J Am Soc Nephrol 12: 2664–2672, 2001 Primary Structure and Expression Studies of Rodent Paracellin-1

STEFANIE WEBER,* KARL P. SCHLINGMANN,* MELANIE PETERS,* LENE NIEMANN NEJSUM,† SØREN NIELSEN,† HARTMUT ENGEL,‡ KARL-HEINZ GRZESCHIK,‡ HANNSJO¨ RG W. SEYBERTH,* HERMANN-JOSEPH GRO¨ NE,§ ROLF NU¨ SING,* and MARTIN KONRAD* *Department of Pediatrics, Philipps University, Marburg, Germany; †Department of Cell Biology, Institute of Anatomy, University of Aarhus, Aarhus, Denmark; ‡Institute of Human Genetics, Philipps University, Marburg, Germany; and §Deutsches Krebsforschungszentrum, Heidelberg, Germany.

Abstract. The novel member of the claudin multigene family, lengths and splice-site loci. By radiation hybrid mapping, the paracellin-1/claudin-16, encoded by the gene PCLN1, is a renal murine Pcln1 gene was assigned directly to marker D16Mit133 tight junction that is involved in the paracellular trans- on mouse 16 (syntenic to a on human port of magnesium and calcium in the thick ascending limb of chromosome 3q27, which harbors the human PCLN1 gene). Henle’s loop. Mutations in human PCLN1 are associated with Mouse multiple-tissue Northern blot showed Pcln1 expression familial hypomagnesemia with hypercalciuria and nephrocal- exclusively in the kidney. The expression profile along the cinosis, an autosomal recessive disease that is characterized by nephron was analyzed by reverse transcriptase–PCR on micro- severe renal magnesium and calcium loss. The complete cod- dissected nephron segments and immunohistochemistry of rat ing sequences of mouse and rat Pcln1 and the murine genomic kidney. Paracellin-1 expression was restricted to distal tubular structure are here presented. Full-length cDNAs are 939 and segments including the thick ascending limb of Henle’s loop, 1514 bp in length in mouse and rat, respectively, encoding a the distal tubule, and the collecting duct. The identification and putative open-reading frame of 235 amino acids in both species characterization of the rodent Pcln1 provide the basis for with 99% identity. Exon-intron analysis of the human and further studies of paracellin-1 function in suitable animal mouse genes revealed a 100% homology of coding exon models.

Members of the claudin multigene family have been identified fied in the kidney, especially in the thick ascending limb of as essential components of the endothelial and epithelial tight Henle’s loop (TALH) (4). junction barrier. In 1998, claudin-1 and claudin-2 were first Mutations in PCLN1 were found to be associated with described by Furuse et al. (1) and were colocalized with the familial hypomagnesemia with hypercalciuria and nephrocal- ubiquitous tight junction protein occludin to cell-to-cell adhe- cinosis (FHHNC, OMIM 248250), a hereditary renal disease sions in chicken liver. Since then, a great number of with urinary magnesium and calcium loss and progression to with related structure were identified in tight junction strands end-stage renal failure (4,7,8). As concluded from the pheno- of multiple tissues (2,3); some of them associated with hered- type of FHHNC patients, an important role of paracellin-1 in itary diseases in humans (4–6). All claudins represent mem- the regulation of paracellular transport of divalent cations in brane proteins with four transmembrane domains, two extra- the TALH was suggested. Little is known about human mag- cellular loops, and intracellular N- and C-termini. Recently, nesium transport mechanisms; therefore, the identification of paracellin-1/claudin-16, which is encoded by the gene PCLN1, paracellin-1 represented a major step to elucidate the mecha- was characterized as a new member of this protein family in nisms of renal magnesium reabsorption. The bovine PCLN1 humans. Although other members of the claudin family show gene was recently identified, and PCLN1 deletions were de- a wide tissue distribution, paracellin-1 was exclusively identi- scribed in Japanese Black cattle affected by an autosomal- recessive renal disorder (9,10). Renal histology in these ani- mals showed a tubular atrophy associated with interstitial nephritis, pointing to an additional role of paracellin-1 for renal Received May 25, 2001. Accepted July 10, 2001. Correspondence to Dr. Martin Konrad, Department of Pediatrics, Philipps epithelial integrity. University Marburg, Deutschhausstrasse 12, D-35037 Marburg, Germany. To provide the basis for further studies of paracellin-1 in Phone: 49-6421-2862789; Fax: 49-6421-2865724; E-mail: konradm@ suitable animal models we performed a cloning strategy of the mailer.uni-marburg.de homologous genes in mouse and rat. We here present the S.W. and K.P.S. contributed equally to this work. cDNA sequences of mouse and rat Pcln1, the murine genomic 1046-6673/1212-2664 Journal of the American Society of Nephrology structure, and the results of RNA and protein expression Copyright © 2001 by the American Society of Nephrology studies. J Am Soc Nephrol 12: 2664–2672, 2001 Paracellin-1 in Mouse and Rat 2665

Materials and Methods BLAST programs that are supplied by NCBI and Infobiogen web sites Cloning of Mouse and Rat Pcln1 cDNA (http://www.ncbi.nlm.nih.gov; http://www.infobiogen.fr). Protein pat- To obtain the cDNA sequences of mouse and rat Pcln1, a search tern and statistical protein structure analysis were performed by using with the coding sequence of human PCLN1 was performed in mouse the PROSITE protein pattern search tools (www.expasy.ch/prosite). and rat EST databases, which yielded five different mouse EST clones (EMBL/GenBank/DDBJ Accession no. BB499021, BB501602, BB Determination of Exon-Intron Boundaries of Mouse 0496375, BB498218, and AV380228) sequenced from the 3'-end Pcln1 corresponding to bases 790 to 989 of the human PCLN1 cDNA To determine the exon-intron boundaries of murine introns two, (EMBL/GenBank/DDBJ Accession no. AF152101) and two different three, and four, exon-specific primers were designed and the inter- rat EST clones comprising the 3'-end of rat Pcln1 (EMBL/GenBank/ vening intron sequences amplified from mouse genomic DNA with DDBJ Accession no. AI412107 and AW142781). the Expand Long Template PCR System (Roche Diagnostics GmbH, By 5'–RACE-PCR (rapid amplification of cDNA ends) the lacking Mannheim, Germany). The fragments were run on an agarose gel, and 5'-region was amplified from mouse kidney Marathon Ready cDNA the approximate sizes were determined by using a half-logarithmic (Clontech, Palo Alto, CA) with the Advantage II Polymerase Mix function. The exon-intron boundaries of these introns were sequenced (Clontech) by using the mouse gene–specific primer mmGSP1 (for all directly with exon-specific primers. primer sequences see Table 1). A Southern blot procedure was used to The exon-intron boundaries of intron one were determined by using identify the correct PCR-fragment (ECL-3'–oligolabeling and detec- the Mouse Genome Walker Kit (Clontech). The nested PCR reaction tion systems; Amersham Pharmacia Biotech, Uppsala, Sweden) (re- yielded a single fragment that was subcloned into the pCR2.1-TOPO sults not shown). A mouse gene–specific primer, mmGSP2, was vector and sequenced from both strands. Alignment with human designed from the 5'-RACE product, and an additional 3'-RACE-PCR PCLN1 was performed with the CLUSTALW-program (http://clust- performed to obtain the complete full-length downstream sequence. alw.genome.ad.jp). The human gene structure was obtained by search The 784-bp and 485-bp PCR fragments were agarose gel purified, in human genomic databases with the human cDNA sequence and by subcloned into a pCR2.1-TOPO vector (Invitrogen, Groningen, The location of the human PCLN1 gene in GenBank contigs AC009520 Netherlands), and sequenced from both strands by using the ABI and AC073963. PRISM 310 Genetic Analyzer (Applera, Norwalk, CT). The 5'– and 3'–RACE-PCR from rat kidney cDNA were performed Chromosomal Assignment of Mouse Pcln1 analogously with rat gene–specific primers rnGSP1 and rnGSP2, The genetic map position of mouse Pcln1 was determined by respectively. The 814-bp and 871-bp PCR fragments were subcloned segregation analysis by using gene-specific amplification of clone and sequenced as described above. DNA from the Whitehead Institute Center for Genomic Research The cDNA sequence of full-length murine Pcln1 was deposited mouse radiation hybrid mapping panel (11). The primer pair mm-Int4- with GenBank accession no. AF323748 and the full-length rat cDNA F/R spanning intron 4 of mouse Pcln1 was used for amplification sequence with accession no. AF333099. (fragment size, Ϸ1350 bp). Data vectors, which were based on two independent PCR analyses of the entire panel, with data arranged in Analysis of cDNA and Amino Acid Sequences the order specified for the Whitehead Institute/MIT Center for Ge- The final cDNA and amino acid sequences obtained were com- nome Research, Mouse EST RH Mapping Project, Public Data Re- pared with human and bovine PCLN1 by using the FASTA and lease 3 (April, 2000) were submitted to two-point maximum-likeli-

Table 1. Oligonucleotide sequences used for PCR, rapid amplification of cDNA ends–PCR, and reverse transcriptase–PCR

Primer Designation Sequence Product Length (bp)

mmGSP 1 5Ј-GGGTAGTTCCTCTCAGGCCCAACATCT-3Ј 784 mmGSP 2 5Ј-CTGTGGATGTTTACGTCGAACGCTCCT-3Ј 485 rnGSP 1 5Ј-AGGAGCGTTCGACGTAAACATCCACAG-3Ј 814 rnGSP 2 5Ј-TTGACTGCGTGAAGTTCCTACCGGATG-3Ј 871 mm-Int1F-GW-1 5Ј-TCTTCTTCAGTACGCTGCCTGCTTCT-3Ј mm-Int1F-GW-2 5Ј-TTGGCCATATTCTCCACTGGGTTTTT-3Ј mm-Int1R-GW-1 5Ј-TATGGAGTCGTACTCATCGCAGGTTC-3Ј mm-Int1R-GW-2 5Ј-GAATCCCATCAAAAGCGTTTGTTACAC-3Ј mm-Int2-F 5Ј-ACCTGCGATGAGTACGACTCC-3ЈϷ3300 mm-Int2-R 5Ј-GATCATCAGTGCTCGAGTTACCAC-3Ј mm-Int3-F 5Ј-CTTTGTTGCAGGGACCACATTACTC-3ЈϷ3650 mm-Int3-R 5Ј-AGCATACCACACAGAACCGATGATT-3Ј mm-Int4-F 5Ј-CTCACCTGCTGTTTGTACCTCTTC-3ЈϷ1350 mm-Int4-R 5Ј-TGCAGTTGAATAGGGCTTCC-3Ј rn-PC-RT-F 5Ј-TTGTTGCAGGGACCGTATTACTCA-3Ј 248 rn-PC-RT-R 5Ј-GGGTAGTTCCTCTCAGGTCCAACA-3Ј rn-␤A-RT-F 5Ј-GAGTACAACCTCCTTGCAGCTC-3Ј 329 rn-␤A-RT-R 5Ј-TTGTAGAAAGTGTGGTGCCAAA-3Ј 2666 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2664–2672, 2001

hood analysis (http://www.genome.wi.mit.edu/cgi-bin/mouse_rh/ Solubilization of Rat Kidney Proteins rhmap-auto/rhmapper.cgi). Mouse-human-synteny was analyzed by Rat renal tissue was disrupted in ice-cold 50 mM Tris-HCl, pH 7.4, using NCBI and MGI mouse-human homology databases (http:// that contained 1% Triton X-100, 0.5% Na-deoxycholacid, 0.1% SDS, www..ncbi.nlm.nih.gov:80/Homology/mouse16.html; 5 mM pefablock, 25 ␮M pepstatin A, 10 ␮M leupeptin, 1mM Na- http://www.informatics.jax.org/searches/linkmap.cgi). orthovanadat, 10% glycerol, and 2 mM ethylenediaminetetraacetic acid by the use of an ultra turax device and a Potter-Elvehjem Northern Blot Analysis of Mouse Pcln1 Transcription homogenizer. Final homogenization was achieved by ultrasound. The Poly(A)ϩ RNA from eight adult mouse tissues was studied by resulting homogenate was centrifuged at 10,000 g for 30 min to multiple-tissue Northern analysis (Clontech). Hybridization was car- separate undisrupted tissue. The supernatant containing solubilized Ϫ ried out overnight at 55°C by using 10 ng of a PCR-amplified proteins was stored at 80°C until use. Protein concentration was fragment comprising the full-length mouse Pcln1 coding sequence, determined with bicinchoninic acid according to the protocol of the radiolabeled with Redivue 32P (Rediprime II random prime labeling supplier (Pierce, Rockford, IL) with BSA as protein standard. system; Amersham Pharmacia Biotech, Buckinghamshire, UK). Rap- id-hyb buffer (5 ml) (Amersham Pharmacia Biotech) was used for Sodium Dodecyl Sulfate–Polyacrylamide Gel hybridization. The membrane was then washed four times with de- Electrophoresis and Immunoblot Assay creasing SSC concentrations (2X/0.1% sodium dodecyl sulfate Protein samples (100 ␮g/lane) were run on 4 to 20% polyacryl- [SDS]–0.1X/0.1% SDS) and increasing temperatures (from room amide gradient gels (GradiGels, Gradipore, North Ryde, Australia) temperature up to 65°C). The membrane was autoradiographed at and then transferred onto a nitrocellulose membrane by electroblotting Ϫ70°C for 72 h by using two intensifying screens and a BioMax with a discontinuous buffer system. The membrane was blocked with MS-film (Eastman Kodak, Rochester, NY). A positive control (dot 5% nonfat dry milk in Tris-buffered saline (TBS; 25 mM Tris, 150 blot with linearized and denatured plasmid DNA containing the mu- mM NaCl, pH 7.5) at 4°C overnight and then incubated in 5% normal rine Pcln1 full-length coding sequence) and a negative control (dot goat serum for 30 min. After washing with 0.1% Tween 20/TBS blot with linearized and denatured plasmid DNA containing the mu- (TTBS), a polyclonal antibody directed against paracellin-1 peptide rine Cldn1 full-length sequence encoding claudin-1) were hybridized (1:1000 in 5% goat serum/TTBS) was applied for4hatroom and washed in the same hybridization bottle. Thereafter, the Northern temperature. This antibody was raised in rabbits against the C-termi- blot was probed with a Rediprime II random prime labeled human nal peptide 282-VSMAKSYSAPRTETAKMYAVD-302 of human ␤-actin cDNA probe (PCR fragment supplied by the manufacturer) to paracellin-1 coupled to keyhole limpet hemocyanin. Peptide sequence compare loading in each lane. homology to rat paracellin-1 was 95% (20/21). No sequence homol- ogy to mouse and rat claudin-1 (GenBank Accession BC002003 and Microdissection of Rat Nephron Segments AF195500, respectively) and mouse claudin-2 (GenBank Accession AF072128) was observed. For detection of bound antibodies, a goat Kidneys from healthy male Wistar rats (180 to 200 g; M & B, Eiby, anti-rabbit IgG peroxidase conjugate (1:7500 in 5% goat serum/ Denmark) were perfused with saline via the abdominal aorta before TTBS) was added to the membranes and incubated at room temper- perfusion with collagenase A in Dulbecco’s modified Eagle’s medium ature for 45 min. After four wash cycles with TTBS, visualization was (Life Technologies, Tåstrup, Denmark) with 0.1% bovine serum al- achieved by enhanced chemiluminescence technique (Amersham bumin, 1 ml/L penicillin-streptomycin solution (Sigma Chemical, Pharmacia Biotech) with exposure times of 30 s to an autoradio- Taufkirchen, Germany), and 660 U/L insulin. Kidney slices were graphic film (Hyperfilm ECL; Amersham Pharmacia Biotech). In a further incubated for 25 min at 37°C with gentle shaking. The slices control experiment, the primary antibody was omitted. In this exper- were moved to Dulbecco’s modified Eagle’s medium that contained iment, no protein bands were visible after the staining procedure (not 10% fetal calf serum, 1 ml/L penicillin-streptomycin solution, and 660 shown). U/L insulin. Tubules were dissected on ice by using a Leica MZ 12.5 microscope (Leica, Bensheim, Germany). Immunohistochemistry ␮ RNA Extraction Rat kidneys were fixed in 4% paraformaline and 7- m-thick sec- tions were cut on a microtome (Rotationsmikrotom 3455 Leitz; From the dissected rat nephron segments, mRNA was extracted Leica). For staining of paracellin-1 protein with the polyclonal anti- from single segments of approximately 1 mm in length. Segments body described above the alkaline phosphatase anti-alkaline phospha- were not pooled, and mRNA was extracted by using poly-T coated tase technique was used as described earlier (12). The specificity of magnetic beads (Dynabeads mRNA DIRECT Micro kit; Dynal, Oslo, immunolabeling was tested by omitting the primary antibody. Norway). Results Reverse Transcriptase–PCR Isolation and Characterization of the Mouse Pcln1 Reverse transcriptase–PCR (RT-PCR) was performed on mRNA Gene and Sequence Comparison of Mouse, Rat, Cattle, (directly on the magnetic beads) by using SuperscriptII (Life Tech- and Human Paracellin-1 nologies). RT negative control reactions were performed on half of the Screening of EST databases with human PCLN1 (GenBank extracted mRNA. A fluorescence multiplex PCR (30 cycles) was accession no. AF152101) revealed partial Pcln1 cDNA clones carried out by using sequence-specific primers for rat Pcln1 (rn-PC- RT-F/R) and ␤-actin (rn-␤A-RT-F/R). ␤-actin was included to con- for both mouse and rat. 5'- and 3'-RACE procedures yielded the firm the presence of cDNA, and a PCR negative control reaction was full-length cDNAs for mouse and rat Pcln1 of 939 and 1514 bp performed. The forward primers were fluorescence labeled. PCR in length, respectively. products were analyzed by capillary electrophoresis by using the ABI Like human PCLN1, mouse Pcln1 comprises five coding PRISM 3100 Genetic Analyzer (Applera). exons. Analysis of the murine exon-intron boundaries revealed J Am Soc Nephrol 12: 2664–2672, 2001 Paracellin-1 in Mouse and Rat 2667 identical coding exon sizes for exons 2 to 5. The adjacent termini for both mouse and rat paracellin-1. Statistical protein intronic sequences show high similarity between the murine analysis revealed three putative protein kinase C phosphoryla- and human genes (Figure 1). tion sites in mouse and rat paracellin-1 (AA position 40, 217, Both mouse and rat cDNA sequences share a suitable trans- and 225) and a C-terminal microbody targeting signal (TRV, lational start site that corresponds to Met71 of human paracel- AA postion 233 to 235). Sequence comparison of rodent para- lin-1 (GenBank accession no. AAD43096), but they lack a cellin-1 in protein databases yielded similarity with a great further upstream start site. Therefore, both rodent Pcln1 genes number of claudins of different species and also to hypothetical encode a protein of 235 amino acids with a predicted molecular proteins of Caenorhabditis elegans (ZK563.4; GenBank acces- weight of 26 kD, lacking the first 70 amino acids when com- sion no. AAA81150) and Drosophila melanogaster (CG6398; pared with the human amino acid sequence. Although the GenBank accession no. AAF48766) (UCSC Computational nucleotide among rodent and human Biology Target99 alignment (13)). PCLN1 downstream from this suitable translational start site is very high, the upstream 5' rodent sequences differ to a great Chromosomal Assignment of Mouse Pcln1 and Mouse/ extent when compared with the human 5' sequence. On the Human Synteny Map amino acid level, mouse and rat paracellin-1 (Met1-Val235) Radiation hybrid panel analysis for mouse Pcln1 indicated a show 99% homology to each other and 92% and 91% to the direct linkage to marker D16Mit133 on mouse chromosome human homologue (Met71-Val305), respectively (Figure 2A). 16q. D16Mit133 maps 20.2 cM distal to the centromere of Bovine paracellin-1 is found as a 235–amino acid transcript mouse chromosome 16. As shown in Figure 3, this mouse and a longer 254–amino acid transcript in the GenBank with chromosomal region is syntenic to human chromosome 3q27- accession no. BAA87045 and BAA82553, respectively. Mouse q28, which harbors the human PCLN1 gene. and rat paracellin-1 are both 88% homologous to the 235- amino acid transcript. Tissue Expression of Mouse Pcln1 Statistical Analyses Northern blot analysis including eight murine tissues re- In analogy to the human protein, hydrophilicity plots predict vealed that Pcln1 is expressed exclusively in the kidney as a four transmembrane segments and intracellular N- and C- 1.2-kb and an alternative 1.0-kb transcript (Figure 4). The

Figure 1. Genomic organization of the murine Pcln1 gene. The comparison with the human gene demonstrates highly conserved splice-sites. Human PCLN1 intron sizes were taken from two genomic clones (GenBank accession no. AC009520 and AC073963), which harbor PCLN1. Mouse Pcln1 intron sizes were calculated after PCR amplification and separation by agarose gel electrophoresis by applying a half-logarithmic function. 2668 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2664–2672, 2001

Figure 2. (A) Amino acid sequence alignment of human (hs), bovine (bt), mouse (mm), and rat (rn) paracellin-1 as predicted from the cDNA sequences. Human methionine 1 (Met1) and Met71 (GenBank accession no. AAD43096) are depicted by black boxes. Met1 of mouse and rat paracellin-1 (GenBank accession no. AAK49518 and AAK52459, respectively) correspond to human Met71 and bovine Met1 (GenBank accession no. BAA87045). Identical amino acids are shaded in gray. (B) Putative protein structure model based on hydrophilicity analysis comparing rodent, bovine, and human amino acid sequences. Rodent paracellin-1 is predicted to be a transmembrane protein with a short intracellular N-terminus, a long intracellular C-terminus, and two extracellular loops. Black circles represent identical amino acids among mouse, rat, cattle, and human. The light gray circles of the long N-terminus demonstrate the predicted structure of human paracellin-1 as reported in the GenBank with accession no. AF152101. However, sequence comparisons among the different species support the theory that human paracellin-1 shares the short N-terminus with both rodent and the bovine protein. (C) Model of the thick ascending limp of Henle’s loop (TALH). The paracellular transport of magnesium and calcium is mediated by the tight junction protein complex at specialized cell-cell contact sites. Tight junctions seal the intercellular space but also confer selectivity to the transepithelial flux of ions and molecules. Human paracellin-1 has been localized to the tight junction complex of the TALH and has been identified to be essential for renal magnesium and calcium reabsorption in this tubular segment (4). positive control with blotted Pcln1 single-stranded DNA also tively in the cortical and medullary TALH and in the cortical gave a strong signal whereas the negative Cldn1 control did not and outer medullary collecting duct (CD) (Figure 5). No signal hybridize with the radiolabeled probe. was seen in the proximal convoluted and proximal straight tubule. Renal Expression of Rat Paracellin-1 The expression and distribution of the gene product paracel- Fluorescence RT-PCR on microdissected rat nephron seg- lin-1 was studied in rat kidney by using a polyclonal paracel- ments with rat-specific primers showed Pcln1 mRNA distinc- lin-1 antibody. Immunoblot analysis of a rat kidney preparation J Am Soc Nephrol 12: 2664–2672, 2001 Paracellin-1 in Mouse and Rat 2669

Figure 3. Mouse/human synteny map. Comparison of radiation hybrid mapping of mouse Pcln1 on mouse Chr16q with the syntenic region on human Chr3q. The relative genetic (gen) positions of the murine genes Fgf12, Zfp148, and Apod surrounding Pcln1 on mouse Chr16 are compared with the gen, cytogenetic (cyt), and physical (phys) positions of the homologous human genes FGF12, ZNF148, and Figure 4. Northern blot analysis of mouse Pcln1 expression with APOD on Chr3 (Mouse Genome Informatics, The Jackson Labora- mRNA from eight adult mouse tissues. A mouse full-length Pcln1 tory, http://www.informatics.jax.org/searches/linkmap.cgi; Ensembl coding sequence fragment was used as a probe (A). The arrows Server, http://www.ensembl.org; NCBI Map Viewer, indicate two renal transcripts of 1.2 and 1.0 kb for Pcln1. Hybridiza- http://www.ncbi.nlm.nih.gov/cgi-bin/Entrez/maps.cgi). On mouse tion with a radiolabeled ␤-actin probe served as control for mRNA Chr16, Pcln1 maps directly to marker D16Mit133 20.2 cM distal to loading of each lane (B). For positive and negative controls, dot blots the centromere. with linearized and denatured plasmid DNA containing the full-length Pcln1 and Cldn1 coding sequences were hybridized together with the multiple-tissue blot (C). The molecular size markers (kb) are depicted for paracellin-1 expression revealed a strong band at 50 kD and at the side of the membrane. a second but weaker protein band with an apparent molecular weight of 35 kD (Figure 6A). In rat kidney sections, paracel- lin-1 expression in the TALH and in the distal tubule (DT) was consensus sequence, which determines the translational start detected (Figure 6, B and C). Some staining was also observed point (14). For human paracellin-1, Simon et al. (4) already in the CD, consistent with the results of RT-PCR analysis of rat discussed the possibility of the second methionine in-frame nephron segments. No staining was seen in glomeruli and in (Met71) to be the original translational start site because it is proximal nephron segments. analogous to the start sites of other claudins. Comparing hu- man and bovine PCLN1, both share only one start codon Discussion in-frame encoding human Met71 and bovine Met1 (GenBank The recent identification of paracellin-1 as a new member of accession no. AB035210), before a downstream sequence of the claudin tight junction protein family and its involvement in high homology (91% on the amino acid level). a hereditary renal disease characterized by substantial loss of Genetic analysis of PCLN1 in humans revealed a polymor- magnesium and calcium and deterioration of renal function phism at amino acid position 55 that would result in a preter- promoted scientific interest in paracellin-1 expression studies. minal translation stop and consecutively to a truncated para- We here present the cloning of mouse and rat Pcln1 and the cellin-1 protein. This polymorphism is frequently found in results of mRNA and protein expression analysis. healthy individuals (8); therefore, it is suggested that the un- As it was shown for human and bovine PCLN1, mouse and derlying nucleotide sequence is not coding. rat Pcln1 are also highly homologous, which indicates evolu- The sequence data of mouse and rat Pcln1 provide further tionary sequence conservation. Homology is not only seen on arguments for paracellin-1 being shorter than reported in the the amino acid level but also on the DNA level including GenBank submission AF152101, as both lack a methionine exon-intron boundaries. However, all four species differ con- that corresponds to human Met1 but share a methionine that siderably in their 5'-UTR. Both human and bovine PCLN1 corresponds to human Met71 and bovine Met1 (GenBank have more than one in-frame start codon in a suitable Kozak accession no. AB035210). The similarity of the downstream 2670 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2664–2672, 2001

A kidney-specific expression of PCLN1 was also observed by Northern blot analysis in humans (4). These findings cor- relate with the renal-specific phenotype of the majority of FHHNC patients. Nevertheless, a subgroup of FHHNC patients also presents with ocular symptoms such as horizontal nystag- mus, severe myopia, or congenital coloboma (15,16), and it still remains to be studied whether defects in paracellin-1 are also responsible for these ocular abnormalities. For bovine PCLN1, expression was observed in the kidney but also to a minor extent in the lung (9). However, cattle with complete PCLN1 deletions do not show pathophysiologic changes in their lungs (9), and the role of bovine PCLN1 for respiratory function is unknown. No pulmonary PCLN1 tran- script was found in human and murine Northern blot analysis. Immunoblotting of whole rat kidney preparations with a polyclonal antibody directed against a C-terminal peptide of the paracellin-1 protein demonstrated two specific protein bands with approximate molecular weights of 50 kD and 35 kD, respectively. In contrast, Western blot analysis of human paracellin-1 revealed a single band with an apparent molecular mass of 36 kD (4). However, this result was observed after overexpression of human paracellin-1 (hsMet1-Val305) in bac- teria. The 50-kD band found in rat kidney corresponds to the doubled predicted molecular weight of 26 kD for rat paracel- lin-1 (rnMet1-Val235) and might represent paracellin-1 dimers. Expression of paracellin-1 along the nephron studied by RT-PCR and immunohistochemical analysis was observed in cortical and medullary TALH, DT, and CD. Previous immu- Figure 5. Fluorescent multiplex reverse transcriptase–PCR on dis- nohistochemical studies of paracellin-1 in human kidney dem- sected rat kidney tubule segments analyzed by capillary electrophore- onstrated expression predominantly in the TALH (4). In all sis. Fluorescent peaks at 248 bp denote Pcln1, and peaks at 329 bp species examined so far, TALH expression of paracellin-1 is denote ␤-actin fragment amplification. Pcln1 mRNA was detected in confirmed. This expression pattern is in accordance with the medullary and cortical TALH (mTALH and cTALH, respectively), cortical and outer medullary collecting duct (CCD and OMCD, re- phenotype of FHHNC-affected patients who present with a spectively). No Pcln1 mRNA expression was observed in proximal dysfunction of the paracellular reabsorption of divalent cations convoluted or proximal straight tubule (PCT and PST, respectively). in this nephron segment. ␤-actin was present in all samples, which confirms the presence of However, there are differences between species concerning cDNA, except in the PCR negative control (H2O), which was blank paracellin-1 expression in the CD. In contrast to RT-PCR and for both Pcln1 and ␤-actin. RT negatives were also blank (data not immunohistochemistry on rat kidney, no CD signal was seen in shown). Please note the difference in scale intensity. RT-PCR on rabbit nephron segments, and paracellin-1 expres- sion has not been described so far in the CD of human kidney PCLN1 sequence is extremely high among all four species. In (4). One might speculate that there is a difference in the addition, the multiple alignment of paracellin-1 and other extension of the expression pattern with respect to more distal members of the claudin family demonstrates that all claudins tubule segments among different species. As there is no major share the short cytoplasmic N-terminus in protein prediction paracellular transport of magnesium and calcium in the distal models with rodent paracellin-1 (1,2). convoluted tubule and CD, the maintenance of renal ion trans- Tissue-specific transcription of murine Pcln1 was studied by port mechanisms does probably not depend on paracellin-1 multiple-tissue Northern blot analysis, which revealed expres- expression in these segments. On the other hand, the phenotype sion exclusively in the kidney. Interestingly, two transcripts of of FHHNC-affected patients points to a possible role of para- 1.2 kb and 1.0 kb were detected. As the murine Pcln1 cDNA cellin-1 for tubular integrity as these patients frequently de- sequence is 939 bp in length without polyadenylation, the 1.2 velop end-stage renal disease that is histologically character- kb transcript is considered to represent 3'-polyadenylated full- ized by renal calcium deposits, interstitial inflammation, length Pcln1-mRNA. The shorter fragment might be due to glomerular sclerosis, and tubular atrophy. For FHHNC pa- alternative splicing with a deletion of sequence information tients, loss of renal function has often been related to the and/or differences in 5' or 3' UTR length. It is less likely that progression of medullary nephrocalcinosis. On the other hand, the shorter transcript represents a highly homologous, yet various hereditary tubulopathies associated with severe unidentified, gene that is exclusively expressed in the kidney. nephrocalcinosis (e.g. hyperprostaglandin E syndrome/antena- J Am Soc Nephrol 12: 2664–2672, 2001 Paracellin-1 in Mouse and Rat 2671

Figure 6. (A) Immunoblot assay of paracellin-1 with solubilized rat kidney proteins using a polyclonal antibody directed against the C-terminus of paracellin-1. A prominent band for paracellin-1 was detected at the molecular weight of 50 kD, and a second weaker signal was detected at 35 kD. (B and C) Light micrographs from 7-␮m sections of rat kidney tissue. Immunostaining for paracellin-1 was observed in the TALH (B), the DT, and CD (C). There was no immunostaining of proximal tubular segments. Control experiments with competitive application of C-terminal paracellin-1 peptide resulted in complete inhibition of staining (data not shown). Magnifications: ϫ100 in B; ϫ400 in C. tal Bartter syndrome (17) and distal renal tubular acidosis (18)) junctions with no sequence similarity to occludin. J Cell Biol do not proceed to renal insufficiency. 141: 1539–1550, 1998 Loss of renal function with tubular atrophy and interstitial 2. Morita K, Furuse M, Fujimoto K, Tsukita S: Claudin multigene nephritis are also seen in Japanese black cattle affected by a family encoding four-transmembrane domain protein compo- complete loss of the PCLN1 gene. Taken together, these find- nents of tight junction strands. Proc Natl Acad Sci USA 96: ings support the theory that paracellin-1 might represent an 511–516, 1999 important structural protein beside its role for paracellular 3. Rahner C, Mitic LL, Anderson JM: Heterogeneity in expres- reabsorption of magnesium and calcium in the TALH. sion and subcellular localization of claudins 2, 3, 4, and 5 in the rat liver, pancreas, and gut. Gastroenterology 120: 411– In summary, the identification of the two rodent analogous 422, 2001 genes of human PCLN1 provides a useful tool for further 4. Simon DB, Lu Y, Choate KA, Velazquez H, Al-Sabban E, Praga studies on paracellin-1 and tight junction physiology in suitable M, Casari G, Bettinelli A, Colussi G, Rodriguez-Soriano J, animal models. Magnesium handling in the kidney is precisely McCredie D, Milford D, Sanjad S, Lifton RP: Paracellin-1, a regulated and paracellin-1 is the first protein identified to be renal tight junction protein required for paracellular Mg2ϩ re- essentially involved in renal magnesium transport mechanisms; sorption. Science 285: 103–106, 1999 therefore, these studies will be helpful for understanding mag- 5. Sirotkin H, Morrow B, Saint-Joire B, Puech A, Das Gupta R, nesium homeostasis. Furthermore, analysis of paracellin-1 in Patanjali SR, Skoultchi A, Weissman SM, Kucherlapati R: Iden- embryonal and fetal rodent tissues might give insight into tification, characterization, and precise mapping of human gene expression patterns during early developmental stages and encoding a novel membrane-spanning protein from 22q11 region enlighten its possible role for renal development and tubular deleted in velo-cardio-facial syndrome. Genomics 42: 245–251, integrity. 1997 The availability of the mouse gene structure will be valuable 6. Wilcox ER, Burton QL, Naz S, Riazuddin S, Smith TN, Ploplis for the generation of a mouse model for FHHNC by gene B, Belyantseva I, Ben-Yosef T, Liburd NA, Morell RJ, Kachar targeting in embryonic stem cells. Site-directed mutagenesis of B, Wu DK, Griffith AJ, Friedman TB: Mutations in the gene PCLN1 in a target vector might allow the investigation of encoding tight junction claudin-14 cause autosomal recessive phenotypic effects of single amino acid exchanges, truncated deafness DFNB29. Cell 104: 165–172, 2001 gene products, and complete loss of paracellin-1. 7. Weber S, Hoffmann K, Jeck N, Saar K, Boeswald M, Kuwertz- Broeking E, Meij IIC, Knoers NV, Cochat P, Sulakova T, Bonzel Acknowledgments KE, Soergel M, Manz F, Schaerer K, Seyberth HW, Reis A, Konrad M: Familial hypomagnesaemia with hypercalciuria and We thank Ulla Pechmann for excellent technical assistance. MK nephrocalcinosis maps to chromosome 3q27 and is associated was supported by the Deutsche Forschungsgemeinschaft (Ko1480/3 to 2). SW was supported by the Stiftung P. E. Kempkes (Kennziffer with mutations in the PCLN-1 gene. Eur J Hum Genet 8: 414– 30/99). 422, 2000 8. Weber S, Schneider L, Peters M, Misselwitz J, Ro¨nnefarth G, References Bo¨swald M, Bonzel KE, Seeman T, Sula´kova´ T, Kuwertz- 1. Furuse M, Fujita K, Hiiragi T, Fujimoto K, Tsukita S: Claudin-1 Bro¨king E, Gregoric A, Palcoux J-B, Tasic V, Manz F, Scha¨rer and -2: Novel integral membrane proteins localizing at tight K, Seyberth HW, Konrad M: Novel paracellin-1 mutations in 25 2672 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2664–2672, 2001

families with familial hypomagnesemia with hypercalciuria and 14. Kozak M: Interpreting cDNA sequences: Some insights from nephrocalcinosis. J Am Soc Nephrol 12: 1872–1881, 2001 studies on translation. Mamm Genome 7: 563–574, 1996 9. Hirano T, Kobayashi N, Itoh T, Takasuga A, Nakamaru T, 15. Praga M, Vara J, Gonzalez-Parra E, Andres A, Alamo C, Hirotsune S, Sugimoto Y: Null mutation of PCLN-1/claudin-16 Araque A, Ortiz A, Rodicio JL: Familial hypomagnesemia results in bovine chronic interstitial nephritis. Genome Res 10: with hypercalciuria and nephrocalcinosis. Kidney Int 47: 659–663, 2000 1419–1425, 1995 10. Ohba Y, Kitagawa H, Kitoh K, Sasaki Y, Takami M, Shinkai Y, 16. Benigno V, Canonica CS, Bettinelli A, von Vigier RO, Trutt- Kunieda T: A deletion of the paracellin-1 gene is responsible for mann AC, Bianchetti MG: Hypomagnesaemia-hypercalciuria- renal tubular dysplasia in cattle. Genomics 68: 229–236, 2000 nephrocalcinosis: A report of nine cases and a review. Nephrol 11. Van Etten WJ, Steen RG, Nguyen H, Castle AB, Slonim DK, Ge Dial Transplant 15: 605–610, 2000 B, Nusbaum C, Schuler GD, Lander ES, Hudson TJ: Radiation 17. Kockerling A, Reinalter SC, Seyberth HW: Impaired response to hybrid map of the mouse genome. Nat Genet 22: 384–387, 1999 furosemide in hyperprostaglandin E syndrome: Evidence for a 12. Komhoff M, Grone HJ, Klein T, Seyberth HW, Nusing RM: tubular defect in the loop of Henle. J Pediatr 129: 519–528, Localization of cyclooxygenase-1 and -2 in adult and fetal hu- 1996 man kidney: Implication for renal function. Am J Physiol 272: 18. Wrong O: Nephrocalcinosis. In: Oxford Textbook of Clinical F460–F468, 1997 Nephrology, 2nd Ed, edited by Davison AM, Gru¨nfeld JP, Kerr 13. Karplus K, Barrett C, Hughey R: Hidden Markov models for detecting DNS, Ritz E, Winearls CG, Oxford, Oxford University Press, remote protein homologies. Bioinformatics 14: 846–856, 1998 1998, pp 1376–1396