibly associated with this un-
usualgenomic structuremay beresponsiblefor thehigh
incidence of PKD1 within the population. If this element
is proven to be responsible for the gene’s mutability,
PKD1 will be the first example of a disease that results
from this novel mechanism of somatic mutation.
The indistinguishable clinical presentation of patients
with PKD2 suggests a two-step process in this disease
as well. Whereas the initial step is certainly a germline
mutation of PKD2, the nature of the second event is not
yet known. Presently it cannot be excluded that there
may be an uncharacterized unstable element hiding
within the genomic structure of PKD2 that results in
frequent somatic inactivation in a manner analogous to
that of PKD1. The relative infrequency of PKD2 argues
against this hypothesis, however. The phenotypic simi-
larity of the disorders may offer an alternative hypothe-
sis. Investigators have postulated that the gene prod-
ucts of PKD1 and PKD2 may be interacting partners of
a common pathway. It has been suggested that the
function of PKD1may betoregulate theactivityof PKD2
(Mochizuki et al., 1996). If this is correct, somatic inacti-
vation of PKD1 may be the second step that leads to
clonal expansion in PKD2 and possibly other forms of
ADPKD. This model predicts that the frequency with
A notable but unexplained clinical feature of all forms of
ADPKD is the phenotypic variability exhibited by family
members who share a common mutation. The focal na-
ture of renal cysts isan extreme example of this variabil-
ity since all renal tubular epithelial cells within a kidney
have an identical mutation. These features suggest that
other factors are required for cyst formation. Identifica-
tion of these other components may explain the patho-
genesis of the disease and possibly provide new ave-
nues for developing therapies.
To evaluate the molecular mechanisms responsible
for cystogenesis, we developed a novel method for iso-
lating epithelial cells from individual cysts. This tech-
avoids biases resulting from the use of cultured cells.
We have used this approach toanalyze multiple kidneys
from females and have shown, using two independent
methods, that renal cysts in ADPKD are clonal in origin.
Our datasuggest thatsomaticmutationof thepreviously
normal PKD1 allele is the second, rate-limiting step in
cystogenesisand islikely toaccount for thefocalforma-
tion of renal cysts. The rate at which one acquires these
“second hits”probably plays a major role indetermining
the course of disease and probably explains much of
its phenotypic variability. It may be speculated that a
similar two-hit mechanism may also be responsible for
causing the extrarenal manifestations of PKD1.
Loss of the normal allele of PKD1 in cystic tissue
suggests a molecular recessive mechanism of disease
similar to that seen with numerous tumor suppressor
genes. If our model is correct, however, the frequency
of “second hits” in cystic epithelia must be extremely
high to account for the thousands of cysts that are ob-
served. This isin contrast tothe small numberof tumors
found inother inherited disordersof the kidney in which
somatic mutation of the normal allele is required for
disease. Wilms’ tumor (Huff et al., 1991) and von Hippel
Lindau’s disease (Foster et al., 1994) are representative
Our data suggest that there is indeed a high rate of
somatic mutation in PKD1. This is best illustrated by
sample JHU273 (Table 1). Of the 13 cysts that were
shown to be monoclonal, 2 had loss of alleles for KG8
and EJ1; 1 had a unique 2 bp deletion; and 8 had no
detectable mutation of the normal allele. The last group
didnot havea common X-chromosomeinactivationpat-
tern, suggesting that they had arisen from at least two
independent progenitor cells (Figure 1). Assuming that
thesomaticmutationsarose afterX-chromosome inacti-
vation was completed, a minimum of four independent
mutations must have occurred in the 13 cysts that were
Molecular Basis of Focal Cyst Formation in PKD1
was added to each cyst DNA sample prior to PCR to control for the
completeness of HpaII or HhaI digestion. In most cyst samples,
the male-specific product was not amplified after HpaII or HhaI
digestion, as expected.
which cysts form in PKD1, PKD2, and PKD3 will be
determined by the rate of somatic mutation of PKD1.
Theidenticalnumberof renalcysts observedinthethree
disorders is consistent with this hypothesis.
In this study, we have determined that random inacti-
vation of the normal PKD1 allele in somatic tissue is the
likely molecular explanation for the observed clinical
variability and the focal formation of renal cysts in the
most common form of ADPKD. Both the high incidence
of the disease in the population and the large number
of second hits that must occur to account for the num-
ber of cysts observed suggest that unique features of
the PKD1 gene structure may be responsible for its
mutability. Our data suggest a molecular recessive
mechanism of disease since both alleles are mutated in
renal cysts. These findings have important implications
for investigators interested in developing models sys-
tems and suggest that treatment strategies directed at
replacement of polycystin may prevent cyst formation
and end-stage renal disease.
Loss of Heterozogosity Assay Using KG8
The microsatellite KG8 that is present in the 3?UTR of PKD1 mRNA
was used to distinguish the mutant and normal alleles (Snarey et
al., 1994). Cystic DNA samples served as template for PCR amplifi-
cation. PCR amplification was performed for 28 cycles using an
identical protocol as that used for the androgen receptor except
that thereaction included a final concentration of 1 ?M of eachKG8
primer (KG8F8, 5?-CTCCCAGGGTGGAGGAAGGTG-3? [bp 13,925–
13,945; HUMPKD1A, GenBank accession number L33243] and
KG8R8, 5?-GCAGGCACAGCCAGCTCCGAG-3? [bp 14,014–14,034]),
5 nM32P–end-labeled KG8F8, and 1.5mM MgCl2. The PCR products
werediluted with 5?volume ofsequencing loadingbuffer andsepa-
rated in a 6% denaturing polyacrylamide gel. The dried gel was
examined by autoradiography using X-Omat XAR film (Kodak) at
?80?C overnight with an intensifying screen.
Either 200 ng of genomic DNA (isolated from whole blood using
the Puregene kit) or 5 ?l of cyst DNA was used as template for
amplification of a 540 bp product (EJ1) using primers FQF28
FQR35 (5?-ATGGGCCACGGGAAGATCC-3?, 12,977–12,995). PCR
was performed as follows: denaturation at 94?C for 5 min; 35 cycles
of 94?C for 30 sec, 62?C for 30 sec, and 72?C for 30 sec; and a final
extension of 72?C for 10 min. The total PCR volume was 30 ?l using
2 unitsofTaq DNApolymerase(Boehringer Mannheim), 0.2?ldCTP,
and a final MgCl2concentration of 1.5 mM.
Heteroduplex analysis was performed using Hydrolink Mutation
Detection Enhancement gels (MDE, AT Biochem) following the
manufacturer’s protocol. Urea was added to the gel to a final con-
centration of 15% to minimize band broadening. The radiolabeled
PCR products were initially denatured by heating at 95?C for 5 min-
utes and then allowed to cool to room temperature gradually over
1–2 hr before loading. Gels were run at 700 V for 14–16 hr, dried,
and placed on X-Omat XAR film (Kodak) at room temperature or on
a Phosphoimager cassette (Molecular Dynamics).
Preparation of Cystic Epithelial Cells from a Single Cyst
Cystic kidneys were processed within 24 hours of removal from the
patient and were maintained at 4?C. The surface of the cyst was
first rinsed with PBS and then its contents were drained by needle
and syringe.The needle was left inserted in the cyst for theduration
of the washing and incubation steps. The cavity of intact cysts was
rinsed a minimum of three times with Ca2?-and Mg2?-free PBS. The
last rinse from some cysts was collected and stored on ice. PBS
containing2 mMEDTA wasthen injected intothelumen of theintact
cyst.After a20min incubation,thecystwas massagedseveral times
to assist in detachment of the epithelial cells from the basement
membrane. Cysts that maintained the extraction solution (PBS/
EDTA) in their lumina for the duration of the incubation period were
considered intact. Only the epithelial cells of intact cysts were har-
vested by drainage. The cystic epithelial cellsin the extraction solu-
tion (PBS/EDTA) and the last wash solutions were centrifuged at
1500 rpm (Beckman)for 15 min.The pellet was usedfor DNA prepa-
ration using the Puregene DNA extraction kit (Gentra) according to
the manufacturer’s protocol. To assist in DNA precipitation, 10 ?g
of glycogen was added to each sample.
The polymorphic locus EJ1 was identified by heteroduplex analysis
using the primers FQF28–FQR35 as described above. This PCR
product spans part of exons 45 and 46 as well as the 90 bp intron
between them and contains the coiled-coil domain of polycystin.
Three heteroduplex patterns weredetected ina screenof 20 normal
and 45 affected individuals. The PCR products that gave each pat-
tern were cloned into pCRII (Invitrogen) and then sequenced to
determine the molecular basis for the polymorphism. A1 was found
to contain the published sequence (HUMPKD1A, GenBank acces-
sion number L33243), whereas A2 had a conservative base pair
change (C-to-T) at position 12,617, which did not alter the amino
acid sequence (leucine, amino acid4136). A3 hadeight consecutive
guanines inintron 45insteadoftheseven contained inthepublished
genomic sequence (51,325–51,331; HUMPKD1GEN, GenBank ac-
cession number L39891). All three alleles were used for mixing
EJ1 was amplified from clones of each of the alleles using the
PCR conditions described above. Four microliters of radiolabeled
EJ1 amplifed using cyst DNA as template was mixed in separate
tubes with 4 ?lof coldproduct derivedfrom eachof thethree cloned
alleles. Heteroduplex analysis was performed as described above.
The products obtained using sample C12 as template were cloned
intopCRII andassayedusing A1–A2inmixing studies.Twoindepen-
dent clones containing the mutant allele were sequenced using
an automated sequencer (Applied Biosystems, Inc.) and standard
The 2 bp deletion identified in the C12 mutant clones created a
new MwoI site. To test for the presence of this site, a nested 172
bp PCR product was amplified using primers C12F1 (5?-CTC
The DNA samples were digested with 10 units of HpaII or HhaI in
a volume of 10 ?l overnight at 37?C. The enzyme was inactivated
by heating at 95?C for 10 min and then used directly as template
for PCR amplification. A set of control amplifications without prior
restriction digestion was done in parallel using identical reaction
conditions for samples from donors JHU93, JHU188, JHU244, and
JHU304. In all cases, two alleles were amplified from each sample.
PCR was performed for 28 cycles of 94?C for 20 sec, 65?C for 20
sec, and 72?C for 20 sec after an initial denaturation at 94? for 5
minutes. The PCR mixture (40 ?l) contained a final concentration of
10 mM of each androgen receptor primer (AR1 and AR2), 5 nM
ratories), and 2 mM MgCl2. The sequences of the primers were
obtained from sequences reported by Tilley et al. (1989): AR1,
5?-GCTGTGAAGGTTGCTGTTCCTCAT-3? (bp 485–508; HUMARA,
GenBank accession numberM21748), andAR2, 5?-TCCAGAATCTG
TTCCAGAGCGTGC-3?(bp 230–253). The PCRproducts werediluted
with 5? volume of sequencing loading buffer (95% formamide, 10
mMEDTA[pH8.0], 0.1%bromophenol blue,and0.1%xylenecyanol
FF)and separated in a 6% denaturing polyacrylamide gel.The dried
gel was examined by autoradiography using X-Omat XAR film (Ko-
dak) at ?80?C overnight with an intensifying screen. An aliquot of
DNA from a male with a different number of CAG repeats at the
androgen receptor locus (so that the alleles could be distinguished)
TGCCCAGGGTGCAGC-3?, genomic position 51,350–51,367; HUM-
PKD1GEN) and C12R1 (5?-GAGGTGGAGGGGTGCGAG-3?, cDNA
position 12,760–12,777; HUMPKD1A) and the following templates:
peripheral blood leukocyte DNA of the same donor, plasmid DNA
containing the EJ1 alleles cloned from C12, and the original EJ1
products obtained from C12 in two independent reactions. PCR
products were digested with MwoI (New England Bio Labs) ac-
cording to the manufacturer’s specfications and analyzed on a 4%
Nusieve gel (FMC Bioproducts).
polymorphicCAG repeatinthehuman androgen-receptorgene cor-
relates with X chromosome inactivation. Am. J. Hum. Genet. 51,
The American PKD1 Consortium (1995). Analysis of the genomic
sequence for the autosomal dominant polycystic kidney disease
(PKD1) gene predicts the presence of a leucine-rich repeat. Hum.
Mol. Gen. 4, 575–582.
Blyth, H., and Ockenden, B.G. (1971). Polycystic disease of kidney
and liver presenting in childhood. J. Med. Genet. 8, 257–284.
Bogdanova, N., Dworniczak, B., Dragova, D., Todorov, V., Dimitra-
kov, D., Kalinov, K., Hallmayer, J., Horst, J., and Kalaydjieva, L.
(1995). Genetic heterogeneity of polycystic kidney disease in Bul-
garia. Hum. Genet. 95, 645–650.
Daoust, M.C., Reynolds, D.M., Bichet, D.G., and Somlo, S. (1995).
kidney disease. Genomics 25, 733–736.
de Almeida, S., de Almeida, E., Peters, D., Pinto, J.R., Tavora, I.,
Lavinha, J., Breuning, M., and Prata, M.M. (1995). Autosomal domi-
nant polycystic kidneydisease-evidence for the existence of a third
locus in a Portuguese family. Hum. Genet. 96, 83–88.
The European Polycystic Kidney Disease Consortium (1994). The
polycystic kidney disease gene encodes a 14 kb transcript and lies
within a duplicated region on chromosome 16. Cell 77, 881–894.
Evan, A.P., and McAteer, J.A. (1992). Cyst cells and cyst walls. In
The Cystic Kidney, K.D. Gardner and J. Bernstein, eds. (London:
Kluwer Academic Publishers), pp. 21–41.
Fearon, E.R., Hamilton,S.R., andVogelstein, B. (1987).Clonal analy-
sis of human colorectal tumors. Science 238, 193–197.
Fick, G.M., Johnson, A.M., Strain, J.D., Kimberling, W.J., Kumar, S.,
Manco-Johnson, M.L., Duley, I.T., and Gabow, P.A. (1993). Charac-
teristics of very early onset autosomal dominant polycystic kidney
disease. J. Am. Soc. Nephrol. 3, 1863–1870.
Fick,G.M.,Johnson, A.M.,andGabow, P.A.(1994).Is there evidence
for anticipationin autosomal-dominantpolycystic kidney diesease?
Kidney Int. 45, 1153–1162.
Foster, K., Prowse, A., van den Berg, A., Fleming, S., Hulsbeek,
M. M., Crossey, P.A., Richards, F.M.,Cairns, P., Affara,N.A., Fergu-
son-Smith, M.A., et al. (1994). Somatic mutations of the von Hippel-
Lindau disease tumour suppressor gene in non-familial clear cell
renal carcinoma. Hum. Mol. Genet. 3, 2169–2173.
Gabow, P.A (1993). Autosomal dominant polycystickidney disease.
N. Engl. J. Med. 329, 332–342.
Gilliland, D.G., Blanchard, K.L., Levy, J., Perrin, S., and Bunn, H.F.
(1991). Clonality in myeloproliferative disorders: analysis by means
of the polymerase chain reaction. Proc. Natl. Acad. Sci. USA 88,
Handler, J.S., Steele, R.E., Sahib, M.K., Wade, J.B., Preston, A.S.,
Lawson, N.L.,andJohnson,J.P. (1979).Toad urinarybladderepithe-
portand reponseto hormones. Proc. Natl. Acad. Sci. USA76, 4151–
Huff, V., Miwa, H., Haber, D.A., Call, K.M., Housman, D., Strong,
L. C., and Saunders, G. F. (1991). Evidence for WT1 as a Wilms
tumor (WT) gene: intragenic germinal deletion in bilateral WT. Am.
J. Hum. Genet. 48, 997–1003.
Hughes, J., Ward, C.J., Peral, B., Aspinwall, R., Clark,K., San Milla ´n,
J.L., Gamble, V., and Harris, P.C. (1995). The polycystic kidney dis-
ease1(PKD1) geneencodesanovel proteinwithmultiple cell recog-
nition domains. Nature Genet. 10, 151–160.
Iakoubova, O.A., Dushkin, H., and Beier, D.R. (1995). Localization
of a murine recessive polycystickidney diseasemutation and modi-
fying loci that affect disease severity. Genomics 26, 107–114.
The International Polycystic Kidney Disease Consortium (1995).
Polycystic kidney disease:the complete structure ofthe PKD1 gene
and its protein. Cell 81, 289–298.
Kaariainen, H. (1987). Polycystic kidney disease in children: a ge-
netic and epidemiological study of 82 Finnish patients. J. Med.
Genet. 24, 474–481.
Haplotype and Family Analysis
The allelic genomic fragments that include both the KG8 and EJ1
loci were amplified from each individual using the primers FQF28
and KG8R8 and 200 ng of peripheral blood leukocyte DNA as tem-
plate. The 50 ?l reaction included 4 units of rTth DNA polymerase,
XL (Perkin Elmer), and a final concentration of 1.1 mM magnesium
acetate and was performed using the following conditions: 95?C for
3 min; 35 cycles of 95?C for 20 sec and 68?C for 2 min; and a final
extension of 68?C for 10 min. The 1.6 kb products were cloned into
pCRII, and then individual clones were analyzed for the variants
they contained at each locus using the microsatellite assay (KG8)
and mixing studies (EJ1) as described above.
Family members of patients donating kidneys were recruited to
participate after receiving permission from the donors. Blood sam-
ples were obtained after receiving informed consent and according
to institutional guidelines. DNA was prepared from whole blood
using thePuregene kit.The KG8microsatellite assay wasperformed
as described above. The marker 3?HVR was also used to confirm
linkage of ADPKD to chromosome 16 markers. Five micrograms of
genomic DNA was digested with PvuII (New England Bio Labs),
electrophoresed overnight through a 1% agarose gel, and then
transferred to nylon membrane (Schleicher and Schuel) using stan-
dard Southern blot protocols. 3?HVR was labeled using Rediprime
(Amersham) and hybridized to Southern blots in hybridization buffer
(0.5 M Na2HPO4[pH 7.2], 1 mM EDTA, 1% BSA, and 7% SDS) at
66?C. Blots werewashed two or three times at 65?C with 0.1% SSC,
0.1% SDS and imaged using a Phosphoimager screen (Molecular
Two-point linkage analyseswereperformed using theMLINK pro-
gram of the LINKAGE 5.1 package. The disease gene frequency
was set at 0.001 and penetrance set at 0.98 for individuals >30
years old. Sex-averaged recombination fractions were used (0.05
for 3?HVR and 0.0 for KG8). The posterior probability of linkage to
PKD1 was determined for pedigree 1 (Figure 5A) using an estimate
of theprior probability oflinkage (?) of0.85 (Narod,1991; Peters and
Sandkuijl, 1992; Ravine etal., 1992; Peters et al., 1993; Mochizuki et
Correspondence should be addressed to G. G. G. We are grateful
to all ADPKD family members for their invaluable participation and
the Polycystic Kidney Research Foundation for assistance in the
collection of tissue samples used in the study. We thank Xiangbin
Zhang, EdwardLee,andMartinDaoust fortheirhelp incystprepara-
tion; Dr. Klaus Piontek and Sidney McGaughey for their help in the
preparation of the manuscript; Dr. Irene Maumenee and Thomas
Mitchell of the Johns Hopkins Hereditary Eye Center for assistance
in the linkage calculations; and Dr. Joseph Handler for helpful dis-
cussions. This work was supported by the Polycystic Kidney Re-
search Foundation (grant 95004), the National Insitutes of Health
(grant DK48006), and the McKutcheon Foundation. G. G. G. is the
Irving Blum Scholar of The Johns Hopkins University School of
Received July 19, 1996; revised September 30, 1996.
Allen, R.C., Zoghbi, H.Y., Moseley, A.B., Rosenblatt, H.M., and Bel-
mont, J. W. (1992). Methylation of HpaII and HhaI sites near the
Molecular Basis of Focal Cyst Formation in PKD1
Kimberling, W.J., Kumar, S., Gabow, P.A., Kenyon, J.B., Connolly,
C.J., and Somlo, S. (1993). Autosomal dominant polycystic kidney
disease: localization of the second gene to chromosome 4q13-q23.
Genomics 18, 467–472.
Mochizuki, T., Wu, G., Hayashi, T., Xenophontos, S.L., Veld-
huisen, B.,Saris, J.J.,Reynolds, D.M., Cai, Y.,Gabow, P.A., Pierides,
A., et al. (1996). PKD2, a gene for polycystic kidney disease that
encodes an integral membrane protein. Science 272, 1339–1342.
Narod, S. (1991). Counselling under genetic heterogeneity: a practi-
cal approach. Clin. Genet. 39, 125–131.
Peral, B., Gamble, V., San Milla ´n, J.L., Strong, C.Sloane-Stanley,
J., Moreno, F., and Harris, P.C. (1995). Splicing mutations of the
polycystic kidney disease 1 (PKD1) gene induced by intronic dele-
tion. Hum. Mol. Gen. 4, 569–574.
Peral, B., San Milla ´n, J.L., Ong, A.C., Gamble, V.,Ward, C.J., Strong,
C., andHarris, P.C. (1996a).Screeningthe 3?region ofthepolycystic
kidney disease 1 (PKD1) gene reveals six novel mutations. Am. J.
Hum. Genet. 58, 86–96.
Peral, B., Ong, A.C.M., San Milla ´n, J.L., Gamble, V., Rees, L., and
Harris, P.C. (1996b). A stable, nonsense mutation associated with
a case of infantile onset polycystic kidney disease 1 (PKD1). Hum.
Mol. Gen. 5, 539–542.
Peters, D.J.M., and Sandkuijl, L.A. (1992). Genetic heterogeneity of
polycystic kidney in Europe. Contrib. Nephrol. 97, 128–139.
Peters, D.J.M., Spruit, L., Saris, J J., Ravine, D., Dandkuijl, L.A.,
Fossdal, R., Boersma, J., van Eijk, R., Borby, S., Constantinou-
Deltas, C.D., et al. (1993). Chromosome 4 localization of a second
gene for autosomal dominant polycystic kidney disease. Nature
Genet. 5, 359–362.
Rao, B.S., Manor, H., and Martin, R.G. (1988). Pausing in simian
virus 40 DNA replication by a sequence containing (dG-dA)27·(dT-
dC)27. Nucleic Acids Res. 16, 8077–8094.
Ravine, D., Walker, R.G., Gibson, R.N., Forrest, S.M., Richards, R.I.,
Friend, K.,Sheffield,L.J., Kincaid-Smith, P., and Danks,D.M. (1992).
Phenotype and genotype heterogeneity in autosomal dominant
polycystic kidney disease. Lancet 340, 1330–1333.
Reeders, S.T., Breuning, M.H., Davis, K.E.., Nicholls, R.D., Jarman,
A.P., Higgs, D.R., and Weatherall, D.J. (1985). A highly polymorphic
DNA marker linked to adult polycystic kidney disease on chromo-
some 16. Nature 317, 542–544.
Scha ¨fer, K.,Gretz, N., Bader, M., Obermu ¨ller, I., Eckardt, K.-U., Kriz,
W., and Bachmann, S. (1994). Characterization of the Han:SPRD
rat model for hereditary polycystic kidney disease. Kidney Int. 46,
Snarey, A., Thomas, S., Schneider, M.C., Pound, S.E., Barton, N.,
Wright, A.F., Somlo, S., Germino, G.G., Harris, P.C., and Reeders,
S.T. (1994). Linkage disequilibrium in the region of the autosomal
dominant polycystic kidney disease gene (PKD1). Am. J. Hum.
Genet. 55, 365–371.
Tilley, W.D. Marcelli, M., Wilson, J.D., and McPhaul, M.J. (1989).
Characterization and expression of a cDNA encoding the human
androgen receptor. Proc. Natl. Acad. Sci. USA 86, 327–331.
Wang, G., Seidman, M.M., and Glazer, P.M. (1996). Mutagenesis in
mammaliancells inducedby triplehelix formation andtranscription-
coupled repair. Science 271, 802–805.
Willman, C.L., Busque, L., Griffith, B.B., Favara, B.E., McClain, K.L.,
Duncan, M.H., and Gilliland, D.G. (1994). Langerhans’-cell histio-
cytosis (histiocytosis X): a clonal proferative disease. N. Engl. J.
Med. 331, 154–160.
Young, SD. L., Krawczyk, S.H., Matteucci, M.D., and Toole, J.J.
Proc. Natl. Acad. Sci. USA 88, 10023–10026.
Zerres, K., Rudnik-Schoneborn, S., and Deget, F. (1993). Childhood
onset autosomaldominantpolycystickidney diseaseinsibs: clinical
picture and recurrence risk. German Working Group on Paediatric
Nephrology. J. Med. Genet. 30, 583–588.