Congenital progressive hydronephrosis (cph) is … progressive hydronephrosis (cph) is caused by an S256L mutation in aquaporin-2 that affects its phosphorylation and apical membrane

Congenital progressive hydronephrosis (cph) is causedby an S256L mutation in aquaporin-2 that affects itsphosphorylation and apical membrane accumulationBradley W. McDill*, Song-Zhe Li*, Paul A. Kovach*, Li Ding†, and Feng Chen*‡

*Renal Division, Department of Internal Medicine, Department of Cell Biology and Physiology, and †Genome Sequencing Center, Washington UniversitySchool of Medicine, St. Louis, MO 63110

Communicated by Mario R. Capecchi, University of Utah, Salt Lake City, UT, March 14, 2006 (received for review December 20, 2005)

Congenital progressive hydronephrosis (cph) is a spontaneousrecessive mutation that causes severe hydronephrosis and obstruc-tive nephropathy in affected mice. The mutation has been mappedto the distal end of mouse chromosome 15, but the mutated genehas not been found. Here, we describe the identification of a singlebase pair change in aquaporin-2 (Aqp2) in cph mutants throughgenetic linkage mapping. The C-T change led to the substitution ofa Ser (S256) by a Leu in the cytoplasmic tail of the Aqp2 protein,preventing its phosphorylation at S256 and the subsequent accu-mulation of Aqp2 on the apical membrane of the collecting ductprincipal cells. The interference with normal trafficking of Aqp2 bythis mutation resulted in a severe urine concentration defect. cphhomozygotes demonstrated polydipsia and produced a copiousamount of hypotonic urine. The urine concentration defect couldnot be corrected by [deamino-Cys1,D-Arg8]-vasopressin (DDAVP, avasopressin analog), characteristic of nephrogenic diabetes insip-idus. The nephrogenic diabetes insipidus symptoms and the ab-sence of developmental defects in the pyeloureteral peristalticmachinery in the mutants before the onset of hydronephrosissuggest that the congenital obstructive nephropathy is most likelya result of the polyuria. This study has revealed the genetic basisfor the classical cph mutation and has provided direct geneticevidence that S256 in Aqp2 is indispensable for the apical accu-mulation, but not the general glycosylation or membrane associ-ation, of Aqp2.

nephrogenic diabetes insipitus � obstructive nephropathy � polarizedtrafficking

Congenital obstructive nephropathy is the most frequentcause of renal failure in infants and children (1). Antenatal

screening detects fetal hydronephrosis in 1 of 100 births, with atleast 20% being clinically significant (2). If left untreated,congenital obstructive nephropathy can cause severe renal fail-ure and death (1). Despite the impact of this devastating disease,the causes in most congenital obstructive cases in humans are notyet known. Besides environmental influence, there are clearlygenetic determinants in congenital obstructive nephropathy (3,4) (OMIM, Online Mendelian Inheritance in Man). Congenitalobstructive nephropathy has been described in a number ofanimal models with spontaneous mutations or targeted geneticmodifications (5). The genetic defects in most, if not all, of thenaturally occurring mutants are still undetermined, let alone themolecular and cellular mechanism by which the genetic defectscause congenital obstructive nephropathy.

Congenital progressive hydronephrosis (cph) was first discoveredas a spontaneous, autosomal recessive trait in C57BL�6J mice atThe Jackson Laboratory in the 1970s (6, 7). Because these mutantswere thought to have renal cysts, they were named juvenile poly-cystic kidney until a later study by Horton et al. (7) found that thecystic dysplasia was caused by progressive hydronephrosis resultingfrom urinary tract obstruction. Horton et al. (7) also mapped cph tothe distal part of the long arm of mouse chromosome 15. A roughchromosomal location of 57.8 cM was assigned to the cph locus by

Mouse Genome Informatics (MGI), largely based on the geneticmapping results from Horton et al. (7) and the relative positionsof the three markers used: caracul (Ca), underwhite (uw), andbelted (bt).

Here, we describe the identification of the cph mutation as asingle base change in codon 256 of aquaporin-2 (Aqp2). Thismutation causes a Ser to Leu substitution, loss of Aqp2 phosphor-ylation at amino acid 256, and the absence of apical accumulationof the protein. We have demonstrated that the mutants have noresponse to a vasopressin analog and produce large quantities ofhypotonic urine, characteristic of patients with nephrogenic diabe-tes insipidus (NDI) (8). The polyuria caused by the Aqp2 mutationin cph mutants likely overwhelms the pyeloureteral peristalticmachinery, resulting in the observed hydronephrosis, obstructivenephropathy, renal failure, and death. This study provides directgenetic evidence that phosphorylation of Aqp2 at S256 is essentialfor its apical membrane accumulation and water reabsorptionfunction in vivo.

ResultsThe cph Mutants Have Apparent Congenital Functional Obstruction ofthe Urinary Tract. The cph mutants appeared grossly normal at birthand made up 27.9% of the pups born in heterozygous intercrosses,very close to the 25% expected for an autosomal recessive mutationfollowing Mendelian inheritance. However, the cph mutants grewslowly and showed a significant size and weight difference frompostnatal day (P) 8 onward (Fig. 1 A and B). About 90% of the cphmutants died between 2 and 4 weeks of age. By 2 weeks, mostmutants also had visibly enlarged abdomens and appeared lethar-gic. Around 10% of the cph homozygotes survived past weaning,with the oldest cph homozygote living for 10 months. The adult cphhomozygotes are either infertile or have modestly reduced fertility.

Although younger mutants (�P14) had unilateral or bilateralhydronephrosis of various degrees, older mutants (�P14) almostalways had severe bilateral hydronephrosis (Fig. 1 C–E). Thisobservation is consistent with the findings in the Horton et al. (7)study but argues against the hypothesis that this is a model ofpolycystic kidney disease (6). However, the finding of hydroureter(arrow in Fig. 1D) and apparent obstruction at the ureterovesicaljunction (UVJ) in �25% of the mutants suggests that these mutantsare unlikely to have complete physical blockage at the ureteropelvicjunction (UPJ) level as proposed (7). Furthermore, we have notfound any correlation between the volume of the bladder and thegenotypes of the mice dissected, suggesting that the defect inthe mutants is likely in a location proximal to the bladder along theurinary tract. Interestingly, �66.3% of the mutants had more severe

Conflict of interest statement: No conflicts declared.

Abbreviations: cph, congenital progressive hydronephrosis; Aqp2, aquaporin-2; NDI, neph-rogenic diabetes insipitus; DDAVP, [deamino-Cys1,D-Arg8]-vasopressin; Pn, postnatal day n.

‡To whom correspondence should be addressed at: Department of Internal Medicine�Renal Division, Campus Box 8126, Washington University School of Medicine, St. Louis,MO 63110. E-mail: [email protected].

© 2006 by The National Academy of Sciences of the USA

6952–6957 � PNAS � May 2, 2006 � vol. 103 � no. 18 www.pnas.org�cgi�doi�10.1073�pnas.0602087103

defects in the right kidney at the time of examination, whereas only22.8% had more severe defects in the left and 10.9% had bothkidneys equally affected.

Blood urea nitrogen (BUN) levels in the mutants were signifi-cantly elevated (Fig. 1F), suggesting renal failure as a result ofobstructive nephropathy. The mutant kidneys had pathologicalchanges, characteristic of obstructive nephropathy, ranging fromparenchymal atrophy, erosion of the renal pelvis, expansion of thepelvicocaliceal space, and dilatation of the collecting ducts (Fig. 1G and H and data not shown). Molding polymers injected into thepelvicocaliceal space were able to travel along the urinary path tothe bladder in both the controls and mutants, although the mutanturinary path is distorted by the hydronephrosis, especially in thepelvicocaliceal space (Fig. 2 A and B). This result argues against theexistence of physical blockage from the kidney to the bladder andsuggests that hydronephrosis is likely caused by a functional ob-struction, namely urine retention in the urinary tract as a result ofa functionally, but not structurally, impaired downward urinetransport. Morphological examination of newborn mutants andtheir control littermates did not reveal any significant changes in thesmooth muscle cells that are important for contractability of thepyeloureteral complex (Fig. 2 C–F) or in the innervating nerves(Fig. 2 G and H) that modulate pyeloureteric peristalsis.

Genetic Linkage Mapping of the cph Locus. Because the cph mutationis in a pure C57BL�6J background, we outcrossed cph heterozy-gotes to three inbred strains (DBA�2J, AKR�J, and MOLD�RkJ)to bring in different genetic backgrounds for testing segregation andlinkage. By backcrossing aphenotypic F1s to confirmed cph het-

erozygotes, we identified heterozygous F1 mice based on theirability to produce mutants. Mapping was done primarily with F2sderived from intercrossing F1 heterozygotes. Previous genetic map-ping efforts using three classical genetic markers on chromosome15: Ca, uw, and bt suggested that the likely arrangement of themarkers is uw-bt-Ca-cph. The cph locus was tentatively assigned byMGI (Mouse Genome Informatics; www.informatics.jax.org) tomouse chromosome 15 at 57.8 cM distal to the Hoxc complex, basedlargely on these results (7). Due to the relatively low number ofinformative recombinations in the original study and the reposi-tioning of reference markers after the sequencing of the mousegenome, we began mapping with markers covering the distal �40Mbp of mouse chromosome 15, from marker D15Mit63 (�65.5Mbp) to the end of the chromosome (�104 Mbp). Both knownmicrosatellite markers found in public databases and novel onesdiscovered through our computational analyses were used. Afterscreening 618 mice representing 1,050 informative meioses, welocalized the mutation to a 0.7-Mbp chromosomal interval proximal(not distal) to the Hoxc complex and between the classical markersbt and Ca. This interval is defined by newly discovered microsat-ellite markers C15LD6 and C15LD5 at positions 99104970 and99804700 (Ensembl build 34), respectively (Fig. 3A and Fig. 7, whichis published as supporting information on the PNAS web site). Thisregion is syntenic to human chromosome 12q13.12.

Identification of the cph Mutation. There are 25 known and pre-dicted genes in the interval between C15LD6 and C15LD5 (Fig. 7,

Fig. 1. The cph mutants have apparent congenital obstruction at multiplelevels. (A) At 2 weeks of age, the mutants are significantly smaller than theircontrol littermates. (Scale bar: 1 cm.) (B) The mutants grow slower than thecontrols. Œ, controls; ■ , mutants. (C–E) The mutants have unilateral or bilateralhydronephrosis and hydroureter (arrow). (F) The blood urea nitrogen (BUN)level in the mutants (P5–P16) is 70.0 � 25.0 mg�dl (n � 35), significantly (P �0.001) higher than that of the controls (P5–P16, 26.8 � 5.0 mg�dl; n � 29). (Gand H) Kidney sections from P14 control and mutant littermates stained withhematoxylin�eosin.

Fig. 2. The cph mutants do not have complete physical obstruction or grossdevelopmental abnormalities in the smooth muscles and nerves along theurinary tract. (A and B) Resin moldings of the pelvicocaliceal space and uretericlumen at P5. (Scale bars: 1 mm.) (C–F) �SMA staining (green) of the smoothmuscles in the pelvic wall (arrows in C and D) and ureteric sections (E and F) atP6. (G and H) Nerve fibers (arrows) on the cross sections of the ureters from P6mice were revealed by the antineurofilament antibody (NF, green). The slideswere counterstained with DAPI to reveal the nuclei (blue).

McDill et al. PNAS � May 2, 2006 � vol. 103 � no. 18 � 6953

GEN

ETIC

S

based on NCBI Genome Mapview and Ensembl mouse genomeserver). At least five of these genes (Gpd1, ACCN2, Racgap1, Aqp5,and Prph1) were given low priority because their mutant allelesproduce distinctively different phenotypes from those observed incph mutants (MGI; www.informatics.jax.org). After careful reviewof the remaining genes, one of these genes, Aqp2, was given highpriority for further analysis due to its known expression in thekidney, its physiological function, and the observations that muta-tions in Aqp2 or its upstream regulator, arginine vasopressinreceptor 2 (AVPR2), cause NDI and kidney defects reminiscent ofobstructive nephropathy (9–11). To find a possible link betweenAqp2 and the cph mutation, we investigated the expression anddistribution of Aqp2 in kidney sections from control and mutantlittermates. Although Aqp2 was predominantly distributed on theapical membrane in the principal cells of the collecting duct in thecontrol samples, especially in the cortex and outer medulla, itsapical accumulation was completely lost at all levels of the collectingduct in all mutants studied (Fig. 3 B and C). There was also anapparent overall increase in Aqp2 levels in mutant collecting ducts.Furthermore, we examined the distribution of Aqp2 in the principal

cells of the Pax3Cre-Cnb1 mutants we have previously studied thatalso have severe congenital obstructive nephropathy (12). Thedistribution of Aqp2 was not changed in the Pax3Cre-Cnb1 mutantsdespite overt collecting duct dilatation (data not shown), suggestingthat the absence of Aqp2 apical accumulation was not simply asecondary result of the congenital obstructive nephropathy.

These findings prompted us to sequence all coding sequences andexon�intron junctions of the Aqp2 gene of the wild-type (WT), cphheterozygous, and cph homozygous mice. A mutation was found innucleotide 767 in the fourth exon of Aqp2 in the mutants (Fig. 3D).The C in codon 256 was changed to a T in the mutants, leading toa Ser to Leu substitution in the protein. This Ser is highly conservedin all vertebrate species analyzed (Table 1, which is published assupporting information on the PNAS web site). This mutationcorrelated perfectly with cph genotype and phenotype: all homozy-gous mutant samples sequenced were homozygous for the muta-tion, all WT mice in the same colony as well as all of the WT inbredmice from C57BL�6J, DBA�2J, AKR�J, and MOLD�RkJ lackedit, and all confirmed cph heterozygotes had both C and T in thisposition. These findings strongly suggest that the Aqp2-S256Lmutation is associated with the cph defects.

The S256L Mutation Affects the Expression of Aqp2 and Its Phosphor-ylation at S256. Previous studies have indicated that phosphoryla-tion of Aqp2-S256 is a key regulatory event in Aqp2 trafficking (13,14, 18–21). However, point mutations at S256 have not yet beenreported in human patients or animal models before this study. Tobetter understand the effects of the S256L mutation at the cellularand molecular levels in vivo, we examined the expression andphosphorylation of Aqp2 in kidney samples from control andmutant mice. RT-PCR using total RNA from whole kidneysdemonstrated a clear increase in Aqp2 transcripts in the mutantkidneys (Fig. 4A), likely as the result of a futile feed-back mecha-nism. Aqp2 protein levels were also increased in the mutants asshown in Western blotting assays (Fig. 4B). This finding is consistentwith the apparent increase of Aqp2 signal in mutant kidney sectionsimmunostained with the anti-Aqp2 antibody (Fig. 3C). This in-crease pertained to both the unglycosylated and glycosylated formsof Aqp2, suggesting that the S256L mutation does not affect theoverall glycosylation pattern of Aqp2. As expected, Aqp2 phos-phorylation at S256 was completely absent in the cph homozygotes,as shown by probing kidney extracts (Fig. 4C) and kidney sections(Fig. 4 D–G) with an antibody that specifically recognizes theS256-phosphorylated form of Aqp2 (14).

cph Heterozygotes Have an Intermediate Phenotype in Aqp2 Distri-bution and Aqp2-S256L Seems to Retain Membrane Association With-out S256 Phosphorylation. The identification of Aqp2-S256L in thecph mutants opens the opportunity to study the consequence of theloss of Aqp2 phosphorylation in vivo. Aqp2-S256L seems to retainits ability to localize to the plasma membrane as revealed byconfocal images of overlapping staining of Aqp2-S256L and E-Cadherin (expressed on the basolateral membrane in the collectingduct principal cells) in the cph mutants (Fig. 5 A–I). These obser-vations suggest that the phosphorylation of S256 is critical for apicalaccumulation but not membrane association of Aqp2. Whereas theoverall pattern of Aqp2 expression in the heterozygotes is similar tothat of the WT mice, the heterozygotes tend to have a higher degreeof basolateral distribution of Aqp2 compared with the WTs,apparently an indication of misrouting of mutant tetramers orchimeric tetramers with both WT and mutant subunits (Fig. 5 A–I).

Besides the vasopressin-regulated Aqp2, two other non-vasopressin-sensitive aquaporins, Aqp3 and Aqp4, are expressed inthe principal cells of the collecting ducts. Although Aqp2 ispredominantly expressed on the apical membrane, Aqp3 and Aqp4are exclusively located on the basolateral membranes (15). Confocalimages of immunostained kidney sections from WT and cph mutantmice showed that the expression level and the basolateral localiza-

Fig. 3. Genetic linkage mapping and positional cloning of cph. (A) Physicalmap of relevant portions of mouse chromosome 15. Our genetic mappingefforts locate the cph locus to the chromosomal interval of �0.7 Mbp, definedby the microsatellite markers C15LD6 and C15LD5. The black triangle indicatesthe chromosomal location of Aqp2. (B and C) Immunostaining of the collect-ing ducts in the outer medulla shows the apical accumulation of Aqp2 in thecontrols (B) but a diffuse staining with no apical accumulation in the mutant(C). (D) Sequence of the fourth exon of Aqp2 reveals the C-T substitution atnucleotide 767 in the homozygous mutants, whereas the heterozygotes haveboth C and T represented at position 767. This substitution results in a Ser toLeu change at amino acid 256 in the cytoplasmic tail of the Aqp2 protein. �,WT allele; c, cph mutant allele.

6954 � www.pnas.org�cgi�doi�10.1073�pnas.0602087103 McDill et al.

tion of Aqp3 and Aqp4 are not changed in the mutants (Fig. 5 J–R),suggesting that the observed loss of Aqp2 apical accumulation is notcaused by general alteration of cell polarity. These results alsosuggest that the three aquaporins in the collecting ducts areregulated independently without any existing mechanism of com-pensatory coordination.

cph Mutants Have both NDI Symptoms and Obstructive Nephropathy.The discovery of the Aqp2-S256L mutation suggests the possibilitythat the congenital progressive hydronephrosis in cph homozygotescould be a result of NDI symptoms, especially polyuria (16). Tostudy this possibility further, we first showed that the cph homozy-gotes failed to concentrate urine as effectively as their WT andheterozygous littermates (Fig. 6A). The failure of urine concentra-tion directly led to higher serum osmolality, a sign of dehydration,in the homozygotes (Fig. 6B). We then subjected these mice to 24-hurine collections in metabolic cages. As expected, the homozygoteshad a much higher urine output than their WT and heterozygouslittermates (Fig. 6C). Although the administration of [deamino-Cys1,D-Arg8]-vasopressin (DDAVP), a synthetic agonist for argi-nine vasopressin receptor 2 (AVPR2), was effective in stimulatingurine concentration in the WT and heterozygous mice, it had no

effects on the homozygotes (Fig. 6D). These findings are consistentwith the hypothesis that the Aqp2-S256L mutation in the cphmutants interrupts the water reabsorption function of Aqp2 andleads to polyuria as well as obstructive nephropathy. Although thewater reabsorption defect in the homozygotes is prominent, therewere no significant differences in excretion of key electrolytes, suchas Na and K, as indicated by urinary sodium�creatinine andpotassium�creatinine ratios (data not shown).

DiscussionThe nature of the cph mutation had been elusive for over threedecades since its discovery in the early 1970s (6). In this study, weused genetic linkage mapping to localize cph to a small chromo-somal interval and identified an S256L mutation in Aqp2 thatcorrelates with the genotypes and phenotypes of the cph mice.

Fig. 4. Absence of S256 phosphorylation and increased Aqp2 expression inthe cph mutants. (A) RT-PCR of total RNA from whole kidneys revealed anincrease of Aqp2 transcripts in the mutants (c�c). (B) Western blot of totalAqp2 protein from control and mutant kidneys. G-Aqp2 indicates the variousforms of glycosylated Aqp2. *, a band that cross-reacts with the secondaryantibody alone and is deemed unrelated to Aqp2. Aqp2 protein levels in themutant samples are elevated. (C) Western blot of proteins from whole kidneyextracts probed with an Aqp2-S256 phosphorylation-specific antibody. Themutant samples (c�c) lost the S256 phosphorylation on Aqp2. (D–G) Immuno-staining of the outer medullary collecting ducts. The Aqp2-S256 phospho-specific antibody (pAqp2) revealed apical distribution of the S256-phosphorylated Aqp2 (red) in the WT principal cells labeled by Dolichosbiflorus agglutinin (green) (F), closely resembling the Aqp2 distribution shownby the nonphospho-specific Aqp2 antibody staining (red) on adjacent sections(D). The mutant (c�c), however, has a complete absence of phospho-specificAqp2 staining (G), but apparently overexpresses the Aqp2-S256L that lacks theS256 phosphorylation (E).

Fig. 5. Aqp2-S256L has membrane association but lacks apical accumulation.Paraffin sections of the outer medulla from P1 mice of WT (A–C), heterozygous(D–F), and homozygous (G–I) littermates were stained with an Aqp2 antibody(A, D, and G) or an anti-E-Cadherin (E-Cad) antibody (B, E, and H). C, F, and Iare merged images of the first two channels. E-Cadherin is expressed on thebasolateral membrane whereas Aqp2 accumulates on the apical membrane ofthe principal cells in the WT. There is very little overlap between the Aqp2staining and the E-Cadherin staining. In the cph homozygotes, however,Aqp2-S256L does not accumulate on the apical membrane but shows a prom-inent basolateral distribution as revealed by the yellow signal produced by theoverlapping red Aqp2 and green E-Cadherin staining on basolateral mem-branes. Although the overall pattern of Aqp2 expression in the heterozygotesis similar to that of the WT mice, the heterozygotes show a higher degree ofbasolateral distribution of Aqp2 compared with the WT (A–I). In cph mutants,both Aqp3 and Aqp4 are expressed on the basolateral membrane and at levelssimilar to those seen in the WT mice (J–U).

McDill et al. PNAS � May 2, 2006 � vol. 103 � no. 18 � 6955

GEN

ETIC

S

Although some cases of congenital hydronephrosis are caused byphysical obstruction of the urinary tract, others are likely due tofunctionally impaired transport of urine from the kidney throughthe lower urinary tract to the outside environment. Besides neu-ronal and muscular abnormalities affecting the pyeloureteral peri-staltic machinery and urine transfer (5), polyuria, as a result ofdefective urine concentrating ability, can also overwhelm the urinetransport system, causing temporary or irreversible hydronephrosis(16, 17). The absence of physical blockage and developmentaldefects of the urinary tract (Fig. 2) are not surprising, consideringthe late onset of Aqp2 expression in embryogenesis and its principalcell-restricted distribution within the urinary system. The observedhydronephrosis in the cph mutants is apparently a result of theinability of the urine transport system to handle the dramaticallyincreased urine production in the mutants (Fig. 6).

Although original characterization of the cph mutants as a modelfor polycystic kidney disease was corrected by the study of Hortonet al. (7), our discovery of the Aqp2-S256L mutation and the findingof NDI defects in these mutants indicate that the obstructivenephropathy is accompanied by, and most likely secondary to, thedefective urine concentrating ability in these mice, analogous to theobservation of hydronephrosis in NDI patients with poorly con-trolled polyuria (16). These studies also underscore the fact thatkidney diseases in many cases have ambiguous pathological pre-sentations and require an understanding of the primary molecularand cellular defects to uncover the true nature of the condition.Hydronephrosis is a common prenatal diagnosis (1) that frequentlyresolves spontaneously. Because urine output is highly variable andinfluenced by multiple factors, it is conceivable that temporarypolyuria may contribute substantially to these spontaneously re-solved hydronephrosis cases.

Although the phosphorylation of Aqp2-S256 has been shown toaffect Aqp2 trafficking (14, 18–21), point mutation of S256 has notbeen reported in human patients or animals before this study (8).As expected, the S256L mutation in Aqp2 disrupts phosphorylationat S256 (Fig. 4C) and prevents the apical accumulation of Aqp2 inthe principal cells of the collecting duct (Fig. 3 B and C). In contrastto the intracellular vesicle retention observed in Aqp2-S256Amutants transfected into cultured cells (19), Aqp2-S256L seems tostill be able to localize to the cell membrane, although the polarizedtransport to the apical surface of the cell is disturbed (Fig. 5 A–I).These observations suggest that the phosphorylation of S256 iscritical for apical accumulation but not membrane association ofAqp2 in vivo. Although the expression level of Aqp2-S256L isincreased in the cph homozygotes, the glycosylation pattern of themutant protein appeared unchanged (Fig. 4B), suggesting that thephosphorylation of S256 may not be required for the glycosylationof Aqp2. Dileucine motifs have been indicated as a basolateraltrafficking signal (22). Recent studies have also suggested theinvolvement of single leucine residues in basolateral trafficking(23), although the requirement of optimal context has not beenruled out. Nevertheless, the possibility exists that L256 in themutant Aqp2 contributes to the shift from apical to basolateraldistribution due to the loss of S256 phosphorylation.

Impaired routing of WT AQP2 after tetramerization with mu-tant AQP2 has been indicated as the molecular basis for theautosomal dominant form of NDI (24, 25). In autosomal recessiveNDI, the absence of symptoms in heterozygotes may be due to thelack of interaction between mutant and WT Aqp2 subunits or theability of the apical sorting signal in the WT subunit to overridethe aberrant sorting signals (26) in interacting mutant subunits.Interestingly, whereas the overall pattern of Aqp2 expression in theheterozygotes is similar to that of the WT mice, the heterozygotes

Fig. 6. The obstructive nephropathy is likely induced by polyuria in the cph mutants. (A) cph mutants have defects in urine concentration. The urine osmolalityin adult mice is significantly lower in the cph homozygotes (480.8 � 258.0; n � 10) versus WT (2535.9 � 283.6; n � 8) and heterozygotes (2192.3 � 289.8; n �16). *, P � 0.001 compared with the other two groups. All measurements are in mOsm�kg unless otherwise stated. (B) Serum osmolality is significantly increasedin the homozygotes (391.9 � 76.7; n � 27) compared with the WT (278.8 � 6.7; n � 6) and heterozygous mice (287.6 � 16.4; n � 20). *, P � 0.001 compared withthe other two groups. (C) cph homozygotes have increased water input (21.6 � 9.4 ml; n � 11) and urine output (15.8 � 8.1 ml) than those of the WT (3.7 � 0.9ml water intake and 1.1 � 0.8 ml urine output, n � 9) and heterozygous (4.0 � 1.1 ml of water intake and 1.2 � 0.6 ml of urine output; n � 11) littermates. *,P � 0.001 compared with the other two groups. (D) DDAVP injection promotes urine concentration in the WT and heterozygotes but not in the cph homozygotes.The postinjection urine osmolality is increased from 2,385.3 � 265.7 (n � 4) and 2,127.3 � 228.2 (n � 11) to 3,030.0 � 554.7 and 2,571.4 � 343.3 for ��� andc�� mice respectively. *, P � 0.05 compared with measurements taken before injection. The osmolality of the post injection urine of the mutant is not significantlydifferent from that measured before the injection (from 614.7 � 222.5 to 584.3 � 260.3; n � 11). Bfr, before DDAVP injection; Aftr, after DDAVP injection.

6956 � www.pnas.org�cgi�doi�10.1073�pnas.0602087103 McDill et al.

tend to have a higher degree of basolateral distribution of Aqp2compared with the WTs (Fig. 5 A–I). Although the ability ofAqp2-S256L to form tetramers with WT Aqp2 subunits has notbeen directly investigated, previous in vitro studies have indicatedthat changes in S256 did not affect tetramerization (27). Therecessive inheritance of cph suggests that there are enough func-tional Aqp2 tetramers on the apical membrane of the principal cellsin the heterozygotes for water reabsorption. In addition to the WThomotetramers, the chimeric tetramers consisting of both WT andmutant subunits may also reach the apical membrane, guided by theWT subunits.

The cph mutants represent a special class of NDI caused byprotein trafficking defects in which most of the Aqp2 proteins aremisrouted to the basolateral membrane rather than being seques-tered intracellularly. Investigation of factors controlling the polar-ized trafficking of Aqp2 may yield strategies for reversing themisrouting and restoring the apical accumulation of Aqp2. The cphmutants would be a nice model for testing such strategies that mayhave potential therapeutic benefits for NDI as well as implicationsfor other trafficking related diseases. On the other hand, whereassurgical models have been instrumental in the study of pathologicalchanges in obstructive nephropathy, the abrupt onset of the ob-struction differs dramatically from those seen in most congenitalobstructive nephropathy cases. The cph mutants have naturallyoccurring hydronephrosis that is 100% penetrant, making theman excellent model for the study of congenital obstructivenephropathy, especially those caused by functional but not physicalobstruction.

Materials and MethodsMice and Microsatellite Markers. We intercrossed cph heterozygotes,obtained from The Jackson Laboratory, to generate phenotypicallynormal mice that are considered controls (either heterozygotes orWT) and mice having kidney defects that are considered mutants(homozygotes). The cph heterozygotes have no detectable abnor-mality and, before the revelation of the mutation, can be identifiedonly by their ability to produce cph homozygous mutants in matingswith confirmed cph heterozygotes. DBA�2J, AKR�J, and Mold�RkJ inbred strains were also obtained from The Jackson Labora-tory for genetic mapping. To identify polymorphic microsatellitemarkers for the genetic mapping of cph, we have used the followingdatabases: MGI, the Ensembl Mouse Genome Server, and theMouse Genetic Map Database from Massachusetts Institute ofTechnology. In addition, we developed a computer program to scanthe mouse genome for DNA fragments consisting of at least 15copies of repeating units of 2–5 nt. By blasting the DNA fragmentcontaining the repeats and unique flanking sequences identified byour computational analyses against the strain-specific genomicfragments within the Celera Discovery System, we were able toidentify unpublished polymorphic microsatellite markers.

Histological analyses, immunostaining, and corrosive castingwere performed as described (12, 28). Antibodies used were asfollows: Aqp2 (polyclonal 1:500; BD Biosciences); neurofilament(monoclonal 1:100; Developmental Hybridoma Studies Bank, Uni-versity of Iowa); �SMA (monoclonal 1:300; Sigma); Aqp2S256phosphorylation specific (polyclonal 1:400; from S. Nielsen); E-Cadherin (monoclonal 1:200; BD Transduction Laboratories);Aqp3 (polyclonal 1:150; Chemicon); Aqp4 (polyclonal 1:400;Chemicon); Alexa Fluor 488- or Alexa Fluor 566-conjugated goatanti-mouse or goat anti-rabbit (1:1,000; Molecular Probes). FITC-labeled dolichos biflorus agglutinin (Vector Laboratories) was usedat 1:100. Images in Fig. 5 were taken with a Zeiss LSM510 confocalmicroscope.

Serological Analysis and Urinalysis. Mice were housed in metaboliccages (Nalge) and given free access to water and powdered chow(Rodent Diet 5001; LabDiet). Mice were allowed to acclimatize inthe cages for 24 h before DDAVP (Sigma) injection at 0.4 �g�kgi.p. Urine samples were collected for the 24 h before injection fromthe metabolic cage and at 2 h after DDAVP injection by sponta-neous voiding or bladder massage. A vapor pressure osmometer(Wescor, Logan, UT) was used to determine urine osmolality.Urine and serum chemistry was analyzed by the Renal ChemistryLaboratory at Washington University School of Medicine.

Western Blotting and RT-PCR. Kidney protein extracts were preparedas described (29), separated by SDS�PAGE, transferred to Immo-bilon-P membranes (Millipore), and probed by the anti-Aqp2antibody and an anti-�-actin antibody (JLA20; DevelopmentalHybridoma Studies Bank, University of Iowa). After incubationwith Alexa Fluor 680-conjugated goat anti-rabbit IgG (MolecularProbes) and IRDye 800CW-conjugated goat anti-mouse IgG(Rockland, Gilbertsville, PA) secondary antibodies, the antibodycomplexes were visualized by the Odyssey Infrared Imaging System(LI-COR). RT-PCR primers for Aqp2 were as follows: 5�-CACATCAACCCTGCTGTGAC-3� and 5�-CAGCTGCATGGT-CAGGAAGAG-3�, 228 bp product). Cycling conditions were 45°Cfor 30 min, 95°C for 5 min, then 30 cycles of 95°C for 20 s, 58°C for35 s, and 72°C for 45 s.

We thank Drs. J. H. Miner, C. Moulson, E. Eicher, H. Liapis, and J.Mandell for helpful information and suggestions; Dr. S. Nielsen (Waterand Salt Research Center, University of Aarhus, Aarhus, Denmark) forthe Aqp2-S256 phosphorylation-specific antibody; and D. Martin forrunning the urine and serum chemistry assays. F.C. is supported in partby institutional funds from the Department of Internal Medicine�RenalDivision at Washington University School of Medicine, a March ofDimes Basil O’Conner Award, and National Institutes of Health (NIH)Grants R21DK64816 and R01DK067386. P.A.K is supported by afellowship from the NIH.

1. Chevalier, R. L. (1999) Pediatr. Nephrol. 13, 612–619.2. Grasso, M. & Gitlin, J. (2001) eMedicine, www.emedicine.com�med�topic3074.htm.3. Feather, S. A., Malcolm, S., Woolf, A. S., Wright, V., Blaydon, D., Reid, C. J., Flinter, F. A.,

Proesmans, W., Devriendt, K., Carter, J., et al. (2000) Am. J. Hum. Genet. 66, 1420–1425.4. Chertin, B. & Puri, P. (2003) J. Urol. 169, 1804–1808.5. Peters, C. A. (2001) Prenatal Diagn. 21, 917–923.6. Fox, S. & Eicher, E. (1978) Mouse News Letter 58, 47.7. Horton, C. E., Jr., Davisson, M. T., Jacobs, J. B., Bernstein, G. T., Retik, A. B. & Mandell,

J. (1988) J. Urol. 140, 1310–1315.8. Fujiwara, T. M. & Bichet, D. G. (2005) J. Am. Soc. Nephrol. 16, 2836–2846.9. Yang, B., Gillespie, A., Carlson, E. J., Epstein, C. J. & Verkman, A. S. (2001) J. Biol. Chem.

276, 2775–2779.10. Yun, J., Schoneberg, T., Liu, J., Schulz, A., Ecelbarger, C. A., Promeneur, D., Nielsen, S.,

Sheng, H., Grinberg, A., Deng, C. & Wess, J. (2000) J. Clin. Invest. 106, 1361–1371.11. Lloyd, D. J., Hall, F. W., Tarantino, L. M. & Gekakis, N. (2005) PLoS Genet. 1, e20.12. Chang, C. P., McDill, B. W., Neilson, J. R., Joist, H. E., Epstein, J. A., Crabtree, G. R. &

Chen, F. (2004) J. Clin. Invest. 113, 1051–1058.13. Noda, Y. & Sasaki, S. (2005) Biol. Cell 97, 885–892.14. Nejsum, L. N., Zelenina, M., Aperia, A., Frokiaer, J. & Nielsen, S. (2005) Am. J. Physiol.

Renal Physiol. 288, F930–F938.15. King, L. S., Kozono, D. & Agre, P. s(2004) Nat. Rev. Mol. Cell Biol. 5, 687–698.16. Zatuchni, J., Armento, D. F. & Menzel, P. H. (1964) Am. J. Med. Sci. 247, 445–450.

17. Laycock, J. F., Barrett, M. C. & Woodrow, D. F. (1991) Int. J. Exp. Pathol. 72, 581–587.18. Fushimi, K., Sasaki, S. & Marumo, F. (1997) J. Biol. Chem. 272, 14800–14804.19. Katsura, T., Gustafson, C. E., Ausiello, D. A. & Brown, D. (1997) Am. J. Physiol. 272,

F817–F822.20. van Balkom, B. W., Savelkoul, P. J., Markovich, D., Hofman, E., Nielsen, S., van der Sluijs,

P. & Deen, P. M. (2002) J. Biol. Chem. 277, 41473–41479.21. de Mattia, F., Savelkoul, P. J., Kamsteeg, E. J., Konings, I. B., van der Sluijs, P.,

Mallmann, R., Oksche, A. & Deen, P. M. (2005) J. Am. Soc. Nephrol. 16, 2872–2880.22. Campo, C., Mason, A., Maouyo, D., Olsen, O., Yoo, D. & Welling, P. A. (2005) Rev. Physiol.

Biochem. Pharmacol. 153, 47–99.23. Rodriguez-Boulan, E., Kreitzer, G. & Musch, A. (2005) Nat. Rev. Mol. Cell Biol. 6, 233–247.24. Kamsteeg, E. J., Bichet, D. G., Konings, I. B., Nivet, H., Lonergan, M., Arthus, M. F., van

Os, C. H. & Deen, P. M. (2003) J. Cell Biol. 163, 1099–1109.25. Kamsteeg, E. J., Wormhoudt, T. A., Rijss, J. P., van Os, C. H. & Deen, P. M. (1999) EMBO

J. 18, 2394–2400.26. de Mattia, F., Savelkoul, P. J., Bichet, D. G., Kamsteeg, E. J., Konings, I. B., Marr, N.,

Arthus, M. F., Lonergan, M., van Os, C. H., van der Sluijs, P., et al. (2004) Hum. Mol. Genet.13, 3045–3056.

27. Kamsteeg, E. J., Heijnen, I., van Os, C. H. & Deen, P. M. (2000) J. Cell Biol. 151, 919–930.28. Graef, I. A., Chen, F., Chen, L., Kuo, A. & Crabtree, G. R. (2001) Cell 105, 863–875.29. Wang, W., Li, C., Kwon, T. H., Knepper, M. A., Frokiaer, J. & Nielsen, S. (2002) Am J.

Physiol. Renal Physiol. 283, F1313–25.

McDill et al. PNAS � May 2, 2006 � vol. 103 � no. 18 � 6957

GEN

ETIC

S