Water channel properties of major intrinsic protein of lens.

The functions of major intrinsic protein (MIP) of lens are still unresolved; however the sequence homology with channel-forming integral membrane protein (CHIP) and other Aquaporins suggests that MIP is a water channel. Immunolocalizations confirmed that Xenopus oocytes injected with bovine MIP cRNA express the protein and target it to the plasma membrane. Control oocytes or oocytes expressing MIP or CHIP exhibited small, equivalent membrane currents that could be reversibly increased by osmotic swelling. When compared with water-injected control oocytes, the coefficient of osmotic water permeability (Pf) of MIP oocytes was increased 4-5-fold with a low Arrhenius activation energy, while the Pf of CHIP oocytes increased > 30-fold. To identify structures responsible for these differences in Pf, recombinant MIP proteins were expressed. Analysis of MIP-CHIP chimeric proteins revealed that the 4-kDa cytoplasmic domain of MIP did not behave as a negative regulator. Individual residues in MIP were replaced by residues conserved among the Aquaporins, and introduction of a proline in the 5th transmembrane domain of MIP raised the Pf by 50%. Thus oocytes expressing MIP failed to exhibit ion channel activity and consistently exhibited water transport by a facilitated pathway that was qualitatively similar to the Aquaporins but of lesser magnitude. We conclude that MIP functions as an Aquaporin in lens, but the protein may also have other essential functions.

W ater C hannel P ro p erties o f M ajor In trin sic P rotein o f Lens* (Received for publication, December 27, 1994) S a b i n e M . M u l d e r s t , G r e g o r y M . P r e s t o n § H , P e t e r M . T . D e e n i , W illia m B . G u g g in o ll, C a r e l H . v a n O s t* * , a n d P e t e r A g re § H $ t h a n n e l a c t i v i t y a n d c o n s i s t e n t l y e x h i b i t e d w a t e r t r a n s p o r t b y a f a c i l i t a t e d p a t h w a y t h a t w a s q u a l i t a t i v e l y s i m i l a r to t h e A q u a p o r i n s b u t o f l e s s e r m a g n i t u d e . W e c o n c l u d e t h a t M IP f u n c tio n s a s a n A q u a p o r i n i n le n s , b u t t h e p r o t e i n m a y a ls o h a v e o t h e r e s s e n t i a l f u n c t i o n s .
Major intrinsic protein (MIP)1 is a 26-kDa protein expressed exclusively in lens fiber cells where it comprises over 60% of the membrane protein. The cDNA encoding MIP was isolated from a bovine lens cDNA library, and hydrophobicity plots predicted that MIP is an integral membrane protein with cytoplasmic amino and carboxyl termini and six bilayer-spanning domains (Gorin et a l , 1984). MIP reconstituted into liposomes exhibits voltage-dependent channels permeable to ions and small mol ecules that may be closed by Ca2H _ and calmodulin (Nikaido and * This study was supported by the Dutch Science Foundation (NWO-SLW-810-405-16.2) and by National Institutes of Health Grants HL33991, HL48268, and DK32753. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked * * advertisement" in accordance with 18 U.S.C, Section 1734 solely to indicate this fact.
* 1 The abbreviations used are: MIP, major intrinsic protein of lens; Pf , coefficient of osmotic w ater permeability; CHIP, channel forming inte gral protein of 28-kDa (CHIP28). Rosenberg, 1985;Girsch and Peracchia, 1985;Peracchia and Girsch, 1989;Shen et a l., 1991), In planar lipid bilayers, MIP forms single channels of various conductances (up to 3 nanosiemens) and a weak selectivity for anions (Ehring e t a l., 1990); Nodulin-26, a homologous protein from soy bean root nodules, was recently shown to behave similarly (Weaver e t a l , 1994). In spite of these studies with reconstituted MIP protein, studies of MIP in native membranes or expressed in oocytes have failed to confirm a physiologic function for MIP. For example, although MIP was initially considered to be the gapj unction protein, immunological, biochemical, and electrophysiological studies failed to identify electric coupling of lens fiber cells via MIP (Swenson et a l , 1989).
MIP was the first identified member of an ancient family of membrane proteins from diverse organisms that now includes more than 20 members. Several MIP family members from animal and plant tissues were shown to function as waterselective membrane channels and are now referred to as the "Aquaporins" (Chrispeels and Agre, 1994). The purification and cDNA cloning of the 28-kDa channel-forming integral mem brane protein, CHIP (Denker et a l , 1988;Preston and Agre, 1991), a MIP homolog from red cells and renal proximal tu bules, permitted the first demonstration of a molecular water channel (Preston et a l , 1992). CHIP is now designated Aqua porin-1 (AQP1), and cDNAs encoding four other Aquaporins have subsequently been isolated from diverse mammalian tis sues (Fushimi et a l., 1993; Ishibashi e t a l , 1994; Ma et a l , 1994; Echevarria et a l., 1994; Hasegawa e t a l , 1994; Jungetf a l , 1994b; Raina et a l , 1995).
The hourglass model was recently proposed to merge struc tural and functional features of the Aquaporins (Jung et a l , 1994a). The model describes two tandem repeats with the first and second half of the molecule oriented at 180° to each other, and each half contains the sequence asparagine-proline-ala nine (NPA) in intracellular loop B and extracellular loop E. In the hourglass model, it was predicted that loop B and loop E fold back into the membrane, forming a single water-selective pore. The striking sequence homology with the Aquaporins suggests that MIP may also be involved in water movement through the lens fiber cell membranes. In this report, we in vestigated the water transporting capacities of MIP using the X en o p u s oocyte expression system, and water channel proper ties qualitatively similar to the Aquaporins were observed.

EXPERIMENTAL PROCEDURES
Plasm id DNA Mutagenesis, DNA Sequencing, and in Vitro R N A Synthesis-Standard molecular procedures were used (Sambrook et al., 1989). The bovine MIP coding sequence (Gorin et a l, 1984) flanked 5' and 3' by Xenopus j3-globin gene untranslated sequences (Swenson et a l, 1989) was transferred into pBluescript IIKS (Stratagene) following digestion with H m dlll and X b a l. The CHIP expression vector was constructed as described (Preston et a l , 1992). These constructs served as templates for site-directed mutagenesis reactions using the Muta-Gene phagemid in vitro mutagenesis kit (Bio-Rad). Table I    This chimorie protoin vectors CI-IIP-MIP-1 and MIP-CHIP-1 were conntrucLod by inserting a B a m B l restriction site at Val-112 of MIP and Thr-120 of CHIP, resulting in the insertion of the amino acids aspartic acid and proiino. MIP(V112Bam) and CHIP(T120Bam) were digested with I k w i l l l the 5'-lmlf of CHIP was ligated to the 3'-half of MIP (OHIP-MIP-1), and the 5'-half of MIP was ligated to the 3'-half of CHIP (MIP-OIIHM.). The chimeric protein vectors MIP-CHIP-2 and CHIP-MIP-2 worti constructed using the megaprimer polymerase chain reac tion method (Bunk and Galinski, 1991) exchanging at Arg-226 of MIP and Arg»234 of CHIP, using the following antisense primers: MIP-CII IP-2, 5 '-GTCTGTGAGGTCACTGCTG-CGAGGGAAGAGGAGA-AAG-3'; ClIIiP-MIP-2, 5'-CTCAGAAACACTCTTGAGC-CGTGGGGCc a (k ;a t g a a g -3'. The chimeric protein, MIP-CHIP (loop A), contains the first exofacial loop (A.) of CHIP (amino acids 34-51) in place of the first exofacial loop of MIP (amino acids 33-43) and was constructed with an 82-base pair msortionul-substiLuLional oligonucleotide primer (not shown) in a sitedim*tod mutagenesis reaction. All mutations were confirmed by enzy matic nucleotide sequencing (U, S. Biochemical Corp.).
Cupped UNA transcripts were synthesized in vitro using T3 RNA polymerase with X baIndigested MIP, CHIP, or mutant expression vector DNA, and the UNA was purified as described (Yisraeli and Melton, 1989).
Preparation o f Oacytvs and Measurement of Pf-Female Xenopus lat'oia were anOHthetized on ice, and stage V and VI oocytes were removed and prepared (Lu et a l } 1990). The day after isolation, oocytes were injected with either 50 nl of water or 0.5-25 ng of cRNA in 50 nl of water. Injected oocytes were maintained for 2-3 days at 18 °C prior to osmotic swelling, membrane isolation, or voltage clamp experiments. Oocyte swelling was performed at 22 °C following transfer from 200 mosM (osmhl) to 70 niosM (osmmit, CHIP) or either 70 or 20 mosM (osmout, MIP) modified Barth's solution diluted with water, Sequential oocyte images were digitized at 5-s intervals for a total of 1 min, and the volumes of the sequential images wore calculated as described (Preston vt «/,, 1993). The change in relative volume with time, diVfV^fdt, was fitted by computer to a quadratic polynomal, and the initial rates of swelling were calculated. The osmotic water permeability (Pn pm/s) was calculated from osmotic swelling data between 5 and 10 s, initial oocyte volume (V,, • 9 X 10 4 cm0), initial surface area (S = 0.045 cm2), and the molar ratio of water (V" « 18 cm'VmolXZhangef a l, 1990) using the formula Oocytc. Membrane Isolation and Immunohlot Analysis-Total oocyte membranes (Preston et a l t 1993) and plasma membranes (Wall and Patel, 1989) were isolated from groups of 4-30 oocytes, solubilized in 1.25% (w/vi SI)S at 60 °C for 10 min, electrophoresed into 12% SDSpolyacrylamide gels (Laemmli, 1970), transferred to nitrocellulose (Davis and Bennett, 1984), incubated with a 1:10,000 dilution of anti-MIP antibody, or a 1:1000 dilution of anti-CHIP antibody (Smith and Agre, 1991), and visualized using the ECL Western blotting detection system (Amersham Corp.). Molecular weights were determined relative to the mobility of prestained SDS-PAGE standards (Bio-Rad).
Anti-M IP Antibody-A synthetic peptide corresponding to the 15 Fig. 1. M embrane topology of MIP show ing sites targ eted for m utagenesis and domain exchanges in this study. SPLICE 1 is the splice site for MIP-CHIP-1 and CHIP-MIP-1; SPLICE 2 is the splice site for MIP-CHIP-2 and CHIP-MIP-2. The proline introduced in CHIP at position Asp-131 is indicated with #, NPA motifs are indicated with asterisks. Filled symbols indicate amino acids that are conserved in more than half of the members of the MIP family from bacteria, yeast, plants and animals (Reizer et al., 1993).
carboxyl-terminal amino acids of bovine MIP (PEVTGEPVELKTQAL) (Gorin et a l , 1984) was coupled to keyhole limpet hemocyanin. Rabbits were injected with 400 fig of conjugated synthetic peptide mixed with Freund's complete adjuvant. After 4 weeks, and every 3 weeks there after, the rabbits were boosted with 200 pg of conjugated synthetic peptide mixed with incomplete Freund's adjuvant. The produced anfci-MIP antiserum was tested for specificity and cross-reactivity by an enzyme-linked immunosorbent assay.
Immunolocalization of MIP in Xenopus Oocytes-Oocytes were fro zen for immunofluorescence microscopy in Tissue-Tek mounting me dium (Miles Inc.). Sections of 5 jam were cut by cryostat, collected on gelatin-coated slides, and fixed at room temperature for 5 min in 1% (w/v) periodate-lysine-paraformaldehyde fixative (McLean and Nakane, 1974). Sections were washed 3 times with TBS (0.9% NaCl, 25 mM Tris, pH 7.4) and incubated overnight at 4 °C with anti-MIP antibody at a dilution of 1:1,000 in TBS containing 10% (v/v) goat serum. After 3 washes with TBS, the sections were incubated for 1 h at 37 °C with affinity-purified fluorescein isothiocyanate-labeled goat-anti-rabbit IgG (Sigma Immuno Chemicals) at a dilution 1:50 in TBS containing 10% (v/v) goat serum. After 3 more washes in TBS, the sections were em bedded in Mounting Medium (Sigma) and analyzed by immunofluores cence microscopy.
Electrophysiology-Studies were carried out as described (Preston et a l, 1992) using a two-microelectrode voltage clamp with Clampex soft ware. Water-and cRNA-injected oocytes were voltage-clamped at a holding membrane potential of -70 mV, repolarized to -90 mV, and

SWOLLEN SHRUNK BACK TO ORIGINAL VOLUME
Fk;, *A. M em brane ion conductance. Voltage clamp was used to measure currents at voltages from -90 mV to +50 mV of water-injected control oocyte«, ooeyten injocted with 25 ng of MIP cRNA, or oocytes injected with 0.5 ng of CHIP cRNA. The currents were determined in 200 mosM modified BarLli'a Holution, The oocytes were swollen by replacing modified Barth's solution with a hypotonic medium, and current measurements wore repeated, Control oocytes wore placed in HaO, and the current was measured after 30 min. Oocytes expressing MIP and CHIP were placed in modified Barth'« notation diluted from 200 to 70 mosM, and currents were measured after 20 and 30 s. Oocytes were shrunk back to their normal volume by replacing thu hypotonic medium with isotonic modified Barth's solution, and currents in oocytes expressing MIP and CHIP were meanured niter 8 min.
lion MIP(AIHIO) did not confer mercury sensitivity to MIP (Fiji, 7). The cysteine at position 14 in the amino-terminal domain of MIP was replaced by a valine, since other Aquaporinn do not contain a cysteine at this position, but the Pf of 014V mutant MIP was not different from wild-type MIP (Fig.  8*4). MIP contains a proline in the second extracellular loop (residue 123 in loop C), a site where other members of the MIP family do not contain this residue. Introducing a proline at this petition in CHIP (D131P) also did not change the Pr (Fig. 8A).
All known Aquaporine except AQP4 contain a proline in the 5th tranmnembrane segment; no proline exists in the 5th transmembrano domain of MIP. When a proline was introduced in MIP (V160P), tho substitution reproducibly enhanced the Pf by 60 ± 20% relative to the P^of oocytes expressing wild-type MIP (Fig. HA). Notably, the amount of MIP protein expressed in oocytes was not different after injection of equal amounts of MIP cRNA and MIPIV160P) cRNA (Fig. 8B). Although not large, this increase in was confirmed in each of five different experiments. A "gain of function" mutation suggests that theP^ of MIP can be increased by introducing subtle conformational changes in the molecule. Nevertheless, replacing the proline in CHIP at this position (P169A) to the corresponding residue of MIP had no measurable influence on the Pf of CHIP (Fig. 8A).
Those results show that MIP exhibits a low Pr value that can be increased by introducing subtle changes in the molecule.
One explanation may be that MIP needs to undergo an activa tion or structural rearrangement to function as a water chan nel, whereas CHIP is constitutively in the activated state. Two sites in the carboxyl-terminal cytoplasmic domain of MIP are phosphorylated in vivo (Lampe and Johnson, 1990). Since these sites are not conserved in other members of the MIP family, the sites may play a role in regulation of MIP function. Mutation of the two putative phosphorylation sites to the corresponding residues in CHIP (S243V, S245E) slightly decreased the Pf (Fig. 8A), but the amount of protein expressed was comparably reduced as compared with wild-type MIP by immunoblot (Fig.  8B). The possibility of stretch-activation was also considered, but if it exists, it was not reproduced by simple increase in volume, since the rate of osmotic swelling did not increase with time (Fig. 9). Thus, if an activation step confers CHIP-level water permeability on MIP, the identity of this step remains unknown.

DISCUSSION
Even though the cDNA encoding MIP was cloned more than a decade ago (Gorin et a l, 1984), the physiological roles of the protein are still not understood. Detailed studies of the homol ogous proteins MIP (Ehring et a l , 1990  al.> 1994), and preliminary studies with CHIP2 revealed large voltage-dependent conductances when reconstituted into pla nar lipid bilayers. Nevertheless, the magnitude of the currents in bilayers containing MIP was far above the conductances measured in normal lens (Mathias et aL, 1979(Mathias et aL, , 1991. There fore, while the planar lipid bilayer studies are highly reproduc ible, their relevance to normal lens physiology remains uncer tain. Moreover, the studies reported here (Fig. 3) and preliminary studies of other investigators (Kushmerick et aL, 1994) failed to confirm ion conductance hy MIP expressed in oocytes.
Soon after the discovery that CHIP is a molecular water channel (Preston et aL, 1992)> several labs evaluated MIP and other homologous proteins for osmotic water permeability. One group reported that MIP is not a water channel (Verbavatz et aL, 1994), but apparently these investigators used the same techniques with which they failed to detect osmotic water per-J. Iiall and P. Agre, unpublished observations. meability of AQP3 , a protein found by two other groups to exhibit high Pf comparable with the other Aquaporins (Ishibashi et al.t 1994;Echevarria et a/., 1994). The studies reported here document that oocytes expressing MIP exhibit osmotic water permeabilities 4-5-fold above control oocytes (Fig. 4) with activation energies identical to the Aqua porins (Fig. 5). These observations are supported by prelimi nary studies of other investigators who also found an increase in Pfof oocytes expressing frog MIP (Kushmerick et aL, 1994) or bovine MIP (Chandy et aL, 1995). MIP may contribute to the maintenance of lens transparency by enhancing uptake of in tercellular water by adjacent lens fiber cells. The narrow geo graphic separation of lens fiber cells (which contain MIP) and lens epithelial cells (which contain CHIP) suggests functional cooperativity. Osmotic gradients provide the driving force for Aquaporin-mediated water transport in kidney and most other tissues (Nielsen et aL, 1993a(Nielsen et aL, , 1993b. It is likely that hydro static forces move water through CHIP in endothelium of the proximal capillary bed (Nielsen et aL, 1993b), and a related process may occur through MIP when the lens shape is rapidly altered by contraction of muscles in the ciliary body to provide fine focus of the corneal image upon the retina.

1/T (i/Kx ltn
A molecular explanation for why MIP is a weaker water   Water Channel Properties of MIP plasma membranes with wavy junctions similar to lens fiber cells (Zampighi et aL, 1989), and if this is important to the function of MIP, the water permeability studies may yield spuriously low values in oocytes. Similar to other bilayer-spanning proteins, MIP may have multiple physiological functions, and it is possible that the primary role of MIP is not water transport. The red cel] band 3 protein is the membrane anion exchanger (AE1), a cytosolic regulator of glycolytic enzyme activity, and the structurally important attachment site for ankyrin on the membrane (for review, see Low (1986)). MIP may also have a structural func tion, since the protein has been shown to enhance adhesion with membranes containing negatively charged phospholipids (Michea et a l, 1994). Also, it is known that some proteins are expressed in lens where their function is unrelated to their functions in other tissues [e.g. crystallins, for review, see Piatigorsky and Wistow (1989) and De Jong et a l (1994)). Thus the extremely high expression of MIP in lens may be far above the level needed for water permeability, since the abundance may be needed for an unrelated function.
A mutation has been identified in mice that may provide insight into other potential functions of MIP, since these mice develop cataracts prior to birth (Muggleton-Harris et a l , 1987). The cat mouse mutation results in lower abundance of MIP mRNA with the major transcript being truncated, and MIP was not detectable in lens by immunocytochemistry (Shiels and Griffin, 1993). The mutation is expressed as a dominant trait and has been mapped to the distal end of chromosome 10 (Muggleton-Harris et a l , 1987), coincident with the MIP locus (Griffin and Shiels, 1992). Mutations in genes encoding struc tural proteins usually produce dominantly inherited disorders, whereas mutations in transporters such as the CFTR are recessively inherited. Consistent with this, mutations in the AQP2 gene were recently identified in homo zygotes and a com pound heterozygote with severe nephrogenic diabetes insipi dus, while the heterozygous relatives were unaffected van Lieburg et a l , 1994). Careful histological analysis of the early stages of disease in the cat mouse may provide clues to the critical function of MIP, which is the first defect leading to the development of cataracts in this animal model.