T m JOURNAL OF BIOICGICAL CHEMISTRY 0 1994 by The American Soeiety for Biochemistry and Molecular Biology, Inc

Vol. 269, No. 14, Issue of April 8, pp. 10417-10422, 1994 Printed in U.S.A.

Functional Independence of Monomeric CHIP28 Water Channels Revealed by Expression of Wild-type Mutant Heterodimers*

(Received for publication, December 7, 1993, and in revised form, January 11, 1994)

Lan-bo Shi, William R. SkachS, and A. S. Verkmane From the Departments of Medicine and Physiology, Cardiovascular Research Institute and Cancer Research Institute, University of California, Sun Francisco, California, 94143-0521

CHIP28 is a mqjor water transporting protein in erythrocytes and kidney which forms tetramers in mem- branes (Verbavatz, J. M., Brown, D., Sabolic, I., Valenti, G., Ausiello, D. A, Van Hoek, A. N., Ma, T., and Verkman, A. S. (1993) J. Cell Biol. 123, 605-818). To determine whether CHIP28 monomers function independently, chi- meric cDNA dimers were constructed which contained wild-type CHIP28 in series with either wild-type CHIP28, a non-water transporting CHIP28 mutant (ClSQW), or a functional but mercurial-insensitive CHIP28 mutant (C189S). Transcribed cRNAs were in- jected in Xenopurr oocytes and plasma membrane expres- sion was assayed by quantitative immunofluorescence. Water channel function was measured by osmotically induced swelling. CHIP28 homo- and heterodimers were targeted to the oocyte plasma membrane and functioned as water channels. Relative osmotic water permeability (Pt) values (normalized for plasma membrane expres- sion of monomeric subunits) were: 1.0 (CHIP28 mono- mer), 0.0 (C189W), 1.07 (C189S), 1.10 (CHIP28-CHIP28 dimer) and 0.52 (CHIP28-Cl89W). The increase in oocyte Pf was linearly related to plasma membrane expression of wild-type CHIP28 and C189S subunits. HgCI, (0.3 mm) inhibited channel-mediated Pf in oocytes expressing wild-type CHIP28 monomers and dimers by 8LW0, but did not inhibit Pf in oocytes expressing C189S. HgCI, inhibited Pf in oocytes expressing CHIP28-Cl89S dimers by 44 * 7%, consistent with one mercurial-sensitive and one insensitive subunit in the heterodimer. These re- sults indicate that despite their assembly in tetramers, monomeric CHIP28 subunits function independently as water channels.

m a n n e l forming Integral Protein of 28 kDa (CHIP281 is the first biological water channel to be identified (1). Functional studies in proteoliposomes reconstituted with purified CHIP28 (2, 31, stably transfected Chinese hamster ovary cells express- ing CHIP28 (4), and Xenopus oocytes (5, 6) indicate that CHIP28 is a selective water channel that excludes small ions and solutes. CHIP28 is present in large quantities in erythro- cytes and various epithelial and endothelial cells involved in fluid transport. Immunolocalization and in situ hybridization

* This work was supported by grants DK35124 and HL42368 from the National Institutes of Health and a grant-in-aid from the American Heart Association. The costs of publication of this article were defrayed

hereby marked “advertisement” in accordance with 18 U.S.C. Section in part by the payment of page charges. This article must therefore be

1734 solely to indicate this fact. $Supported by Physician-Scientist Award CA01614 from the Na-

tional Institutes of Health. Q To whom correspondence should be addressed: 1065 Health Sci-

ences East Tower, Cardiovascular Research Institute, University of California, San Francisco, San Francisco, CA94143-0521. ”el.: 415-476- 8530; Fax: 415-665-3847.

studies showed strong CHIP28 expression in epithelial cells of the kidney proximal tubule and thin descending limb of Henle, choroid plexus, ciliary body, alveolus, intestinal crypt, gallblad- der, placenta, and other tissues (7-10). CHIP28 is homologous to the MIP (Major intrinsic Protein of lens fiber) family of proteins, which includes small proteins from mammals, bacte- ria, and plants (11, 12). Recently, the plant protein ?-TIP has been shown to transport water when expressed in Xenopus oocytes (131, and additional homologous water channels from mammalian tissues have been identified (14-16).

CHIP28 is a hydrophobic protein in which -50% of mono- mers are glycosylated at residue N42 (6). Circular dichroism and Fourier transform infrared spectroscopy data indicate that CHIP28 contains 4045% a-helical secondary structure (17). A model in which four membrane-spanning a-helical domains form a central aqueous channel has been proposed based on analysis of CHIP28 biogenesis and assembly at the endoplas- mic reticulum membrane (18). Evidence from cross-linking studies (19) and freeze-fracture electron microscopy (20) indi- cates that CHIP28 forms tetramers in membranes. Rotary shadowed electron micrographs of CHIP28 in reconstituted proteoliposomes, stably transfected Chinese hamster ovary cells expressing CHIP28, and thin descending limb of Henle showed a tetrameric arrangement of subunits with a greatest diameter of 8.5 nm (20). The role for assembly of CHIP28 mono- mers into oligomeric complexes is not known, but may be re- quired for protein stability and function.

Two lines of evidence suggest that CHIP28 monomers func- tion independently. Radiation inactivation in native kidney vesicles showed a -30-kDa target size for the functional unit of osmotic water permeability (21). The simplest interpretation of these results is that the size of the functional CHIP28 subunit is 30 kDa; however, it is equally plausible that the CHIP28 tetramer is the functional unit and CHIP28 monomers that are damaged by radiation are unable to associate into tetramers (22). A second line of evidence comes from coinjection of Xeno- pus oocytes with cRNA encoding wild-type CHIP28 and mutant proteins that are either non-functional (C189W) or functional but not mercurial inhibitable (C189S, Refs. 23, 24). Although wild-type and mutant monomers appeared not to interact, it was not possible to rule out the formation of homotetramers consisting of all wild-type and all mutant monomers. Nor was it possible to quantify the relative expression of wild-type and mutant monomers.

The purpose of this study was to determine whether indi- vidual monomeric subunits in CHIP28 tetramers function in- dependently andor interact at the functional level. Our ap- proach, based on elegant studies of cooperative interactions in K+ channel gating (251, was to express cDNA dimers of CHIP28. The cDNAs encoded a wild-type CHIP28 monomer linked in- frame to wild-type or various mutant CHIP28 monomers (Fig. 1). By this approach, wild-type and mutant monomers were forced to remain together in a single peptide of -57 kDa size.

10417

10418 CHIP28 Monomers Function Independently Transcribed cRNA was expressed in Xenopus oocytes. Plasma membrane protein expression was assayed by a novel quanti- tative immunofluorescence method, and water transport func- tion was measured from the kinetics of oocyte swelling in re- sponse to osmotic gradients. The results provided direct evidence that although CHIP28 forms tetramers in mem- branes, the individual monomers function independently.

EXPERIMENTAL PROCEDURES cDNA Constructs-The coding sequence of human CHIP28 was sub-

cloned into a modified SP64 vector containing the Xenopus 5'-untrans- lated globin enhancer (Fig. 1) to give plasmid pSP64.CHIP28 as de- scribed previously (6). The CHIP28 mutants C189S and C189W were prepared by site-directed mutagenesis as described previously (24). Tu construct dimeric cDNAS, plasmid pSP64.CHIP28 was used as template for PCR amplification (20 cycles) using sense primer 5'-GCCACCATG- G C C A ~ G A G T T C A A G ~ G ~ G 3 ' and antisense primer 5I-CTAGAT- T C A ~ ~ ~ ~ C ~ C A T C ~ C - 3 ' encoding NcoI and BspHI restric- tion sites, respectively. This PCR' fragment encodes the CHIP28 NH, terminus translation initiation codon (NcoI site) to nucleotide 269 (AAA) just 5' to the COOH terminus stop codon. The PCR fragment was digested with NcoI and BspHI, gel purified, and ligated into the NcoI- linearized, phosphatase-treated plasmid pSP64.CHIP28. TheBspHI re-

firmed by restriction mapping. This chimeric cDNA (pSP64.CHIP28- striction site is destroyed following ligation. Insert orientation was con-

CHIP28) encodes a single tandem repeat of two CHIP28 subunits. Tu prepare cDNAS encoding CHIP28 trimers (and then tetramers),

the procedure was repeated using the digested PCR fragment and NcoI- linearized plasmid pSP64.CHIP28-CHIP28. Tu prepare dimers contain- ing wild-type CHIP28 in series with mutant CHIP28, the C189S and C189W coding sequences were subcloned into pSP64 at NcoI and BamHI sites. The wild-type CHIP28 PCR fragment containing Ne01 and BspHI sites was ligated into NcoI-linearized plasmids pSP64.Cl89S and pSP64.Cl89W.

Tu prepare CHIP28-MIP26 dimers, a MIP26 fragment was PCR am- plified using sense primer 5'-CCCCTGCCATGGGGGAACTGCGGTC-3' and antisense primer 5'-CGGCTGGAGCTCTTACAGGGCCTGGG-3' encoding NcoI and SacI restriction sites, respectively. PCR fragments were digested by NcoI and SacI, gel purified, and ligated into NcoI- and SacI-digested plasmid pSP64. Plasmid pSP64.MIP26 was linearized with NcoI, phosphatase treated, gel purified, and ligated with the wild- type CHIP28 PCR fragment at NcoI and BspHI sites as above.

RNA "bunscription-Complementary RNA was transcribed in vitro using SP6 polymerase (Bethesda Research Laboratories) and 4 pg of plasmid DNAin a 100-pL voiume at 37 "C for 2 h. Diguanosine triphos- phate (Pharmacia LKB B i o t e c ~ o l o ~ Inc.) was included in the reaction mixture for capping. At the completion of the reaction, plasmid DNA was digested with RNase-free DNase (Invitrogenf. After phenol-chloro- form extraction and precipitation, the cRNA was washed with 70% ethanol and suspended in distilled water for oocyte injection.

Oocyte Znjection and Water ZYunsport AssayStage V and VI oocytes from Xenopus laevis were isolated and defolliculated with collagenase (type lA, Sigma, 1 mdml for 2 h at 20 "C) in Barth's buffer (200 mOsm). Oocytes were microinjected with 50-nl samples of cRNA (0-20 ng/pl) and incubated at 18 "C for 24 h. Osmotic water permeability (Pf) in oocytes was measured from the time course of oocyte swelling at 10 "C in response to a 20-fold dilution of the extracellular Barth's buffer with distilled water (26). In some experiments, oocytes were incubated with 0.3 n w HgCl, for 5 min before and during measurements, or HgC1, was added directly to the assay solution at -1 min after buffer dilution. Oocyte Pfwas calculated from the initial rate of swelling, d ( V ~ ~ ) / d ~ , by the relation Pi = [ ~ ( v ~ ~ ) / d t Y r t S ~ , t V ~ ( O s m ~ " ~ - O s ~ ) l , where S/V, = 50 cm", V, = 18 cm3/mol, and O s m ~ ~ - O s ~ = 190 mOsm.

I m m u n o ~ u o ~ s e e n c e - ~ r completion of the water permeability as- say, groups of 10-20 oocytes were immersed in fixation solution (4% paraformaldehyde in PBS containing 0.1 M sucrose) for 4 h. Oocytes were cryoprotected overnight in PBS containing 30% sucrose, embedded in ornithine carbamyl transferase compound, and frozen in dry ice/ ethanol. Serial sections (-6 pm thickness) were mounted on Superfrost' Plus glass slides (Fisher) for antibody staining. Frozen sections were preincubated with 0.5% Triton X-100 for 30 min at 23 "C. Sections were then washed and incubated with immune serum containing a polyclonal

The abbreviations used are: PCR, polymerase chain reaction; PBS, phosphate-buffered saline; WT, wild-type; RRL, rabbit reticulocyte ly- sate.

CHIP28 antibdy fl:500 dilution) (8) or with preimmune (control) se- rum in PBS containing 1% bovine serum albumin for 1 h. %&OEM were rinsed three times with PBS and incubated for 30 min with a fluores- cein-conjugated goat anti-rabbit IgG (Boehringer Mannheim) (1:50) in PBS containing 1% bovine serum albumin. Slides were washed three times with PBS and covered with Dabco (diazabicyclo octane) solution to retard fluorescence quenching and a glass coverslip. Fluorescence was quantified using a Leitz inverted epifluorescence microscope equipped with a fluorescein filter set, a 10 x air objective (numerical aperture 0.4), and a cooled charge coupled device camera (Photomet- ric~). The details of the imaging system, instrument calibration, and background-subtraction software were described previously (27). For each set of oocytes, at least 50 images were obtained from multiple areas of oocytes on at least three separate slides. Background-sub- tracted fluorescence was measured and normalized per unit length of the oocyte plasma membrane.

Cell-free ~ ~ n s l a t ~ n - I n vitro transcribed cRNA was added to a rab- bit reticulocyte lysate (RRL) mkture containing [ ~ S l m e t ~ o ~ n e as de- scribed previously (6). Microsomal membranes prepared from dog pan- creas were added in some experiments at the star t of translation t~ a final concentration of 8 AZm. N-Linked glycosylation in the presence of microsomes was confirmed by translation in the presence of a tripeptide competitive inhibitor of oligosaccharyl-transferase, AcAsn-Tyr-Thr. Translation was performed at 24 "C for 1 h. Samples were analyzed by SDS-polyacrylamide gel electrophoresis, EN3HANCE fluorography (New England Nuclear), and autoradiography.

RESULTS Fig. lA shows the vectors used to transcribe cRNA for oocyte

expression. To enhance expression, the 58-base pair 5"untrans- lated sequence of the Xenopus globin gene (28) was positioned just upstream from the CHIP28 ATG translation initiation site. The dimeric construct encoded a single -57-kDa peptide which contained two consensus sites (N42) for N-linked glycosylation (Fig. 1B). As described under Experimental Procedures, the stop codon of the upstream insert was replaced by a phenylala- nine and followed immediately by the first methionine of the doynstream insert. Previous studies showed that the NH, and COOH termini of CHIP28 are cytoplasmically oriented and that the COOH terminus contains a relatively long stretch of polar residues (1, 18).

The monomeric and dimeric constructs were translated in rabbit reticulocyte lysate in the absence or presence of pancre- atic microsomes (Fig. 2). In the presence and absence of micro- somal membranes, translation of MIP26 produced a single band at 26 kDa, consistent with previous reports that native MIP26 is not glycosylated (29). Translation of monomeric wild- type CHIP28 and the C189W and C189S mutants produced proteins of 28 kDa (non-glycosylated) and 32 kDa (glycosy- lated). N-Linked glycosylation was confirmed by the absence of the 32 kDa band when the acceptor peptide AcAsn-Tyr-Thr was included in the translation reaction mixture. Translation of CHIP28 dimeric constructs produced bands of 57, 60, and 63 kDa, representing the non-glycosylated, singly and doubly gly- cosylated protein, respectively. No monomeric protein was pro- duced. The translated products from the CHIP28-MIP26 (WT- MIP26) chimera underwent only a single glycosylation event as expected. Translation of CHIP28 trimeric and tetrameric con- structs generated chains of 75 and 110 kDa, respectively. How- ever, translation was less efficient, and the products were vari- ably glycosylated in the presence of microsomal membranes.

The cDNA constructs encoding wild-type CHIP28 monomers and multimers were transcribed and the cRNA was expressed in Xenopus oocytes. Fig. 3A shows representative time courses of oocyte swelling in response to dilution of the extracellular buffer with distilled water. Water permeability in oocytes ex- pressing CHIP28 monomers and dimers was strongly in- creased, whereas little functional expression was observed for the trimers and tetramers. Subsequent experiments were therefore carried out with monomers and dimers where expres-

CHIP28 Monomers Function Inokpendently 10419

FIG. 1. cDNA constructs for expres- sion of CHIP28 dimers. A, schematic of vectors showing SP6 promoter, 5'-un- translated sequence of thexenopus globin gene and CHIP28 insertk). See text for details. B, schematic of dimeric protein containing two CHIP28 units in series. The nucleotide and amino acid sequence at the fusion site is shown.

A CHIP28

CHIP28

enhancer gIobGi Nco :*"\'\\\ H I

SP6 insert X

pSP64.CHIP28 pSP64.CHIP28-X (monomer)

BamH I

B

". GAG ATG AAG CCC AAA TTC ATG GCC AGC

E M K P K F M A S amino acid 265 269 1 2 3

sion levels were adequate for quantitative analysis of protein expression and function. Fig. 3B shows the relationship be- tween the amount of cRNA injected and oocyte osmotic water permeability ( P f ) measured at 10 "C. Pf increased linearly with the cRNA amount to 0.25 ng for the CHIP28 monomer and to 0.5 ng for the dimer. The decreased "efficiency" for expression of dimers compared with monomers indicates differences in plasma membrane protein expression and/or single channel water permeabilities.

To relate oocyte water permeability to CHIP28 expression, a quantitative immunofluorescence approach was developed to determine the relative amount of CHIP28 expressed at the oocyte plasma membrane. Oocytes expressing various con- structs were k e d , sectioned and immunostained as described under "Experimental Procedures." All samples were processed in parallel and at least three separate tissue blocks (each con- taining 10-20 oocytes) were prepared for each sample to mini- mize day-to-day and oocyte-to-oocyte variability in sectioning and immunostaining. Oocyte sections were examined as shown in Fig. 4A. A measuring area surrounding a 50-100 p length of membrane was selected, and the background-subtracted in- tegrated pixel intensity was quantified and expressed per unit length of oocyte membrane. All measurements were made in membranes of the oocyte "animal" pole. Fig. 4 shows strong plasma membrane staining in oocytes expressing CHIP28 (A) and C189W ( D ) monomers, and WT-WT ( B ) , WT-C189W ( E ) , and WT-C189S (F) dimers. The low background staining ob- served in the water-injected (control) oocytes (C) was not dif- ferent from that observed in CHIP28-expressing oocytes using preimmune serum (not shown).

The relative water permeability of wild-type and mutant CHIP28 constructs was compared in the next series of studies. Fig. 5A shows representative swelling data. Oocytes expressing the WT-C189W dimer had significant, although less water per- meability than oocytes expressing the WT-WT dimer. Oocytes expressing the C189W monomer had the same water perme- ability as water-injected oocytes even though the C189W pro-

tein was trafficked normally to the oocyte plasma membrane. These results indicate that a non-functional C189W monomer does not inactivate WT CHIP28 in the WT-C189W heterodimer. Thus, the monomeric CHIP28 subunit is likely able to function independently as a water channel.

Quantitative analysis of the functional data for a series of oocytes is provided in Fig. 5B. The ordinate is the increase in oocyte Pf ( A P f ) calculated from the difference between Pf in oocytes expressing the indicated cRNA and Pf in water-injected oocytes. The abscissa is the relative plasma membrane expres- sion of CHIP28 (or mutant) subunits determined by quantita- tive immunofluorescence. Abscissa values have been normal- ized per unit length of oocyte plasma membrane. Background signal (generally ~ 5 % of total signal) obtained from parallel measurements in water-injected oocytes (as in Fig. 4C) has been subtracted. A value of unity has been assigned to oocytes expressing 0.5 ng of wild-type CHIP28 monomer. Fig. 5B shows a linear relationship between A P f and plasma membrane ex- pression for oocytes injected with cRNA encoding WT CHIP28 and C189S monomers and WT-WT and WT-Cl89S dimers. These results indicate similar single channel water permeabili- ties for the WT and C189S monomers when expressed in mo- nomeric or dimeric constructs. In contrast, C189W monomers were expressed at the oocyte plasma membrane but exhibited no increase in P,. In two separate sets of studies, oocytes ex- pressing the WT-C189W heterodimer had 47 = 5% [(A?',Jplasna membrane expression) compared with that of the WT-WT ho- modimer. This values does not differ significantly from 50%. Therefore, a non-functional subunit in a CHIP28 dimeric chi- mera does not affect the function of the wild-type subunit.

MIP26 is homologous to CHIP28 but does not function as a water channel (17). Pf in oocytes expressing MIP26 monomers was not different from that in water-injected oocytes. Pf in oocytes expressing CHIP28-MIP26 heterodimers was 14 2 4 x lo4 c d s ( n = 45). significantly greater than 7.8 2 0.8 x lo4 c d s (n = 38) in control oocytes ( p c 0.02). Thus CHIP28 is functional even when fused with a very different non-water

10420 CHIP28 Monomers Function Independently

A MIP26 CHIP28 C189W C189S

Microsomes - + + I - + + I - + +I- + + Tripeptide - - + - - + - -

1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 + - - +

45 kDb

30 kDD

14 kDD

B WT-WT

Microsomes - Tripeptide - - +

1 2 3

+ + WT-Cl89W WT-Cl89S I - + + I - + + " + - - + 4 5 6 7 8 9

I WT-MIP

60 kDb

45 kDb

30 kDb

14 kD*

C WT-w-WT WT-WT-WT-WT

Microsomes - + + - 1 + + Tripeptide - -

1 2 3 4 5 6 !'*"WP.

+ - - +

92 kD* - 60 kD c

i_ I.

45 kD *

30 kD

14 k D b

(RRL). A, cDNA clones encoding MIP26 (lanes 1-31, CHIP28 (lanes FIG. 2. Translation of cDNAclones in rabbit reticulocyte lysate

4-6), C189W (lanes 7-9), and C189S (lanes 10-12) were translated in RRL in the presence and absence of canine pancreas microsomes as indicated. N-Linked glycosylation of chains resulted in a 3 4 kDa shift in migration as evidenced by the appearance of a 32 kDa band. Glyco- sylation was confirmed by translation in the presence of a tripeptide (AcAsn"Tp-Thr) which competitively inhibits microsomal oligosac- charyltransferase. B, cDNA encoding tandem repeats of WT-WT, WT- Cl89W, WT-C189S, and WT-MIP26 CHIP28 homo- and heterodimers were translated in RRL in the presence and absence of microsomes and tripeptide as indicated. C, clones encoding tandem repeat trimeric (WT- W T - W T ) and tetrameric (WT-WT-WT-WT) CHIP28 coding sequences were translated in RRL to generate -85- and - 100-kDa peptides, re- spectively. In the presence of microsomal membranes (lanes 2,3,5, and 6 ) . chains were variably glycosylated. Inefficient elongation and/or in- ternal initiation of translation resulted in generation of more rapidly migrating chains.

transporting protein. Expression of the CHIP28-MIP26 het- erodimer was too low for quantitative analysis of plasma mem- brane expression.

Studies of mercurial inhibition of a heterodimer containing a mercurial-sensitive monomer ( W T CHIP281 and a mercurial- insensitive monomer (C189S) were carried out. As shown in Fig. f M , addition of HgCl, during the swelling assay strongly inhibited water permeability in oocytes expressing the WT-WT homodimer but had little effect in oocytes expressing the C189S mutant. Intermediate inhibition was observed in oocytes ex-

pressing the WT-C189S heterodimer in which only the WT subunit is sensitive to HgCI,. Percentage inhibition was deter- mined from the initial slope of the swelling curve (before HgCl, addition) and the slope measured at 1-2 min after HgCI, ad- dition. The data obtained for a series oocytes is summarized in Fig. 6B. The increase in Prover that observed in water-injected oocytes (AI',) was strongly inhibited (85-90%) by 0.3 mM HgCI, in oocytes expressing WT CHIP28, WT-WT, and WT-C189W proteins but was not inhibited significantly in oocytes express- ing C189S. AF',. in oocytes expressing WT-C189S protein was inhibited by 44 2 7%, not significantly different from 42.5% (the average between 0 and 85%). This finding strongly supports the conclusion that CHIP28 monomers function independently.

DISCUSSION The purpose of this study was to determine whether CHIP28

monomers functioned independently as water channels. We reported recently that CHIP28 was assembled as tetramers in biological membranes and reconstituted proteoliposomes (20). The strategy adopted in the present study was to examine the function of heterodimers consisting of wild-type and mutant CHIP28 monomers. The association of wild-type and mutant CHIP28 monomers was forced by expression of cDNA chimeric dimers encoding wild-type CHIP28 in series with wild-type or mutant CHIP28. Previous studies in which oocytes were in- jected with cRNA encoding wild-type and mutant CHIP28 monomers (23,24) could not define the functional water trans- porting unit size because association of wild-type and mutant CHIP28 monomers could not be assured. Further, a linear de- pendence of water permeability on the amount of CHIP28 ex- pression gives no information about the functional water trans- porting unit size if all monomers are associated as tetramers. We find here that monomeric subunits in wild-type mutant CHIP28 heterodimers functioned independently based on two experimental approaches: measurement of single channel wa- ter permeability for heterodimers containing CHIP28 and non- functional (C189W) subunit, and HgCI, inhibition of water per- meability in a heterodimer containing CHIP28 and C189S. The results provide strong evidence that CHIP28 monomers func- tion independently and suggest that each CHIP28 subunit con- tains a separate aqueous pore and mercurial inhibition site.

The construction of cDNA chimeras encoding tandem repeat homo- and heterodimers to determine functional subunit size in a membrane-associated multimer was recently introduced for studies of K+ channel function. Tytgat and Hess (25) exam- ined the gating of heterotetramers composed of two or four RCKl K+ channels with different gating properties on a single cDNA. Like CHIP28, both the amino and carboxyl termini of the RCKl channel were cytoplasmically oriented. They re- ported cooperative gating in RCKl channels in which the indi- vidual subunits interacted at a functional level. From similar studies on heterotetramers consisting of TEA sensitive and insensitive RBKl K+ channels, Kavanaugh et al. (30) concluded that bound TEA interacts simultaneously on all subunits. In contrast, our results showed that a mutant CHIP28 subunit affects neither the water transporting function nor the mercu- rial sensitivity of adjacent wild-type CHIP28. However, our data provide no information about whether the membrane as- sembly of CHIP28 in tetramers is required for water transport- ing function and whether isolated CHIP28 monomers exist and are functional. The non-covalent assembly of CHIP28 mono- mers in tetramers probably provides the lowest thermody- namic energy state in membranes.

An important methodological development reported here is the application of quantitative immunofluorescence to measure CHIP28 plasma membrane expression. Oocytes expressing

CHIP28 Monomers Function Independently 1042 1

FIG. 3. Expression of wild-type CHIP28 constructs in Xenopus oocytes.A, time course of oocyte swelling a t 10 "C in response to a 20-fold dilution of the extracellular Barth's buffer with distilled water. Oocytes were microin- jected with 1 ng of cRNA encoding CHIP28 monomers or multimers. Rela- tive oocyte volume was determined by a real-time imaging method (26). B , oocyte P,(S.E., 10-12 oocytes in each group) as a function of the amount of injected cRNA encoding CHIP28 monomers and dimers.

A 1.27 ~

W

5

G 8 1.1

9 W

0

2 W

I-

W U

5 1 .o

0 50 100 150 0 0.1 0.2 0.3 0.4 0.5

TIME (s) INJECTED cRNA (ng)

FIG. 5. Functional analysis of CHIP28 homo- and heterodimers. A, time course of swelling in oocytes injected with water or 1 ng of cRNA encoding WT- WT, WT-C189W, or C189W. Experimental conditions were as in Fig. 3A. B , APr(P,in cRNA-injected oocytes minus P, in water- injected oocytes) as a function of relative plasma membrane immunofluorescence. Data are mean f S.E. for 15-25 oocytes in each group. See text for details.

FIG. 4. Immunofluorescence of oocytes expressing CHIP28 con- structs. Oocyte sections (-6 pm thick- ness) were immunostained with a rabbit polyclonal anti-CHIP28 antibody and a fluorescein-conjugated secondary anti- body as described under "Experimental

cooled CCD camera and hardcopied on a Procedures." Images were recorded by a

video printer. A measuring region has been demarcated in micrograph A. 00- cytes were injected with water (C) or 1 ng of cRNA encoding WT CHIP28 (A),

and WT-Cl89S (F). Scale bar, 40 pm. WT-WT (B) , C189W ( D ) , WT-C189W (E)

A 8

W 1.2- WT-WT 120- CHIP28 A WT-Cl89S

5

P

,..-

6 1.1-

' 0 WT-WT A WT-Cl89W 1 0 0 - 0 C189S

G

g

+/ / & + .. - C189W 20-

W

-,.- A

,,.: yz 80 - ' 1 C189W

r X

0 .... 0 .,.'. .-,'. WT-Cl89W 2 60;

&+ ..,. ,,._,..+- W

- ,-e L- 40 - =r

: /.- ..?. . .j: .*:- 4

.i' .,-. .. ",' ,;, .,?

>I' .E "_.._"." = ,,o i".."" ,"" ..._..- -.-"

water /

I 0 " A I

0 50 100 150 0 0.5 1

various CHIP28 constructs were fmed, sectioned, and immu- nostained in parallel, and plasma membrane fluorescence was quantified by image analysis. A cooled CCD camera was uti- lized for its excellent sensitivity and linear response character- istics. Based on a previous quantitative imaging study of en- dosomal acidification (26), custom C-language software was written to determine the background-subtracted pixel intensityhnit length of oocyte membrane. "he linearity be- tween incremental Pi and plasma membrane immunofluores- cence in WT CHIP28-injected oocytes supported the utility of the approach. However, there were several sources of variabil- ity in the imaging assay, including variations in: protein ex- pression in different oocytes and in different areas of each oocyte, tissue section thickness and geometry, and antibody staining efficiency. For this reason all samples were processed in parallel, measurements were made only from the animal pole in oocytes sectioned near their greatest diameters, and >50

TIME (s) Relative plasma membrane immunofluorescence per unit length

measurements were carried out for each sample using multiple oocytes on separate slides. Notwithstanding these technical concerns, quantitative immunofluorescence provides a practi- cal approach to quantify plasma membrane expression of a foreign protein in Xenopus oocytes for determination of single channel properties.

"here were several assumptions made in the analysis that we believe are justified, yet they cannot be subjected to direct experimental validation. I t was assumed that the anti-CHIP28 antibody stained with equal efficiency wild-type CHIP28, and the C189S and C189W mutants. We believe that this assump- tion is justified because the antibody is primarily directed against the CHIP28 carboxyl terminus (8); the carboxyl termi- nus is not required for function and is located on the opposite side of the membrane from residue C189. I t was assumed that the wild-type mutant heterodimers participate in the formation of membrane-associated tetramers. If, however, the het-

10422 CHIP28 Monomers Function Independently A H9Cb WT-WT B REFERENCES

1 /zz-5:- CHIP28 1. Preston, G. M., and Agre, P. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 11110-

W

"

,./" 11114 , </. Cl89S,,.. 5 ,y'

2. Van Hoek, A. N., and Verkman, A. S . (1992) J. Biol. Chem. 267,18267-18269

1 , C189S )I 3. Zeidel, M. L., Ambudkar, S . V., Smith, B. L., and Agre, P. (1992) Biochemistry

9 31,7436-7440

Biol. Chem. 268.22756-22764 WT-WT -i 4. Ma, T., Frigeri, A,, %ai, S.-T., Verbavatz, J. M., and Verkman, A. S . (1993) J.

WT-Cl89S

0 50 100

%INHIBITION

FIG. 6. Mercurial inhibition of CHIP26 homo- and bet- erodimers. A, time course of swelling in oocytes injected with cRNA encoding WT-WT (1 ng), C189S (0.5 ng), or Wl"C189S (1 ng). At the arrow, HgCl, (final concentration 0.3 m ~ ) was added. B , summary of data (S.E., n = 20-30) for a series of oocytes. Percentage inhibition of incremental P (Mf) was calculated from fitted swelling curve slopes before and 1-4 min after HgC1, addition.

erodimers remained isolated, then the principal conclusion of this study, that CHIP28 monomers functioned independently, would remain valid. It is possible that CHIP28 heterodimers might form large octameric complexes in which wild-type CHIP28 monomers were centrally associated with the mutant subunits at the periphery; however, steric constraints would likely prevent the formation of such large complexes. Even if a multimeric complex did form, the HgC1, inhibition studies of water permeability in oocytes expressing the Wl"C189S heterodimer indicate that monomeric subunits function independently.

We conclude that CHIP28 monomers function independently even though they associate in membranes as tetramers. Each monomer might therefore contain a separate water transport- ing pathway. Analysis of CHIP28 membrane topology (18) and secondary structure content (17) is consistent with the possi- bility that four transmembrane helices in each monomer form a central polar aqueous channel. High resolution structural

5. Preston, B. M., Carroll, T. P., Guggino, W. B., andAgre, P. (1992) Science 266,

6. Zhang, R., Skach, W., Hasegawa, H., Van Hoek, A. N., and Verkman, A. S .

7. Hasegawa, H., Zhang, R., Dohrman, A,, and Verkman, A. S . (1993) Am. J.

8. Sabolic, I., Valenti, G., Verbavatz, J. M., Van Hoek, A. N., Verkman, A. S.,

9. Nielsen, S . , Smith, B. L., Christensen, E. I., Knepper, M. A., and Agre, P. (1993)

10. Bondy, C., Chin, E., Smith, B. L., Preston, G. M., andAgre, P. (1993)Proc. Natl.

12. Wistow, G. J., Pisano, M. M., and Chepelinski, A. B. (1991) IRed Biochem. 11. Verkman, A. S . (1993) Water Chunnek, pp. 74-98, R. G. Landea Co. Austin TX

13. Maurel, C., Reizer, J., Schoeder, J. I., Chrispeels, M. J. (1993) EMBO J. 12,

14. Fushimi, K., Uchida, S . , Hara, Y., Hirata, Y., Marumo, F., and Sasaki, S . (1993)

15. Ma, T., Hasegawa, H., Skach, W., Frigeri, A,, and Verkman, A. S . (1994)Am. J.

16. Hasegawa, H., Ma, T., Skach, W., Matthay, M. M., and Verkman, A. S . (1994)

17. Van Hoek, A. N., Wiener, M., Bicknese, S. , Miercke, L., Biwerai, J., and Verk-

18. Skach, W., Shi, L.-B., Calayag, M. C., Lingappa, V. R., and Verkman, A. S .

20. Verbavatz, J. M., Brown, D., Sabolic, I., Valenti, G., Auaiello, D. A., Van Hoek, 19. Smith, B. L., and Agre, P. (1991) J. Biol. Chem. 266,6407-6415

A. N., Ma, T., and Verkman, A. S . (1993) J. Cell. Biol. 123,605-618 21. Van Hoek, A. N., Hom, M. L., Luthjens, L. H., de Jong, M. D., Dempster, J. A.,

and Van Os, C. H. (1991) J. Biol. Chern. 226,16633-16635 22. Verkman, A. S. , Skorecki, K., and Ausiello, D. A. (1984) Proc. Natl. Acad. Sci.

385387

(1992) J. Cell. Biol. 120, 359-369

Physiol. 264, C2374245

Ausiello, D. A,, and Brown, D. (1992) Am. J. Physiol. 263, C122541233

J. Cell. Biol. 120, 371-383

Acad. Sci. U. S. A. 90,4500-4504

Sci. 16,170-171

2241-2247

Nature 361,549-552

Physiol. 266, C189-C197

J. Biol. Chem. 269,5497-5500

man, A. S . (1993) Biochemistry 32, 11847-11856

(1993) J. Am. Soc. Nephrol. 4,859 (abstr.)

23. Preston, G. M., Jung, J. S., Guggino, W. B., and Agre, P. (1993) J. Biol. Chem. U. S. A. 81,150-154

24. Zhang, R., Van Hoek, A. N., Biwersi, J., and Verkman, A. S . (1993) Biochem-

26. Zhang, R., Logee, K, and Verkman, A. S . (1990) J. Biol. Chem. 266, 15375- 25. Tytgat, H., and Hess, P. (1992) Nature 369,420-423

27. Zen, K, Biwersi, J., Periasamy, N., and Verkman, A. S . (1992) J. Cell Bwl. 119,

268,17-20

istry 32,2938-2941

15378

99-110 data will be required to define the geometry of the aqueous 28. Melton, D. A,, Krieg, P.A., Bebagliati, M. R., Maniatic, T., Zim, K, and Green, channel through CHIP28. 29. Gorin, M. B, Yancey, S . B., Cline, J., Revel, J.-P., and Horwitz, J. (1984) Cell 39,

M. R. (1984) Nucleic Acids Res. 12,7035-7056

Acknowledgment-We thank Dr. Olivier Seksek for assistance in car- 30. Kavanaugh, M. P., Hurst, R. S., Yakel, J., Varnum, M. D., Adelman, J. P., and 4969

rying out the quantitative immunofluorescence measurements. North, R. A. (1992) Neuron 8,493-497