Novel Mutations in Aquaporin-2 Gene in Female Siblings with ...

5 downloads 0 Views 659KB Size Report
versal primers were used in the cycle-sequencing reaction. DNA se- quences were ..... Bichet DG, Arthus M, Lonergan M, et al. 1993 X-linked nephrogenic ...
0021-972X/98/$03.00/0 Journal of Clinical Endocrinology and Metabolism Copyright © 1998 by The Endocrine Society

Vol. 83, No. 9 Printed in U.S.A.

Novel Mutations in Aquaporin-2 Gene in Female Siblings with Nephrogenic Diabetes Insipidus: Evidence of Disrupted Water Channel Function* KATSUMI GOJI†, MICHIO KUWAHARA†, YONG GU, MASAFUMI MATSUO, FUMIAKI MARUMO, AND SEI SASAKI Department of Endocrinology and Metabolism, Kobe Children’s Hospital (K.G.); Second Department of Internal Medicine, School of Medicine, Tokyo Medical and Dental University (M.K., Y.G., F.M., S.S.); and International Center for Medical Research, Kobe University School of Medicine (M.M.), Japan ABSTRACT Novel mutations of the aquaporin-2 (AQP2) gene have been detected in Japanese female siblings with autosomal-recessive nephrogenic diabetes insipidus. The patients were compound heterozygote for point mutations at nucleotide position 374 (C374T) and at position 523 (G523A) in exon 2 of the AQP2 gene, resulting in substitution of methionine for threonine at codon 125 (T125M) and arginine for glycine at codon 175 (G175R). The water permeability (Pf) of oocytes injected with wild-type complementary RNA increased 9.0-fold compared with the Pf of water-injected oocytes, whereas the increases in the Pf of oocytes injected with T125M and G175R complementary

C

ONGENITAL nephrogenic diabetes insipidus (NDI) is a hereditary disorder characterized by a lack of responsiveness to arginine vasopressin (AVP) in the renal tubules. The clinical hallmarks of NDI are polyuria and polydipsia. In the majority of families with NDI, the disease exhibits a hereditary pattern consistent with X-linked transmission. In 1992, X-linked NDI was found to be caused by mutations in the V2 receptor gene located in the Xq28 region (1–3). Subsequently, more than 60 mutations in the V2 receptor gene have been reported in NDI patients (4). However, some families with NDI show an autosomal-recessive mode of inheritance (5), and it has been postulated that the cellular abnormalities in NDI may be not only at the V2 receptor but also in the postreceptor cascade of events that mediate AVP-induced antidiuresis. In 1993, Fushimi et al. (6) isolated a complementary DNA (cDNA) for aquaporin-2 (AQP2), a vasopressin-regulated water channel. Subsequently, a cDNA for human AQP2 was isolated (7). AQP2 is expressed predominantly at the apical region of the principal cells of the collecting duct and the inner medullary collecting duct cells (6, 8, 9). In these cells, vasopressin increases the osmotic water permeability of the

Received February 17, 1998. Revision received May 12, 1998. Accepted May 20, 1998. Address all correspondence and requests for reprints to: Katsumi Goji, Department of Endocrinology and Metabolism, Kobe Children’s Hospital, 1–1-1 Takakuradai, Suma-ku, Kobe 654, Japan. * This work was supported by a Grant-in-Aid from the Ministry of Education, Science, and Culture, Japan; a grant from The Salt Science Research Foundation; and a Grant for Pediatric Research (9C-3) from the Ministry of Health and Welfare, Japan. † These authors contributed equally to this work.

RNA were only 1.7-fold and 1.5-fold, respectively. Immunoblot and immunocytochemistry indicated that the plasma membrane expressions of T125M and G175R AQP2 proteins were comparable to that of the wild-type, suggesting that although neither the T125M nor G175R mutation had a significant effect on plasma membrane expression, they both distorted the structure and function of the aqueous pore of AQP2. These results provide evidence that the nephrogenic diabetes insipidus in patients with T125M and G175R mutations is attributable not to the misrouting of AQP2, but to the disrupted water channel function. (J Clin Endocrinol Metab 83: 3205– 3209, 1998)

apical membrane by triggering exocytosis of AQP2-containing vesicles (9 –12). Identification of AQP2 mutation in a patient with NDI confirmed that AQP2 is necessary for urinary concentration (13). Presently, further mutations of AQP2 were reported in NDI patients (14 –18). Functional expression studies indicated that the defect of these mutations is caused by a lack of plasma membrane appearance of AQP2 (misrouting), rather than an inhibition of water channel function (17, 19). We report here two female siblings with an autosomal-recessive form of NDI. They were found to be compound heterozygote for two missense mutations in the AQP2 gene. Functional expression studies have provided the first evidence that these mutations cause disruption of the water channel function rather than misrouting. These two novel mutations may shed important insight into the understanding of the structure-function relationship in AQP2. Patients and Methods Patients The female proband (patient 1) was delivered to Japanese healthy parents at term after an uncomplicated pregnancy. Family history was negative for NDI, and her parents were not known to be consanguineous. At the age of 4 days, she was admitted to our hospital because of unexplained fever. At admission, hypernatremia (154 mmol/L) and high serum osmolality (322 mOsm/kg H20) were found. Urine osmolality (121 mOsm/kg H20) was low in spite of a high plasma level of AVP (37.4 pg/mL) measured by RIA, and it did not increase in response to sc administration of aqueous vasopressin. Fluid intake ad libitum with a salt-restricted diet and administration of hydrochlorothiazide normalized her serum sodium and osmolality. At the age of 20 months, she was readmitted to our hospital to confirm the diagnosis. She was diagnosed with NDI because of an inability of the urine to concentrate despite a 5.1% weight loss during a 3-h water deprivation test, and an inability of the urine to concentrate after sc administration of aqueous vasopressin.

3205

3206

JCE & M • 1998 Vol 83 • No 9

GOJI ET AL.

When patient 1 was 2 yr old, her sister (patient 2) was born at term. At the age of 4 months, patient 2 was admitted to our hospital because of fever and failure to thrive. At admission, hypernatremia (163 mmol/L) and high serum osmolality (330 mOsm/kg H20) were found. Urine osmolality (70 mOsm/kg H20) was low despite a high plasma level of AVP (65.5 pg/mL); she was received the same diagnosis as her elder sister. Chromosomal studies showed a 46,XX karyotype with t(3;9) (q21/q34) translocation in both siblings and their mother.

DNA sequencing analysis Genomic DNA was obtained from the peripheral blood of the patients by standard methods. PCR amplification of the AQP2 gene was performed with primers described by Deen et al. (13). Because PCR primers for amplification of the AQP2 gene were synthesized with universal sequencing primer binding site added to the 59 end, fluorescent universal primers were used in the cycle-sequencing reaction. DNA sequences were analyzed with an automated DNA sequence analyzer (A.L.F. red DNA sequencer, Pharmacia LKB, Uppsala, Sweden). Nucleotides were numbered with respect to the A of the first ATG of the open reading frame.

Site-directed mutagenesis and in vitro complementary RNA (cRNA) synthesis Mutants of human AQP2 were made with the PCR technique using AQP2 cDNA as a template (6, 20). A fragment between the NcoI site at nucleotide 134 and the StuI site at nucleotide 809 in pAQP2/ev1 was replaced by a PCR fragment coding the mutants. Threonine at position 125 in the amino acid sequence of human AQP2 cDNA was altered to methionine with a mutation primer, 59-CAGCAACAGCATGACGGCTGGCC-39. Glycine at position 175 was altered to arginine with a mutation primer, 59-GGGCCACCTCCTTAGGATCCATT-39. Recently, a mutation of threonine to methionine at position 126, the residue next to threonine 125, was found in a patient with NDI (17). This mutant was also constructed using a mutation primer, 59-CAGCAACAGCACGATGGCTGGCC-39. Mutations were confirmed by DNA sequencing. Capped RNA transcripts of wild-type and mutated AQP2 were synthesized in vitro with T3 RNA polymerase using NotI-digested AQP2 cDNA.

tibody against AQP2 diluted at 1:500, rinsed with PBS, and further incubated for 30 min with FITC-labeled goat antirabbit IgG (1:200, Sigma Chemical Co., St. Louis, MO). Oocytes were imaged with a fluorescent microscope at 3400 magnification (Nikon BIOPHOT, Nihonkougaku, Tokyo, Japan).

Results

Direct sequencing of DNA strands in both of the siblings (patients 1 and 2) revealed that they were compound heterozygote for point mutations in the AQP2 gene; a cytosineto-thymine transition at nucleotide position 374 (C374T) and a guanine-to-adenine transition at position 523 (G523A) were found in exon 2, resulting in a threonine-to-methionine mutation at codon 125 (T125M) and glycine-to-arginine mutation at codon 175 (G175R), respectively. Direct sequencing of exon 2 of the AQP2 gene from each of the parents revealed that the C374T mutation was inherited from the father and the G523A mutation from the mother. Because the C374T and G523A mutations create NlaIII and Eco81I restriction enzyme recognition sites, respectively, the compound heterozygous state of both siblings for the mutations was confirmed by digesting the amplified fragment of exon 2 with NlaIII and Eco81I (Fig. 1). The PCR products of the exon 2 coding region from 80 normal chromosomes from 40 Japanese subjects remained uncut with NlaIII and Eco81I, indicating that these two mutations are not frequent polymorphisms in the population studied.

Measurement of osmotic water permeability of oocyte Oocytes at stages V-VI were obtained from Xenopus laevis. Each oocyte was injected with 40 nL of water (control) or 3 ng wild-type or mutated AQP2 cRNA. Oocytes were incubated for 48 h at 20 C in Barth’s buffer. The osmotic water permeability (Pf) of the oocytes was measured at 20 C from the time course of osmotic cell swelling (20). The oocytes were transferred from 200 mOsm Barth’s buffer to 70 mOsm buffer, and then imaged on a CCD (charge coupled device) camera connected to an area analyzer (Hamamatsu Photonics C3160, Hamamatsu, Japan). Serial images taken at 0.5-sec intervals were stored in a computer. Pf was calculated from the initial 15-sec response of cell swelling, as described.

Immunoblot analysis Lysates and plasma membrane fraction of oocytes were obtained as previously described (19, 21). After being heated at 70 C for 10 min, samples were separated by SDS-PAGE. Oocyte lysates from 0.2 oocytes or plasma membrane from 20 oocytes were applied in each lane. The samples were transferred to Immobilon-P filter (Millipore, Marlborough, MA) using a semidry system. The filters were incubated for 1 h with an affinity-purified antibody against 15 COOH-terminal amino acids of AQP2 (6). The filters were further incubated for 1 h with 125 I-labeled protein A solution, followed by autoradiography.

Immunocytochemistry Oocytes were fixed in 4% paraformaldehyde for 4 h and cryoprotected overnight in PBS containing 30% sucrose. The samples were embedded in OCT compound (Tissue Tek Products, Miles Laboratories, Inc., Elkhart, IN) and frozen in liquid nitrogen. Cryostat sections (6 mm) were incubated for 30 min in PBS containing 1% BSA. After 3 washes in PBS, the sections were incubated for 60 min with affinity-purified an-

FIG. 1. Inheritance of mutations in family of NDI patients. Squares and circles represent males and females, respectively. Patients are indicated with filled circles. Mutations in AQP2 gene were confirmed by restriction analysis. PCR products of exon 2 of AQP2 gene were digested with NlaIII or Eco81I and separated on a 2% agarose gel. Resulting fragment lengths are given in base pairs.

NOVEL MUTATIONS IN AQUAPORIN-2 GENE

cRNA encoding wild-type or mutant AQP2 was injected into oocytes to test their water channel function (Fig. 2). In oocytes injected with water or cRNA encoding AQP2, average Pf values (in cm/sec 3 1024) were 22 6 2 (water), 198 6 8 (wild-type), 38 6 5 (T125M), 33 6 4 (G175R), and 87 6 6 (T126M). The relative Pf value for wild-type AQP2 vs. water control was 9.0, whereas those for T125M, G175R, and T126M were 1.7, 1.5, and 4.0, respectively. Coinjection of either of the mutant cRNAs with wild-type cRNA had no effect on Pf of wild-type AQP2 (data not shown). To examine the size and amount of mutant AQP2 proteins expressed in oocytes, oocyte lysates were immunoblotted (Fig. 3A). A band of 29-kDa AQP2 protein was detected in all lanes except the lane of control oocytes injected with water. Although there were some differences, the intensities of the 29-kDa bands were comparable among oocytes injected with wild-type and mutant cRNAs. An additional broad band of 35 kDa to approximately 45 kDa representing a glycosylated form of AQP2 protein (6, 22) was observed in the lanes loaded with oocytes expressing wild-type, G175R, and T126M AQP2. This broad band was absent in the lane of T125M. To determine whether the mutant AQP2 proteins were localized in the plasma membrane, similar immunoblot studies were performed using the plasma membrane fraction of the oocytes. (Fig. 3B). The two bands of 29-kDa and 35 kDa to approximately 45 kDa were detected in oocytes injected with wild-type and G175R cRNAs, whereas only the 29-kDa band was detected in oocytes injected with T125M cRNA. Both of the bands were detected in the plasma membrane fraction from oocytes injected with T126M cRNA, but their intensities were apparently weaker than the intensities of the bands for wild-type, T125M, and G175R. The expression of AQP2 proteins in the oocyte plasma membrane was further determined by immunocytochemistry (Fig. 4). The plasma membrane was not stained in waterinjected oocytes (Fig. 4A). In contrast, a bright immunofluorescence staining of the plasma membranes was observed in oocytes injected with cRNA of wild-type AQP2 (Fig. 4B), T125M (Fig. 4C), G175R (Fig. 4D), and T126M (Fig. 4E).

3207

FIG. 3. Immunoblot of oocyte lysates (A) and oocyte plasma membrane fractions (B) probed with an affinity-purified antibody against human AQP2. Oocytes were injected with water (control) or 3 ng cRNA of wild-type or mutated human AQP2s (T125M, G175R, and T126M). Samples of lysates from 0.2 equivalent oocytes or samples of plasma membranes from 20 equivalent oocytes were loaded in each lane.

Discussion

FIG. 2. Osmotic water permeability (Pf) of wild-type and mutated AQP2. Oocytes were injected with 40 nL water (control) or 3 ng wild-type or mutant AQP2 cRNAs with alteration of threonine 125, glycine 175, and threonine 126 to methionine (T125M), arginine (G175R), and methionine (T126M), respectively. Pf was calculated from time course of osmotic cell swelling of oocytes. Each bar represents means 1 SE of 26 –30 measurements.

We report here female NDI siblings with compound heterozygote for novel missense mutations, T125M and G175R, in the AQP2 gene. Oocytes injected with either T125M or G175R mutant cRNA showed little increase in water permeability, indicating that missense AQP2 proteins of T125M and G175R were the cause of the NDI in our patients. Immunoblot and immunocytochemistry indicated that the plasma membrane expressions of T125M and G175R AQP2 proteins were comparable to that of the wild-type. These results suggest that, rather than having any significant effect on plasma membrane expression, both of the T125M and G175R mutations distort the structure and function of the aqueous pore of AQP2. Sixteen mutation sites have been found in the human AQP2 gene, including two novel sites reported in this study (13–18) (Fig. 5). Deen et al. (19) and Mulders et al. (17) re-

3208

GOJI ET AL.

JCE & M • 1998 Vol 83 • No 9

FIG. 4. Immunocytochemistry of oocytes injected with cRNA encoding wild-type AQP2 (B), T125M (C), G175R (D), or T126M (E). Water-injected oocyte was also presented as a negative control (A). Oocyte sections were incubated with affinity-purified antibody against AQP2, and immunostained with goat antirabbit IgG conjugated with FITC. Sections were viewed with a fluorescence microscope at magnification 3400.

FIG. 5. Mutation sites of human AQP2 identified in patients with NDI. Solid symbols represent predicted location of mutations. These mutations include L22V, G64R, N68S, V71M, G100X, 439delC, T125M, T126M, A147T, G175R, C181W, P185A, R187C, A190T, S216P, and P262L.

ported that the inability of the mutant proteins (G64R, N68S, T126M, A147T, R187C, and S216P) to facilitate water transport in Xenopus oocytes was mainly caused by impaired routing to the plasma membrane. The positions of these mutants are distributed in the extracelluar (T126M, R187C), transmembrane (A147T, S216P), and intracellular (G64R,

N68S) domains, and none of the mutations they studied showed normal translocation to the plasma membrane. Thus, we have identified for the first time missense mutations that have no apparent effect on plasma membrane expression but decrease water permeability. Interestingly, the amino acid position of T126M, a missense AQP2 mutation identified in one of the patients reported by Mulders et al. (17), is just one amino acid downstream to that identified in our patient (T125M). Accordingly, we constructed and characterized T126M mutant. The water permeability of the oocytes expressing T126M protein was significantly higher than that of oocytes expressing T125M protein and was comparable with about 50% of the water permeability of the oocytes expressing wild-type AQP2. Immunocytochemistry showed a clear AQP2 staining in the plasma membrane of oocytes expressing T126M AQP2. It is difficult to evaluate the amount of AQP2 protein localized at plasma membrane from the finding of immunocytochemistry, however, immunoblot analysis using membrane fraction of the oocytes revealed that the expression of T126M protein was decreased compared with that of T125M protein. Relatively preserved water permeability of oocytes expressing T126M AQP2 could be explained by the decreased expression of T126M protein in the plasma membrane, and, as suggested by Mulders et al. (17), T126M AQP2 protein would be functional water channel.

NOVEL MUTATIONS IN AQUAPORIN-2 GENE

Using immunocytochemistry, we did not detect T126M AQP2 protein in the cytoplasm, whereas Mulders et al. showed labeling of T126M protein in the cytoplasm (17). Therefore, whether T126M AQP2 protein is localized in cytoplasm of oocytes, or whether cytoplasmic T126M AQP2 is not detected by our antibody remains to be clarified. It is noteworthy that the identical substitution of the neighboring amino acids (T125M and T126M) may produce very different effects on human AQP2; i.e. a loss of channel function in the case of T125M, and a misrouting of trafficking in the case of T126M. Bai et al. (22) showed that substitution of asparagine at 123 of rat AQP2 with larger residues (N123D, N123W) decreased oocyte Pf, raising the possibility that asparagine 123 is closely located to the aqueous pore. The missense mutation identified in our patients (T125M) corresponds to the position only two amino acid downstream of the asparagine 123 of rat AQP2. Because methionine residues have larger lateral moiety than threonine, replacement of threonine residue at 125 with methionine would interfere with the formation of the aqueous pathway in human AQP2. Immunoblot analysis failed to detect the glycosylated band of AQP2 in T125M (Fig. 3). In rat AQP2, the consensus sequence for N-glycosylation (N-X-S/T) is present at positions 124 –126 (N-A-T), and it has been shown that asparagine 124 is the glycosylation site (22). The corresponding consensus sequence for glycosylation is N-S-T at positions 123–125 in human AQP2. Thus, it is quite reasonable that the T125M mutation lacks glycosylation. However, the disrupted water channel function of the T125M mutant may not be caused by the lack of glycosylation, because a previous study demonstrated that glycosylation itself is not necessary for the expression of water channel function in rat AQP2 (22). In a deduced membrane topology of AQP2, glycine 175 is located at the transition from the transmembrane to the third extra cellular domains. The mechanism for the decrease in water channel function by substitution of arginine for glycine at 175 is unknown at present. Glycine is an important amino acid in determining the three-dimensional structure of polypeptide, and this residue is usually well conserved in members of a given protein family. The mutation causing substitution of glycine may distort the structure of AQP2. In summary, we identified two missense mutations in the AQP2 gene that have shown for the first time a loss of channel function with little effect on the intracellular trafficking. Our conclusions were drawn from the observations of functional expression of human AQP2 in Xenopus oocytes. Xenopus oocytes have been shown to be useful to the expression studies of membrane proteins. However, there could be some discrepancies in the intracellular trafficking and glycosylation between oocytes and mammalian cells. Therefore, further studies on functional expression of AQP2 in mammalian cells will be needed to elucidate the molecular details of intracellular localization and channel function of AQP2.

3209

Acknowledgments We thank Dr. K. Fushimi for valuable discussion.

References 1. Pan Y, Metzenberg A, Das S, Jing B, Gitschier J. 1992 Mutations in the V2 vasopressin receptor gene are associated with X-linked nephrogenic diabetes insipidus. Nat Genet. 2:103–106. 2. Bichet DG, Arthus M, Lonergan M, et al. 1993 X-linked nephrogenic diabetes insipidus mutations in North America and the Hopewell hypothesis. J Clin Invest. 92:1262–1268. 3. Holtzman EJ, Harris HW, Kolakowski LF, Guay-Woodford L, Botelho B, Ausiello DA. 1993 A molecular defect in the vasopressin V2-receptor gene causing nephrogenic diabetes insipidus. N Engl J Med. 328:1534 –1537. 4. Tsukaguchi H, Matsubara H, Mori Y, et al. 1995 Two vasopressin type 2 receptor gene mutations R143P and DV278 in patients with nephrogenic diabetes insipidus impair ligand binding of the receptor. Biochem Biophys Res Commum. 211:967–977. 5. Langley JM, Balfe JW, Selander T, Ray PN, Clarke JTR. 1991 Autosomal recessive inheritance of vasopressin-resistant diabetes insipidus. Am J Med Genet. 38:90 –94. 6. Fushimi K, Uchida S, Hara Y, Hirata Y, Marumo F, Sasaki S. 1993 Cloning and expression of apical membrane water channel of rat kidney collecting tubule. Nature. 361:549 –552. 7. Sasaki S, Fushimi K, Saito H, et al. 1994 Cloning, characterization, and chromosomal mapping of human aquaporin of collecting duct. J Clin Invest. 93:1250 –1256. 8. Nielsen S, DiGiovanni SR, Chrisstensen EI, Knepper MA, Harris HW. 1993 Cellular and subcellular immunolocalization of vasopressin-regulated water channel in rat kidney. Proc Natl Acad Sci USA. 90:11663–11667. 9. Nielsen S, Chou CL, Marples D, Christensen EI, Kishore BK, Knepper MA. 1995 Vasopressin increases permeability of kidney collecting duct by including translocation of aquaporin-CD water channels in plasma membrane. Proc Natl Acad Sci USA. 92:1013–1017. 10. Sabolic I, Katsura T, Verbabatz JM, Brown D. 1995 The AQP2 water channel: effect of vasopressin treatment, microtubule disruption, and distribution in neonatal rats. J Membr Biol. 143:165–177. 11. Yamamoto N, Sasaki S, Fushimi K, et al. 1995 Vasopressin increases AQP-CD water channel in the apical membrane of collecting cells without affecting AQP3 distribution in Brattleboro rat. Am J Physiol. 268:C1546 –C1551. 12. Marples D, Knepper MA, Christensen EI, Nielsen S. 1995 Redistribution of aquaporin-2 water channels induced by vasopressin in rat kidney inner medullary collecting duct. Am J Physiol. 269:C655–C664. 13. Deen PM, Verdijik MAJ, Knoers NVAM, et al. 1994 Requirement of human renal water channel aquaporin-2 for vasopressin-dependent concentration of urine. Science. 264:92–95. 14. van Lieburg AF, Verdijk MAJ, Knoers NVAM, et al. 1994 Patients with autosomal nephrogenic diabetes insipidus homozygous for mutations in the aquaporin 2 water-channel gene. Am J Hum Genet. 55:648 – 652. 15. Bichet DG, Arthus M-F, Lonergan M, et al. 1995 Autosomal dominant and autosomal recessive nephrogenic diabetes insipidus: novel mutations in the AQP2 gene. J Am Soc Nephrol. 6:717 (Abstract 2168). 16. Canfield MC, Tamarappoo BK, Moses AM, Verkman AS, Holtzman EJ. 1997 Identification and characterization of aquaporin-2 water channel mutations causing nephrogenic diabetes insipidus with partial vasopressin response. Hum Mol Genet. 6:1865–1871. 17. Mulders SM, Knoers NVAM, van Lieburg AF, et al. 1997 New mutations in the AQP2 gene in nephrogenic diabetes insipidus resulting in functional but misrouted water channels. J Am Soc Nephrol. 8:242–248. 18. Hochberg Z, van Lieburg A, Even L, et al. 1997 Autosomal recessive nephrogenic diabetes insipidus caused by an aquaporin-2 mutation. J Clin Endocrinol Metab. 82:686 – 689. 19. Deen PMT, Croes H, van Aubel AMH, Ginsel LA, van Os CH. 1995 Water channels encoded by mutant aquaporin-2 genes in nephrogenic diabetes insipidus are impaired in their cellular routing. J Clin Invest. 95:2291–2296. 20. Kuwahara M, Fushimi K, Terada Y, Bai L, Marumo F, Sasaki S. 1995 cAMPdependent phosphorylation stimulates water permeability of aquaporin-collecting duct water channel protein expressed in Xenopus oocytes. J Biol Chem. 270:10384 –10387. 21. Wall DA, Patel S. 1989 Isolation of plasma membrane complexes from Xenopus oocytes. J Membr Biol. 107:189 –201. 22. Bai L, Fushimi K, Sasaki S, Marumo F. 1996 Structure of aquaporin-2 vasopressin water channel. J Biol Chem. 271:5171–5176.