A mouse model for cystic biliary dysgenesis in autosomal recessive polycystic kidney disease (ARPKD)


  • Potential conflict of interest: Nothing to report.


Autosomal recessive polycystic kidney disease (ARPKD) is an important cause of liver- and renal-related morbidity and mortality in childhood. Recently, PKHD1, the gene encoding the transmembrane protein polyductin, was shown to be mutated in ARPKD patients. We here describe the first mouse strain, generated by targeted mutation of Pkhd1. Due to exon skipping, Pkhd1ex40 mice express a modified Pkhd1 transcript and develop severe malformations of intrahepatic bile ducts. Cholangiocytes maintain a proliferative phenotype and continuously synthesize TGF-β1. Subsequently, mesenchymal cells within the hepatic portal tracts continue to synthesize collagen, resulting in progressive portal fibrosis and portal hypertension. Fibrosis did not involve the hepatic lobules, and we did not observe any pathological changes in morphology or function of hepatocytes. Surprisingly and in contrast to human ARPKD individuals, Pkhd1ex40 mice develop morphologically and functionally normal kidneys. In conclusion,our data indicate that subsequent to formation of the embryonic ductal plate, dysgenesis of terminally differentiated bile ducts occurs in response to the Pkhd1ex40 mutation. The role of polyductin in liver and kidney may be functionally divergent, because protein domains essential for bile duct development do not affect nephrogenesis in our mouse model. Supplementary material for this article can be found on the HEPATOLOGYwebsite (http://www.interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2005.)

Autosomal recessive polycystic kidney disease (ARPKD [MIM 263200]) or polycystic kidney and hepatic disease 1 (PKHD1) is an inherited disorder of the kidney and liver with an estimated incidence of 1 in 20,000 live births.1, 2 Its principal manifestations involve the fusiform dilatation of renal collecting ducts and tubuli and fibrosis of the hepatic portal triad. The only signs potentially displayed in utero, albeit often not visible in second trimester fetal sonography, are enlargement and increased echogenicity of both kidneys as well as oligohydramnios.3 Up to 30% of the affected neonates die shortly after birth as a result of severe pulmonary hypoplasia and secondary respiratory failure. In ARPKD patients surviving the neonatal period, the prognosis is more optimistic, and the disease phenotypes may vary significantly.4 Frequent complications include systemic hypertension, end-stage renal disease, and portal hypertension resulting from congenital hepatic fibrosis.

In 1994, the ARPKD gene was mapped to human chromosome 6p21-cen,3 and subsequent linkage data showed no evidence for genetic heterogeneity, despite the variable clinical phenotypes. Two independent strategies finally identified the PKHD1 gene. Ward et al.5 characterized the mutation in the PCK rat model for polycystic kidney disease and identified germline mutations in ARPKD patients in the human PKHD1 orthologue. Independently, an international consortium isolated the ARPKD gene by positional cloning.6–8 PKHD1 was shown to encode a novel large transmembrane protein with a minimum of 86 exons and many alternative splice variants. The longest open reading frame was shown to encode a 67-exon transcript for a predicted protein of 4074 amino acids, which was termed polyductin or fibrocystin. Polyductin/fibrocystin contains mostly extracellular domains with several Ig-like, plexin, transcription factor motifs, which are shared by hepatocyte growth factor receptor and the plexin superfamily involved in cellular adhesion, repulsion, and proliferation, and multiple parallel beta-helix 1 repeats. There is only one predicted transmembrane domain and a very short intracellular C-terminal tail. Analysis of differentially spliced messenger RNA (mRNA) transcripts suggests that there also may be secreted polyductin/fibrocystin isoforms devoid of a transmembrane domain.8

We recently identified the murine Pkhd1 orthologue and showed that the overall protein structure and the complex pattern of splicing is highly conserved.9 We then aimed to investigate the role of polyductin/fibrocystin during mouse development and therefore constructed mice with a targeted Pkhd1 germline mutation.


ARPKD, autosomal recessive polycystic kidney disease; PKHD1, polycystic kidneys and hepatic disease-gene 1; PCK, polycystic kidney disease; mRNA, messenger RNA; cDNA, complementary DNA; ES, embryonic stem; PCR, polymerase chain reaction; GAPDH, glyceralaldehyde-3-phoshate dehydrogenase; TGF-β1, transforming growth factor beta-1; RT, reverse transcription.

Materials and Methods

Genomic Library Screening and Isolation of a Murine Pkhd1 Gene Fragment.

Library plating and plaque lifting was performed using standard procedures. A partial human ARPKD complementary DNA (cDNA) probe spanning exon 419 was used to screen 4.5 × 105 recombinant lambda phages of a murine 129SVJ genomic library (Stratagene, Heidelberg, Germany). Hybridization and washing were performed at intermediate stringency with the final wash in 0.25× standard saline citrate (SSC), 0.0125% sodium dodecyl sulfate at 62°C for 40 minutes. A single lambda phage clone covering a 14-kb insert of the murine Pkhd1 locus spanning exons 40 to 44 was isolated, and the insert was verified by DNA sequencing.

Construction of the Pkhd1 Targeting Vector, Polymerase Chain Reaction Assays, and Generation of Gene-Disrupted Mice.

From the isolated lambda phage, a partial 7.6-kb genomic Pkhd1 fragment harboring murine exon 40 (homologous to the human exon 41) was subcloned into pBluescript II SK (−) (Stratagene), and site-directed mutagenesis was performed to create a unique XhoI site within exon 40. The targeting vector was designed to disrupt exon 40 by a LacZ-PGK-neo cassette, flanked by 5.5-kb 5′ homologous and 2-kb 3′ homologous 129SVJ genomic sequences. As described previously,6, 10 25 μg linearized plasmid DNA was electroporated into 5 × 107 R1 embryonic stem (ES) cells, cultured, and selected by G418.

Correctly targeted ES cells were microinjected into blastocysts and transferred into CD-1 foster mice. Two independent ES mouse lines were generated by germline transmission from chimeric founders and were used to breed the Pkhd1−/− mutation on a C57/BL6 background. Wild-type and mutated animals described in this study always represent littermates with +/+ and −/− genotypes, respectively, obtained from mating +/− animals.

Southern blot analysis was used to identify homologous recombination in ES cells and to genotype the first generations of mice. Genomic DNA was digested with BamHI and separated on 0.8 % aragose gels. A [32P]-dCTP random prime-labeled probe comprising a 425-bp genomic SacI–HindIII fragment of the Pkhd1 gene (Fig. 1A) was used to identify the 4.2-kb wild-type allele and the 2.2-kb mutated allele.

Figure 1.

PKHD1 targeting vector and genotyping of mutant mice. (A) Schematic representation of the partial PKHD1 gene locus and the targeting construct. Black boxes indicate PKHD1 exons 40 through 44, and the gray box indicates the probe used for Southern hybridization, which recognizes the 4.2-kb wild-type and the 2.2-kb mutant PKHD1 BamHI fragment. (B) Southern blot and PCR analysis of tail biopsy DNAs from the progeny of mating heterozygotes. The positions of primer pairs for amplification of the wild-type and mutant PCR fragments are shown in panel A. wt, wild-type; ko, mutated. (C) Northern blot of kidney and liver mRNA extracted from wild-type and Pkhd1ex40-mutated mice (ko) using exon 40 (left panel) and exons 65 to 67 as probes. (D) RT-PCR amplification of C-terminal Pkhd1 fragments from wild-type and mutant mice (ko) indicates that the transcript in mutant mice encodes the C-terminal transmembrane domain. (E) RT-PCR analysis of kidney and liver RNA indicating exon 40 skipping in PKHD1ex 40 mice. Primer pairs were designed matching to sequences flanking exon 40 as indicated in the right margin. Neg. means RT-PCR without adding reverse transcriptase. (F) Schematic representation of the polyductin mutation introduced into Pkdh1ex40 mice.

After expanding the two knockout lines, mouse genotyping was performed by polymerase chain reaction (PCR). Three primers were used to detect wild-type and null-mutant alleles: mEx40 sense (5′-CTAAGA GTGAGAAGATGCTGG-3′), mEx40 antisense (5′-CTGGGAGATCAACGCTGCTC-3′) and PGK-polyA down (5′-CTGCTCTTTACTGAAGGCTCTTT-3′), which amplified 188-bp (wild-type) and 400-bp (null-allele) fragments. PCR was performed with Taq polymerase (Invitrogen, Karlsruhe, Germany) and with the following profile: 94°C for 45 seconds, 53°C for 30 seconds, 72°C for 30 seconds for 35 cycles, and finally 72°C for 5 minutes.

RNA Isolation, Northern Blots, and Reverse Transcription PCR.

Tissues from mice at 3, 5, 10, and 14 weeks of age were disrupted with a rotor-stator homogenizer, and total cellular RNA was extracted by standard protocols. Thirty micrograms total RNA per lane was separated on a 1.2% formaldehyde agarose gel and transferred to nylon membranes. The blot was hybridized with a cDNA probe encoding exon 40 of murine Pkhd1, and a second hybridization was performed using a cDNA probe spanning the C-terminal domain of Pkhd1 (encompassing the 3′ region of exon 65 to 67). Final washes were done in 0.5× standard saline citrate, 0.1% sodium dodecyl sulfate at 65°C for 20 minutes. For loading control, blots were reprobed with a mouse GAPDH (glyceralaldehyde-3-phoshate dehydrogenase) probe.

For reverse transcription, 1 μg total RNA was primed with 1 μg oligo d(T)12–18, and transcription was performed by Superscript II RT following the manufacturer's description (Invitrogen). Five percent to 20% of the reaction (1–4 μL) was added to 1 μL 10 mmol/L deoxyribonucleotide triphosphate, 5 μL 10× PCR buffer, 1.5 μL of sense and antisense primer each (50 pmoles/μL), 5 units Taq polymerase (Roche, Mannheim, Germany), and 36.5 to 39.5 μL H2O.

PCR was run with 35 cycles (1 minute at 94°C, 45 seconds at 53°C to 55°C, and 1 to 2 minutes at 72°C). Primers mEx40 sense (see above) and mEx40 antisense (5′-CTCATAGCTCCCACCAGT GCG-3′) were used to amplify a 167-bp cDNA fragment harbouring murine Pkhd1 exon 40 (homologous to human exon 41). A 385 bp cDNA fragment spanning Pkhd1 exons 39 - 41 was amplified using the primers mEx39 sense (5′-GCCAACAGTTGCTCTCCTCAGC-3′) and mEx41 antisense (5′-CATTGCTGTCCACCTTCAGAC-3′). To create a 774-bp cDNA fragment including Pkhd1 exons 38 to 43, the primers mEx38 sense (5′-GCAGCTGCA CATGCAGGAGAC-3′) and mEx43 antisense (5′-GACAGTCCCTCAGCTGCAG-3′) were used. To amplify the cDNA fragment encoding amino acids 3701 to 4059, primers mEx62 sense (5′-CTGTGGGGGCCCTACTAGTGAC-3′) and mTAA antisense (5′-CATGCC CACAGCTGTTACTG-3′) were used and to amplify the cDNA fragment encoding amino acids 3873 to 4059 primers mEx65 sense (5′-GCGTGGATCCTTTAAGAAAAGCAA-3′) and mStop antisense (5′-GGCCCTTAAGTTACTGGATGGTTTCT-3′), respectively. Collagen type I was amplified using the primers ColI sense (5′-ACCTGTGTGTTCCCTACTCA-3′) and ColI antisense (5′-GACTGTTGCCTTCGCCTCT-3′), Collagen type III was amplified using the primers ColIII sense (5′-AATGGTGGTTTTCAGTTCAGC-3′) and ColIII antisense (5′-TGGGGTTTCAGAGAGTTTGGC-3′). As a control, β-actin mRNA was amplified using 5′-TGGAATCCTGTGGCATCCATGAAAC-3′ and 5′-TAAAACGCAGCTCAGTAA-CAGTCCG-3′ primers or GAPDH mRNA was amplified using GAPDH sense (5′-ACCACAGTCCATGCCATCAC) and GAPDH antisense (5′-TCCACCACCATGT-TGCTGTA).

Histological Preparations and Immunostaining.

Organs were dissected and fixed in 4% (wt/vol) paraformaldehyde in phosphate-buffered saline. Staining with hematoxylin-eosin and Sirius staining were performed using standard protocols. Bars shown in the histological slides represent 100 μm.

For immunohistochemical stainings, paraffin-embedded tissue sections were deparaffinized and incubated in 0.6% H2O2/methanol for 30 minutes, then washed, microwave treated, pretreated with proteolytic enzyme, and blocked with avidin according to manufacturer's instructions (Linaris, Wertheim, Germany). After incubation in 10% horse serum/phosphate-buffered saline for 20 minutes, Ki-67 antiserum was diluted 1:30 (Dako, Hamburg, Germany), collagen III antibodies 1:100 (Collagen Type III, Sigma, Deisenhofen, Germany), and transforming growth factor beta-1 (TGF-β1 (Santa Cruz, Heidelberg, Germany) 1:100 in phosphate-buffered saline. Incubation was performed for 1 hour at 37°C. After washing, the sections were incubated with biotinylated secondary antibodies. Immunoreactions were visualized using diaminobenzidine as a substrate (Vectastain ABC Elite Kit, Vector Laboratories, Burlingame, CA). To quantify proliferative activity, Ki-67–immunopositive nuclei were counted in 1,000 intrahepatic bile duct epithelial cells. To quantify soluble TGF-β1 from liver extracts, a commercially available ELISA kit was used, following precisely the manufacturer's instructions (Bender Medsystems GmbH, Vienna, Austria).

Electron Microscopy.

Transmission electron microscopy was performed as described previously.10 Briefly, specimens were fixed in 2.5% glutaraldehyde/0.1 mol/L sodium cacodylate (pH 7.4) and postfixed in 1% osmium tetroxide/0.1 mol/L sodium cacodylate. Embedding was done in Epon 812 over 1 week, and 1-μm semi-thin sections were stained for light microscopy with toluidine blue 0. Finally, 60-nm thin sections were cut and stained with uranyl acetate and lead citrate.

Serum Assays.

Whole blood was obtained from 6 to 7 mice each between 8 weeks and 14 months of age, and the serum was recovered after coagulation by centrifugation for 5 minutes. Serum parameters were measured by an autoanalyzer (Johnson and Johnson, Neckarsgemünd, Germany) as previously described.11


Generation of Pkhd1ex40-Mutated Mice.

We previously characterized the murine Pkhd1 gene and performed detailed expression analyses in embryonic and adult tissues.9 This study also provided evidence that differentially spliced Pkhd1 mRNAs are expressed in a tissue-specific manner. Because analyses by multiple tissue Northern blots and in situ hybridization indicated that our probe for exon 40 (homologous to exon 41 of the human gene) hybridized with the largest, most abundant transcripts in liver and kidney and also with differentially spliced smaller transcripts in testis, we decided to disrupt exon 40 by insertion of a LacZ-PGK-neo selection cassette (Fig. 1A). Genotyping of neomycin-resistant stem cell clones was performed by Southern blotting (data not shown), and 2 independent stem cell clones were used for germline transmission of the mutation. Initially, genotyping the mice from tail biopsies was performed by Southern blot, but subsequently 2 PCRs specific for the wild-type and mutated alleles were designed for routine genotyping. Primer pairs and the Southern probe are indicated schematically in Fig. 1A, and a representative blot and PCR gel resulting from wild-type, heterozygous, and homozygous mutant mice are shown in Fig. 1B.

Pkhd1 mutated mice were born at the expected Mendelian frequency. Of 130 progeny, we received 30 wild-type (23%), 66 heterozygous (51%), and 34 homozygous (26%) mutants. Mice of all genotypes were fertile and, when autopsies were performed 5 weeks after birth, showed no phenotypic abnormality based on macroscopic inspection. Surprisingly, we did not detect polycystic transformation of the kidneys in Pkhd1-deficient mice. To determine the effect of the Pkhd1ex40 mutation on mRNA expression, we analyzed RNA extracted from kidney and liver by Northern blots and reverse transcription (RT)-PCR (Fig. 1C-E). We previously observed the highest levels of Pkhd1 mRNA expression in kidney and lower levels in liver.9 Northern blot hybridization using exon 40 as a probe and RT-PCR amplification using an exon 40 primer pair failed to detect Pkhd1 mRNA expression in kidney and liver of homozygous mutants (Fig. 1C,E). However, a C-terminal cDNA probe matching exons 65 to 67 clearly indicated expression of a very large transcript in the kidney of Pkhd1ex40 mice similar in size to the wild-type transcript. RT-PCR analyses using C-terminal primer pairs matching amino acids 3701 to 4059 and 3873 to 4059, respectively, and indicated consistently that Pkhd1ex40 mice express a transcript encoding the transmembrane domain (Fig. 1D). Further RT-PCRs with different primer pairs matching sequences flanking exon 40 indicated expression of an mRNA transcript generated by skipping exon 40 (Fig. 1E). We therefore cloned the RT-PCR products shown in Fig. 1E and determined by sequencing that the largest message transcribed from the mutated Pkhd1ex40 gene is generated by skipping exon 40, leading to an in-frame deletion of amino acid residues 2160 to 2223. Consistent with our results from RT-PCRs, we failed to detect any LacZ staining in heterozygous or homozygous mutant mice, indicating that the entire exon 40 including the LacZ cassette is skipped (data not shown). The location of the mutation in the mouse Pkhd1 protein is schematically shown in Fig. 1F.

Ductal Plate Malformation in Pkhd1ex40 Mice.

Because human ARPKD patients suffer from cystic biliary dysgenesis and polycystic kidneys, detailed histological analyses of all internal organs with special emphasis on liver and kidneys were performed. As shown in Fig. 2A through D, histological liver sections of homozygous neonatal Pkhd1ex40 mice showed severe malformation of the ductal plate when compared with wild-type littermates. The embryonic architecture of bile ducts encirculating the portal veins was retained and, in addition, mild duct ectasia was present. This phenotype was never detected in wild-type or heterozygous littermates. Systematic analysis of embryos at gestational stages E12.5, E14.5, E16.5, and E18.5 showed that ductal plate malformation was already present in embryonic mice at stage E16.5 and started first at the portal tracts located near the porta hepaticae and was completed in the peripheral portal tracts in newborn mice (Fig. 6, see supplemental figures at (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). These data suggested that development of biliary dysgenesis in Pkhd1ex40 mice paralleled remodeling of the ductal plate architecture. We further observed progressive deterioration of the mice. Significant bile duct proliferation and ectasia accompanied by fibrosis of the portal tracts developed between 5 and 14 weeks after birth (Fig. 2E-H). Progressive portal fibrosis became especially evident on Sirius stains (Fig. 2B,D,F,H) and again was never observed in wild-type or heterozygous littermates.

Figure 2.

Histological analysis of liver changes in Pkhd1ex40 mice. Liver sections from adult wild-type littermates 14 weeks after birth (A-B), from Pkhd1ex40 mice 1 day (C-D), 5 weeks (E-F), and 14 weeks after birth (G-H). Panels A, C, E, and G show hematoxylin staining, and panels B, D, F, and H Sirius staining. PV, portal vein; bd, bile duct; lob, hepatic lobule; cv, hepatic central vein.

Cystic biliary dysgenesis progressed even further in mice at the age of 14 months (Fig. 3). All homozygous animals showed macroscopically visible polycystic livers. The size of cystic liver degeneration was slightly variable, differing from multiple small cysts and sometimes also huge liver cysts (Fig. 3B-C). The animal shown in the lower panel of Fig. 3C also showed an enlarged spleen and ascites, indicating portal hypertension. Liver histology of Pkhd1ex40 mice at age 14 months again showed maximal polycystic liver transformation. However, the liver changes were confined to abnormal bile duct architecture and fibrosis in the portal tracts, and the lobular hepatic parenchyme remained entirely normal (see Fig. 3D,F,H). Only in end-stage cystic transformation occasional secondary degeneration of hepatic lobules was present between biliary cysts.

Figure 3.

Macroscopic and histological analysis of Pkhd1ex40 mice at 14 months of age. (A) Normal liver in a wild-type mouse and (B, C) polycystic livers in homozygous mutant littermates. Histological liver sections of the same animals are shown in panels D-I and kidney sections in panels J-M. bd, bile duct; pv, portal vein; hep, degenerated hepatic lobule between cystic bile ducts; cd, collecting duct; gm, glomerulum.

In contrast to the liver changes, we never observed any pathological changes in kidneys of mutant Pkhd1ex40 mice (Fig. 3J-M). Collecting ducts (Fig. 3J-K), renal tubules, as well as glomeruli (Fig. 3L-M) were indistinguishable between wild-type and homozygous mutant littermates even at age 14 months. Extrahepatic bile duct and gallbladder development was not affected by the mutation (data not shown).

Further detailed examination of the cholangiocellular ultrastructure by electron microscopy showed normal cholangiocytes in Pkhd1ex40 mutant mice (Fig. 7A-B, supplemental figures). Ultrastructurally, cysts were lined by a single layer of normally differentiated cuboidal epithelial cells with well-formed microvilli and cell junctions. Cystic cells were not enlarged and showed no flattening of the convoluted basement membrane. In contrast to a recently published study in the PCK rat model,12 the number of mitochondria was not altered, and we did not observe any ciliary abnormalities in cholangiocytes of Pkhd1ex40 mutants. Ultrastructural analysis of 65 cilia from 8 different mutated mice showed no deviation from the normal 9+0 internal structure of microtubules (inset), characteristic of primary cilia, and no signs of apical degeneration (Fig. 7C, supplemental figures).

It is known from ARPKD patients who survive the neonatal period that hepatocellular function is typically normal, but kidney serum parameters deteriorate gradually to end-stage renal failure.4, 13 Therefore, liver and kidney serum parameters were measured in wild-type and Pkhd1ex40-mutated mice. We found that the mean values and standard deviation of serum creatinine, phosphate, potassium, creatine kinase, urea, glutamic oxaloacetic acid, glutamic pyruvic transaminase, alkaline phosphatase, gamma-glutamyltransferase, total protein, and bilirubin values when measured at different ages between 8 weeks to 14 months did not differ between wild-type and homozygous mutant littermates (data not shown). From these data, we concluded that the Pkhd1ex40-mutation in mice leads to bile duct malformation with progressive cystic transformation and congenital fibrosis of the hepatic portal tracts, but not to polycystic kidney transformation or any other impairment of renal function.

Progressive Portal Fibrosis in Pkhd1ex40-Deficient Mice.

In comparison with wild-type mice (Fig. 4A), Pkhd1ex40-deficient mice progressively accumulate fibrillar collagen in portal tracts. Immunohistochemical stainings exemplarily shown in Fig. 4B visualized collagen type III deposition, particularly around malformed bile ducts, and the same results were obtained by collagen I staining (data not shown). Deposition of fiber bundles increased over time and caused severe portal tract fibrosis in animals 14 weeks after birth. Consistently, we measured increased synthesis of collagen I and III mRNA in the liver of mutated mice by RT-PCR (Fig. 5A).

Figure 4.

Progressive portal fibrosis, TGF-β1 expression and proliferation of cholangiocytes in Pkhd1ex40 mice. Panels show indirect peroxidase immunostainings of liver sections with primary antibodies indicated in the left margin. (A, C, E) Wild-type mice, (B, D, F) Pkhd1ex40 mice. PV, portal vein; bd, bile duct.

Figure 5.

Collagen, TGF-β1 mRNA expression and TGF-β1 protein in wild-type and Pkhd1ex40 mice. (A) Panels show agarose gels loaded with RT-PCR amplification products. Col I, collagen type I; Col III, collagen type III; TGF-β1; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; M, molecular weight marker; wt, wild-type mice; mut, Pkhd1ex40 mice. (B) TGF-β1 protein measured by a quantitative ELISA in liver protein extracts of E19.5, 7-day-old and 3-week-old wild-type (gray bar) and homozygous mutant littermates (black bars).

Although the precise molecular mechanisms controlling human biliary morphogenesis are unclear, it is known that soluble factors secreted from cholangioblasts direct mesenchymal cell growth and limit epithelial proliferation during biliary ductal remodeling. In particular, immunoreactivity for TGF-β1 is present in the developing intrahepatic bile ducts and down-regulated in the fully differentiated biliary tree.14 We therefore immunostained liver sections for TGF-β and detected no immunoreactivity in the portal tracts of control sections derived from normal postnatal mice (Fig. 4C bile ducts marked by arrows). In contrast, significant TGF-β1 immunoreactivity and increased TGF-β1 mRNA synthesis was present in bile duct epithelial cells of Pkhd1ex40-mutated mice (see Fig. 4D and Fig. 5A), resembling the staining pattern of primitive bile ducts of the ductal plate.

Consistently, we also measured significantly enhanced TGF-β1 concentrations in liver protein extracts of mutated mice (Fig. 5B). Although the portal tracts constitute only a minor portion of the whole liver, TGF-β1 levels were approximately 40% higher in mutant versus wild-type mice. Consistent with an embryonic phenotype, we also observed continuous proliferation of bile duct epithelial cells in adult Pkhd1ex40-mutant mice. Approximately 2% Ki-67 immunoreactive nuclei were present in bile duct epithelial cells of mutated mice, compared with less than 0.1% in wild-type mice (Fig. 4E-F).


In this report we describe a mouse strain, Pkhd1ex40, with a targeted mutation in the Pkhd1 gene. Both human and murine Pkhd1 genes encode a large transmembrane protein, referred to as polyductin or fibrocystin, subject to a complicated pattern of exon assembly and alternative exon usage. We chose to mutate exon 40, because it is part of the most abundant mRNA expressed in liver and kidney.9 The Pkhd1 transcript expressed in kidney and liver of Pkhd1ex40 mice is only slightly modified, encoding a protein lacking amino acid residues 2160–2223 due to skipping exon 40. Analyses by RT-PCR and LacZ staining indicated that the entire exon 40 including the targeting cassette is removed by splicing from the mutated transcript. Surprisingly, homozygous Pkhd1ex40 mice undergo dysgenesis of intrahepatic bile ducts starting at the time of embryonal remodeling of the ductal plate and develop portal fibrosis in the absence of any kidney disease.

The ductal plate initially forms as a single cell layer around the later portal tracts in immediate contact with the periportal hepatoblasts (reviewed by Zerres et al.2). In a centrifugal manner from the liver hilum to the periphery, portal mesenchymal cells interpose between the ductal plate and the hepatocellular parenchyme and deposit matrix in response to TGF-β1 and TGF-α secretion from cholangioblasts. Thereby, the ductal plate becomes physically separated from the periportal hepatocellular parenchyme.15 Then, the ductal plate duplicates through continued cholangioblast proliferation to form a primitive cylindric bile duct encircling almost the entire portal tract. It is assumed that physical separation of the ductal plate provides a signal to the cholangioblasts to undergo apoptosis and a series of subsequent architectural remodeling steps to form mature bile ducts longitudinally orientated along the portal veins. Consistently, it has been shown that the contact to mesenchyme is essential for differentiation into biliary epithelium of isolated liver cells in vitro.16

Analyzing Pkhd1ex40 mice showed that the failure of embryonic ductal plates to undergo architectural changes after successful delineation from the periportal hepatocytes retains the cholangiocytes in a proliferative and TGF-β1-immunoreactive state. It appears that embryonic cholangioblasts in Pkhd1ex40 mice continuously stimulate mesenchymal cells to synthesize and accumulate excessive matrix. Based on these findings, we hypothesize that PKHD1 may act as a matrix receptor during maturation of intrahepatic bile ducts. Previously, we showed that PKHD1 is expressed already at very early stages during hepatic development specifically in cells of the ductal plate.9 The spatially and temporally appropriate expression patterns and the bile duct defects in Pkhd1ex40 mice are in support of our hypothesis of PKHD1 acting as a matrix sensor and signal receptor during bile duct development.

Unexpectedly, we failed to detect any changes in kidney morphology or function in Pkhd1ex40 mice, although the PKHD1 mRNA in kidney harbors the same mutation as the mRNA in liver. Kidney morphology was analyzed in both newborn and adult mice, and all serum tests indicated normal glomerular and tubular kidney functions. Very recently, polyductin/fibrocystin was found to be located to primary cilia, and ciliary defects were described in a rat model of ARPKD.12, 17 Because we detected ultrastructurally normal primary cilia in the liver of Pkhd1-mutated mice, it is possible that ciliary defects may not cause the entire spectrum of biliary pathology.

Because the mutation destroys a conserved protein motif adjacent to parallel beta-helix 1-repeats, we conclude that this motif may be redundant for nephrogenesis and adult kidney function but essential for intrahepatic bile duct development. We also can exclude that deficient PKHD1 function becomes apparent at later adult stages, because we have now observed normal kidney morphology and function in mice even at the age of 14 months. Thus, the mutation in Pkhd1ex40 mice affects functions of polyductin in liver and kidney development in a different manner.

Interestingly, studies of ARPKD families have identified a mild and a severe phenotype in the absence of genetic heterogeneity.3 Based on these earlier findings and our study, it is therefore likely that there is a genotype–phenotype correlation resulting from specific mutations in different PKHD1 motifs. Consistently, our previous analysis of mutations in a large cohort of ARPKD patients showed that patients with 2 truncating mutations displayed a severe phenotype with perinatal or neonatal demise.18 However, in the absence of data concerning functional relevance of different PKHD1 protein motifs, it is impossible to precisely interpret the frequent missense mutations. In this respect, our study provides further evidence that the precise phenotype of hepatic and kidney disease in ARPKD patients correlates with the genotype.

In summary, in this context, it is interesting to note that a second detailed screen of PKHD1 sequence changes detected mutations in an ARPKD cohort and also in patients with Caroli′s disease.19 We therefore recommend that neonatal patients and children with congenital hepatic fibrosis and Caroli′s disease should be screened for PKHD1 mutations even in the absence of renal pathology.


The authors thank Anke Röper and Jörg Bedorf for excellent technical assistance with immunohistochemistry and electron microscopy and Andrea Jacob for maintenance of the animal colony.