Background of the Invention

Aqueous fluids account for approximately half of a normal adult's body weight. These fluids contain oεmotically active solutes. The proper concentrations of solutes within bodily fluids are maintained within narrow limits despite large variations of both solute and water intake by changes in the volume of water excreted per day. Thus, proper renal processing of aqueous fluids, including modulation of water excretion, is critically important to the maintenance of good health.

Renal processing of the body's solutes and water content first involves filtering of blood at glomeruli to separate retained blood cells and proteins from filtered solutes and water. The majority of the filtered solutes and water are returned to the body's circulation via selective absorption by renal tubules. In the proximal portion of renal tubules, water reabsorption occurs as a result of active reabsorption of solutes. In contrast, in distal portions of the tubules, solute and water reabsorption occur by separate processes. When an excess

of body water is present, there is reabsorption of body solutes and excess water flows through the distal nephron to the bladder as dilute urine. In periods of dehydration, water is osmotically reabsorbed such that a concentrated hypertonic urine is formed. Osmotic reabsorption of water in the distal nephron segment, called the collecting duct, is modulated by antidiuretic hormone (ADH also referred to as vasopressin) .

Changes in collecting duct water permeability are accomplished through control of the water permeability of the apical membranes of epithelial cells that line this segment. ADH causes the insertion of water channels into epithelial cell apical membranes. These water channels were originally contained in vesicles within the cytoplasm of these epithelial cells. ADH causes water channel insertion by fusion of the water channel- containing vesicles with the apical membrane. Removal of the ADH stimulus causes removal of water channels from the apical membrane by retrieval of the water channel- containing membrane into the epithelial cell cytoplasm. A variety of diseases are associated with retention of excess body water. These include, for example, liver failure, heart disease and syndrome of inappropriate ADH secretion (SIADH) . Conversely many diseases are associated with the bodies inability to retain water.

This inability can lead to life threatening dehydration and severe electrolyte imbalance. These diseases are difficult to manage therapeutically because of the kidney's disassociation of solute and water reabsorption, as discussed above.

One of the major problems preventing effective removal of excess body water in these diseases is the lack of ability to selectively block renal water reabsorption. Presently, these diseases are treated with a variety of diuretic agents. However, currently

available diuretics block renal tubular water reabsorption by inhibiting tubular solute reabsorption. This is accomplished by altering the flux of salt ions across the membranes of various segments of the renal tubules. Thus, in order to remove excess body water, there is an obligatory loss of large amounts of body solutes that results in the depletion of body ionic stores, especially sodium, potassium and chloride ions. It is apparent, therefore, that there has been a longstanding need for better and more selective means of controlling the body's water regulation. Despite the need, it has been extremely difficult, or impossible, to screen compositions for their ability to selectivity alter water flow across ADH water channels. This is largely due to the fact that water flow through the water channels and lipid bilayer is so rapid that it nearly precludes any practical way of measuring such flow.

Summary of the Invention

This invention relates to Applicants' discovery that ADH water channels in vesicles derived from renal apical membranes are permeable to protons. As a result, the ability of a candidate composition to alter the permeability of ADH-elicited water channels can be assessed by determining the effect that the composition has on the proton permeability of ADH water channels. One embodiment of the invention relates to a polypeptide which significantly alters proton permeability of ADH water channels. This alteration can include one which significantly inhibits or increases the proton permeability of the ADH water channels. The polypeptide can be a diuretic which inhibits ADH-elicited water channels, preferably one which selectively inhibits the ADH-elicited water channels. Further it can be an antibody, including polyclonal sera, a monoclonal

antibody or a biologically functional fragment of an antibody directed to an ADH water channel associated protein. An example of such an antibody is an anti-55/53 kD antibody. In a still further embodiment, the composition can be a peptide which binds to an ADH associated protein.

In another embodiment, the invention provides a method for therapeutically treating a patient having a disease associated with excess or depleted body water. This method comprises administering to the patient a therapeutically effective amount of a composition which is capable of altering the permeability of the ADH- elicited water channels.

Brief Description of the Figures Figure 1 is a plot of relative fluorescence versus time illustrating the effect of two temperatures on 2 pH change in F-dextran loaded vesicles containing ADH water channels.

Figure 2 is a plot of relative fluorescence versus time illustrating the effect of buffer concentration on the rate of pH change in F-dextran-loaded vesicles containing ADH water channels.

Figure 3 is a plot of relative fluorescence versus time illustrating the effect of parachloromercuribenzene sulfonate (pCMBS) on the rate of pH change as detected by fluorescence in F-dextran loaded vesicles containing ADH water channels.

Figure 4 is a plot of relative fluorescence versus time illustrating the symmetry of inward and outward proton fluxes in F-dextran loaded vesicles containing ADH water channels.

Figure 5 is an Arrhenius plot illustrating the effect of varying temperature on vesicle proton permeability (P _H+ ) •

Detailed Description of the Invention The subject invention relates to compositions and methods for modulating the permeability of an ADH- elicited water channel. More specifically, modulation of permeability refers to a significant alteration in the flow of water through the aqueous pores that constitute the ADH water channels of renal apical membranes. Such alteration could occur through a number of mechanisms, including direct blockade of the channels as well as modification of proteins or lipids that in turn change the structure of the channels to decrease their permeability to water.

Endocytic vesicles containing ADH water channels can be obtained from a variety of sources. These include, for example, the homogenized papillas of kidneys from Brattleboro rats. New Zealand white rabbits and domestic cattle. A convenient source for ADH water channel- containing vesicles is the intracellular vesicles derived from the apical membranes of ADH-stimulated toad urinary bladders. The membranes of these retrieved vesicles are derived from vesicles that store high concentrations of ADH water channels. These vesicles, called aggrephoreε, are large storage vesicles located immediately beneath the apical membrane. Techniques for obtaining a partially purified vesicle fraction containing such vesicles have been published. See Harris, ^' H. W. et al., J. Me b. Bio. 96:175-186 (1987) , the teachings of which are incorporated by reference.

Suitable vesicles could also be prepared by other techniques. For example, ADH water channel components

could be incorporated into artificial liposomes containing an entrapped fluorophore.

The ADH water channel-containing vesicles are suspended in an aqueous medium sufficient to allow dissociation of protons (H ⁺ ) and anions, such as hydroxyl (OH") ions, from their parent compositions. A wide variety of aqueous solutions, including pure water, can be used in this assay.

Compositions can be tested to determine their ability to alter the permeability of ADH-elicited water channels. These compositions preferably comprise polypeptides which bind to components of the ADH water channel. These polypeptides include, for example, proteins, polyclonal antisera, monoclonal antibodies or biologically functional fragments of same. The distinction in size between peptides and polypeptides is not always uniform in the literature. Therefore, as used herein, polypeptide means any oligomer of two or more peptides. The modulating effects of- a composition can be tested, for example, using a suspension of vesicles containing ADH-elicited water channels. Therefore, the polypeptide to be tested must be sufficiently soluble in an aqueous medium to allow a sufficient amount to dissolve so that the modulating potential of the composition can be determined. Incubation conditions and duration will depend on the nature of individual composition to be tested.

More specifically, the vesicles containing ADH- elicited water channels are suspended in an aqueous medium, preferably one that is stirred continuously. Proton flux through the ADH water channels in these vesicles is created by establishing a significant proton concentration differential between the suspending mediuir.

and intravesicular medium. This concentration differential establishes a driving force to create proton flux through the water channels if the water channels remain proton permeable. The proton concentration differential can be created conveniently by rapidly changing the pH of the aqueous suspending medium compared to the intravesicular medium.

For example, acid can be added directly to the suspending medium to lower the pH thereby providing a suspending medium having a significantly higher concentration of protons than the intravesicular medium. Alternatively, an aliquot of alkali, such as sodium hydroxide, could be added to the suspending medium to raise the pH thereby decreasing the proton concentration of the suspending medium compared to the intravesicular proton concentration. Alternatively, the proton concentration of the extravascular medium could be altered by initiation of a chemical reaction that produces or consumes protons. Because proton flux is rapid, it is preferred to detect a change in intravesicular volume by rapid detection means, such as optical, nuclear magnetic resonance (NMR) or certain dyes. The experimental work described herein was conducted using a pH-senεitive fluorophore, fluorescein, which was bonded to dextran to assist in retaining the fluorophore within the water channel-containing vesicles. This technique has been described in literature. See for example, Harris, H. W. , Wade, J. B. and Handler, J.S., "Fluorescent Markers to Study Membrane Retrieval in Antidiuretic Hormone-Treated Toad Urinary Bladder", published in Am. J. Phvsiol. 251.C274-C284 (1986) , the teachings of which are incorporated by reference. Fluorescein is a preferred pH-sensitive fluorophore because its intravesicular fluorescence correlates linearly with pH over the pH

range of 6.0 to 8.0, permitting accurate estimation of intravesicular pH.

In order to slow the rate at which intravesicular pH changes, a buffer is employed. The buffer is loaded into the vesicles in a manner similar to the loading of the pH-sensitive fluorophore or other pH indicator. Suitable buffers have a group with a dissociation constant (pKa) within the physiological pH range. Examples of suitable buffers include HEPES, Tris and sodium phosphate buffers. Since proton permeability has been found to be closely related to water permeability of ADH water channels, the detection of proton permeability serves as a convenient measure of water permeability. The ability of a particular composition to alter the permeability of ADH water channels is assessed by paired experiments in which one aliquot of vesicles is treated with the candidate composition and the other aliquot serves as its untreated paired control. The rate of proton equilibration across these two groups of vesicles is compared after each is subjected to an identical pH gradient. The rate of proton equilibration is recorded and the data analyzed to derive an equation that describes these rates. The rates of these two groups of vesicles are then compared to determine if the candidate composition alters proton flow across the membranes of these vesicles. If it does, it will also alter water permeability.

It is desirable, of course, that candidate compositions possess additional properties beyond their ability to alter the proton and water permeability of ADH water channels. Particularly preferred compositions will have a selective effect in altering the permeability of ADH-elicited water channels. That is, such selective compositions will alter ADH-elicited water channels without significantly effecting other renal reabsorption

processes, such as reabsorption of solutes such as sodium, potassium and chloride ions. Candidate compositions found to alter ADH-elicited water channels can be tested for their effects on other renal processes by employing perfused renal segments, and/or renal epithelial cells. Such techniques are well know to those skilled in the art. See, for example, Hoch, B. , Gorfein, P. C, Linzer, D. , Fusco, M. J. and Levine, S.C., "Mercurial Reagents Inhibit Flow Through ADH-Induced Water Channels in Toad Bladder" , Am. J. Phvsiol.

256.F948-F953 (1989), the teachings of which are hereby incorporated by reference.

Candidate compositions should also be non-toxic, as defined by their lack of effects on other bodily transport systems, cell and organ machinery or higher physiologic functions. The presence or lack of such effects can be tested in live mammals, such as Brattleboro rats, by techniques known to those skilled in the art. Examples of the kind of candidate compositions encompassed by this application include polypeptides, particularly antibodies, which bind to an ADH water channel associated protein. Comparison of SDS-PAGE protein bands in either unstimulated or ADH-stimulated granular cell apical membranes has suggested that protein bands of 55, 53, and 17 kDa are ADH water channel components of the toad bladder granular cell apical membrane. Methods for isolating these proteins is described in detail below. As is also presented in the Examples which follow, ho ologs of the 55' kDa and 53 kDa ADH water channel proteins of toad cells exist in purified mammalian water channel containing vesicles.

Antibodies in this context is intended to include polyclonal sera, monoclonal, single chain and chimeric antibodies. Further, fragments of an antibody which

retain the capacity to alter the permeability of the ADH water channels are included. Antibody fragments are a preferred embodiment of the invention due to their increased ability to filter through renal tubules. Antibodies can be derived through techniques known to those skilled in the art, such as immunization of mammals, for example rabbits or goats, with an ADH-water channel protein and, in the case of monoclonal antibodies, the production of hybridoma cell lines from antibody secreting lymphocytes.

Further, one can make peptides through standard peptide synthesis methods. Such peptides would mimic the binding site of an antibody which has the capacity to alter the permeability of the ADH water channels, thus the peptide would also have the capacity to alter the permeability of the ADH water channels. One can also make small molecules which have the capacity to bind to binding sites on the ADH water channel associated proteins. Candidate compositions found to alter the ADH- elicited water channels can be used to treat patients having a disease associated with excess or decreased body water. Such treatments will involve the administration to the patient of therapeutically effective amounts of the composition capable of altering ADH-elicited water channels. In most cases, it will be preferred to employ such a composition which is also selective in its alteration of ADH-elicited water channels. Thus, the selective composition will alter ADH-elicited water channels without having any significant effect on other renal processes and which is non-toxic, as defined above.

Administration of such compositions can be by medically accepted techniques, including intravenous, enteral, etc. Appropriate amounts or dosages will vary from individual to individual and by disease, of course.

Appropriate dosages can be calculated by those skilled in the art taking such factors into account.

The invention is further illustrated by the following specific examples.

EXAMPLE 1

Proton Permeability at 5° and 25°

Endocytic vesicles from the bladder of Dominican toads were ADH simulated and loaded with fluorescein bound to dextran (F-dextran) following the procedures of Harris et al. , "Apical Membrane Vesicles of ADH-

Stimulated Toad Bladder Are Highly Water Permeable," Am. J. Phvsiol. 2_58_:F234-F243 (1990); Harris, H.W., et al.. "Fluorescent Markers to Study Membrane Retrieval in Antidiuretic Hormone-Treated Toad Urinary Bladder," Am. J. Phvsiol.. 251.C274-C284 (1986). Briefly, the procedures were as follows.

Dominican toads (Bufo marinus) , obtained from National Reagents, Bridgeport, Connecticut, were employed. Urinary bladders of Bufo marinus were prepared and mounted on cannulaε as small sacs. After a 20 minute interval of ADH stimulation (50 mU/ml) , the apical surfaces of bladders were exposed to solutions containing 50 mg/ l of F-dextran (Av. Mol. wt. 70,000 Daltons) and 0-20 mM NaCl, 30 ιr_M KC1, and 1-20 mM HEPES, pH 8.0, for an interval of 5 minutes followed by termination of ADH stimulation and incubation for an additional 10 minutes. The solution containing the F-dextran was then removed, the bladders rinsed to remove all extracellular F- dextran. The intermediate pellet fraction obtained from centrifugation of the cell was used. F-dextran loaded vesicles were washed repeatedly to remove F-dextran and partially purified by differential centrifugation. In

the absence of ADH, not detectable fluorescence was present in this vesicle fraction despite apical membrane exposure to F-dextran.

Standard cuvette fluorescence measurements were performed using an SLM-Aminco 500C spectrofluorimeter (excitation wavelength 499 nm, emission wavelength 520 nm, slits 2 nm) . Stopped-flow measurements were performed in a High Tech stopped-flow device (50 msec dead time) connected to a Photon Technologies Alphascan Fluorimeter. One syringe contained a concentrated vesicle suspension and antifluorescein antibody while the other contained buffer with sufficient HC1 to change the final extravesicular pH as indicated. Data from 3-5 determinations performed on a single day were averaged. Results were fitted to single exponentials and τ was converted to permeability using the following equation:

J _H ⁺ = (PH+) ^X (ΔC) x (A)

where J _H + is the flux rate of protons in mole/sec, P _H + is the permeability coefficient of protons in cm/sec, ΔC is the concentration gradient for protons across the vesicle membrane at the start of the experiment in mole/cm ³ , and A is the surface area of a single F-dextran containing vesicle (7.1 x 10' ¹⁰ cm ² ). See Harris, H.W. , et al.. "Isolation and Characterization of Specialized Regions of Toad Urinary Bladder Apical Plasma Membrane Involved in the Water Permeability Response to Antidiuretic Hormone," J. Memb. Biol. , 9_6.:175-186 (1987). J _H + was calculated by multiplying 1/τ by the amount of buffer (in moles) in an individual vesicle (single vesicle volume was 1.41 x 10" ¹ " cm ³ ) . T is the mathematical term representing the time for the process to be 2/3 completed. Estimates of vesicle volume and surface area were calculated from

di ensions of vesicles loaded with horseradish peroxidase (HRP) instead of F-dextran. Electron microscopic examinations of HRP-loaded vesicles showed that 89% were spherical in shape with a radius of 7.5 x 10" ⁶ cm and a volume/surface area ratio of 2.5 x 10"* cm.

The effect of changing extravesicular pH from 8.0 to 6.0 on intravesicular fluorescence at 25°C and 5 ^β C was determined. The extravesicular pH was lowered by abrupt addition of a small volume of 2N HC1. The results are plotted in Fig. l. As can be seen from Figure 1, J _H + was markedly slowed by the temperature reduction from 25°C to 5°C.

EXAMPLE 2 The Effect of Buffer Concentration on Proton Permeability

The procedure and materials of Example 1 were employed, except as where indicated differently. Vesicles were loaded, prepared and maintained in 20 mM ("high buffer") or 2 mM ("low buffer") HEPES buffer at pH 8.0. The temperature was maintained at 25°C. The pH outside the vesicles was abruptly lowered to 6.0 and fluorescence was measured over time for the high and low buffer systems. The results are plotted in Figure 2 which illustrates that the rate of decay of the imposed pH gradient varied with changes in the intravesicular buffer concentration. When the lumens of F-dextran- loaded vesicles were buffered with 20 mM HEPES, the time constant of the decline in pH; was given by a T of 7.2 sec. and yielded a P _H + of 2.5 x 10 ³ cm/sec. When vesicles were loaded and maintained in 2 mM HEPES, the T was 0.8 sec. giving a P _H + of 2.2 x 10 ^"3 cm/sec. The close agreement of p _H + values under conditions of low and high intravesicular HEPES concentrations indicates that, in

thiε range of added buffer, the added buffer itself and not other vesicle components dominates vesicle buffer capacity. The change in T and the constancy of P _H + as a function of intravesicular buffer concentration indicates that the fluorescence measurements reflected actual proton flux across the vesicle membrane.

EXAMPLE 3 Effect of pCMBS on Proton Permeability

The effect of pCMBS on the rate of pH change in F- dextran loaded vesicles was determined. The materials and procedures of Example 1 were employed, except as noted. In vesicleε containing 17.6 mM HEPES, pH 8.0, the extravesicular pH was abruptly lowered to 6.0 after one hour preincubation at 4°C in the absence ("control") or presence ("pCMBS") of 1 mM pCMBS. The resultε are plotted in Figure 3. Aε illuεtrated, preincubation of vesicles with pCMBS inhibited 54% of p _H + in F-dextran loaded vesicleε. Control p _H waε 3.9 ± 0.5 x lθ ^"3 cm/εec (n = 8 veεicle preparationε) , while that in vesicles treated with 1 mM pCMBS was 1.8 ± 0.3 x 10 ^~3 cm/sec (n = 5; p < 0.005 compared to control; unpaired t teεt) .

The inhibitory effect of pCMBS waε not εhared by another sulfhydryl reagent, n-ethylmaleimide (NEM, 2 mM; vesicles were pretreated for 1 h with NEM in HEPES buffer on ice). In paired experiments, p _H + was 5.6 ± 0.2 x 10 ^'3 cm/sec in the absence and 6.2 ± 0.9 x 10 ^~ ~ cm/sec (NS compared to control) after pretreatment with NEM. Unlike pCMBS, NEM pretreatment has no inhibitory effect on water flux acrosε the toad bladder or human erythrocyte membrane. See Ojciuε, D.M. and Solomon, A.K. , "Sites of p-chloromercuribenzeneεulfonate inhibition of red cell urea and water tranεport", Biochiτn. Biophvs. Acta 942:

73-82 (1988) . Phloretin inhibits both ADH-simulated urea flux in toad bladder and J _H + across artificial planar lipid bilayer. See Gutknecht, J., "Proton/Hydroxide Conductance and Permeability through Phospho-lipid Bilayer, "Proc. Natl. Acad. Sci. USA. 84.:6443-6556

(1987); Levine, S.D., et al.. "Effect of Phloretin on Water and Solute Movement in the Toad Urinary Bladder," J. Clin. Invest.. 5_2_:1435-1442 (1973). Phloretin was without effect, however, on apical membrane vesicle J _H +. In paired studies, control P _H + was 2.4 ± 4.0 x 10 ^"3 cm/sec; for phloretin-treated veεicleε P _H + was 2.6 ± 0.4 x 10' ³ cm/sec. Vesicleε were exposed to phloretin for 30 min. prior to assay; stopped flow studieε were performed in the continued preεence of phloretin.

EXAMPLE 4

The Effect of Inward and Outward Proton Fluxes

The effect of inward and outward proton fluxes waε determined employing the procedures and materials of Example 1, except as noted. Vesicles were preincubated at pH 8.0 and diluted into pH 6.0 (tracing A), or preincubated at pH 6.0 and diluted to pH 8.0 (tracing B) . The results were plotted in Figure 4. As can be seen, the rate of proton flux waε the same whether the luminal contents of the vesicles or the outside medium was more acidic at the outset of the experiment (P _H + for pH 8.0 to 6.0, 3.9 ± 0.5 x 10 ³ cm/sec; p _H + for pH 6.0 to 8.0; 3.9 ± 0.2 x 10" ³ cm/sec) .

EXAMPLE 5

The Effect of Varying Temperature on Proton Flux The effect of varying temperature on proton flux was studied by constructing an Arrhenius plot for the proton permeation across membranes of toad bladder vesicles containing ADH water channels following the procedures and applying the materials of Example 1, except as noted. Measurements were taken at various temperatures, as indicated, and the results are plotted in Figure 5. Each point represents mean ± SE of 3-4 different experiments; correlation coefficient for the fitted line was 0.979. The calculated activation energy of 3.6 kcal/mole indicates proton flux via channels containing water; valueε greater than 10 are obtained when protonε diffuεe acroεε lipid bilayer. (Gutknecht, J. , "Proton/Hydroxide Conductance and Permeability Through Phospholipid Bilayer," Proc Natl. Acad. Sci. USA. 84:644-3-6556 (1987); Kachadorian, W.A., et al.. "Temperature Dependence of ADH-Induced Water Flow and Intramembranouε Particle Aggregates in Toad Bladder, "Science. 205:910- 913 (1979)).

Example 6

Production of Antibodies Against ADH Water Channels which Inhibit Proton Permeability of Such Water Channels ADH water channel containing vesicleε were purified uεing either a density shifting protocol or by flow sorting for specific experiments. For density shifting, vesicles containing water channels retrieved from the apical membrane were loaded with horseradiεh peroxidaεe (HRP) .

After homogenization of granular cellε, vesicles were partially purified on a linear 29-62% sucrose

gradient. Fractions highly enriched for HRP containing vesicles were then incubated with diaminobenzidine (DAB) and Hj0 ₂ . Under these conditions, HRP catalyzes the polymerization of DAB into a dense polymer that increases the density of HRP-containing vesicles.

These denser vesicles were then separated from their remaining contaminants by repetition of the original sucrose gradient sedimentation step. Flow sorting purification of water channel containing vesicles waε accomplished after loading with fluorescein dextran (F-

Dex) instead of HRP. F-Dex containing vesicles were then separated from contaminants on a Becton Dickinson FACStar Plus flow cytometer as described by Verbavatz, J-M. , et al.. edited S. Jard and R. Jamison Collogue INSERM/Libbey Eurotext pp 105-115, and Valenti, G., et al. "Polyclonal antibodies in the study of ADH-induced water channels in frog urinary bladder", Am. J. Physiol. (Renal) 30:F437- F442 (1991) , the teachingε of which are incorporated by reference. In either case, purified vesicle proteins were then fractionated by SDS-PAGE. (Calamita, G., et al. "Selected polyclonal antibodies and ADH challenge in frog urinary bladder: a label fracture study", Am. J. Phvεiol. (Renal) 3_1:F267-F274 (1992)).

SDS-PAGE gelε containing veεicle proteinε were fixed in a mixture of 40% methanol, 10% acetic acid and 50% water and εtained with 1% Coomassie blue R-250. These fixed proteins were then excised from the gel and used as antigenε for the production of anti-55/53 kD antisera. Rabbit polyclonal anti-55/53 kD antisera was prepared by repeated immunization of New Zealand white rabbits with a mixture of Coo aεεie blue εtained 55 and 53 kD protein gel εliceε and complete Freund'ε adjuvant. After three separate intradermal immunizations containing approximately 20 icrogramε of protein each, rabbit sera

was collected using standard methods and tested for reactivity against purified 55 and 53 kD proteins immobilized on nitrocellulose membranes by the western blotting technique. Animals producing reactive anti- 55/53 antibodies were then titered by dilution of the antisera. These animals then received additional intradermal injections of 20 microgra s of purified 55 and 53 kD proteins emulsified in mineral oil. Large amounts of antisera were then collected by standard techniques.

The antisera was tested for its functional and structural specificity against water channels in TB, mouse kidney (MK) and red blood cells (RBC) . To quantitate inhibition of TB P _f , anti-55/53 kD or preimmune antisera were added to solutionε bathing either the apical surface of paired ADH stimulated toad bladders fixed briefly by exposure to 1% glutaraldehyde (Eggena P. Endocrinol. 91:240, 1972) or to the cytoplasmic surface of water channel containing endosomes prepared from TB homogenateε. Apical addition of anti-55/53 kD antisera maximally inhibited TB P _f by 78 ± 25% (n=9) after 20 minutes but had no effect on water channel endosome P _f (n=3) . Western blots demonεtrated thiε antiserum recognized specifically proteins of 55 and 53 kD in TB, 46, 38 and 30 kD in MK and 55 and 29 kD in RBC. Studies, using immunocytochemistry of mouεe kidney tissue sections, revealed that anti-55/53 kD antisera localized to the apical membranes of renal tubules.

Example 7

Purification and Characterization of ADH Water Channel Proteins

Experimental Procedures

Experimental Methodε Water channel-containing vesicles (WCV) were purified as described previously using a density shifting protocol (Harris, H.W. , et al. J. Membr. Biol. 96:175- 186, (1987)). WCV retrieved from the apical membrane were loaded with horseradish peroxidase. After homogenization of granular cells, WCV were purified partially on a linear 29-62% sucroεe gradient. Fractionε highly enriched for horεeradiεh peroxidase-containing WCV were then incubated with dia inobenzidine (DAB) and H ₂ 0 ₂ . Under these conditions, horseradish peroxidase catalyzes the polymerization of DAB into a dense polymer that increaseε the denεity of horεeradiεh peroxidaεe- containing vesicles. Theεe denser WCV were then separated from the remaining contaminants by repetition of the original sucrose gradient sedimentation step. Purified WCV proteins were then fractionated by SDS-PAGE (Harriε, H.W. , et al., Am. J. Phvεiol. 26l:C143-C153. (1991)). SDS-PAGE gelε containing veεicle proteinε were either fixed in a mixture of 40% methanol, 10% acetic acid, and 50% water and strained with 1% Coomassie Blue R-250, or gel proteins were transferred to ^' nitrocellulose or polyvinylidene difluoride (PVDF) membraneε. Stained SDS-PAGE gelε were dried on a heated gel dryer for autoradiography or selected portions of protein bands excised from the gel with a razor blade for peptide

mapping analysis. Alternatively, nitrocellulose and PVDF membranes containing vesicle proteins were used for immunoblotting and carbohydrate analyses or compositional and amino-terminal analyses as described below. Granular cell proteins were ³² P-labeled as described previously (Konieczkowski, M, and Rudolf, S.A. , J. Pharmacol. Exp. Ther. 234:515-521, (1985)) by in vitro incubation of isolated intact toad bladder sacs with [ ³² P] orthophosphate. The carbohydrate content of protein bands were assessed by the method of O'Shannesεy et al . (O'Shannessy, D.J. et al. _r Anal. Biochem. 163:204-209, (1987)) using reagents provided by the Glycan detection Kit (Boehringer Mannheim) . Nitrocellulose-bound proteinε were oxidized with sodium metaperiodate, washed and reacted with digoxigenin-succinyl-amidocaproic acid hydrazide. Digoxigenin-labeled glycoconjugates were subsequently detected by enzyme-linked immunoasεay uεing alkaline phosphataεe-conjugated sheep anti-digoxigenin Fab fragments. Reactivity of 55-, 53-, and 17 kDa protein bands were compared with that exhibited by the glycoprotein transferrin assayed under identical conditions.

Peptide mapping analyεeε waε carried out aε detailed by Elder et al. (Elder, J. H. , et al. , J. Biol. Chem. 252:6510-6515, (1977)) using plastic-backed celluloεe thin layer plateε on a flat bed electrophoreεiε unit. When the peptide compositions of two plates were being compared, they were electrophoresed and chromatographed together. All autoradiography waε carried out at -60°C using enhancing screens. Peptide mobilities were determined by multiple autoradiograhpy exposures and measurementε of the mobility of each peptide with reεpect to the origin.

Excised protein bands fixed on PVDF membranes were rinsed briefly in 20% methanol, 80% water, then hydrolyzed at 107°C for 24 h in vacuum-sealed glass vials containing 6 N HCl. For each sample, an identical size PVDF membrane from the same blot containing no stainable protein was included as a control. Preliminary studies using purified proteins (0-galactosidase and albumin) demonstrated that transfer and hydrolysis on PVDF membranes did not produce any significant changes in their amino acid composition. These data are consistent with more extensive studies performed on a wide variety of PVDF bound proteins by Tarr et al. (Techniques in Protein Chemistry II, Chap. 13, Academic Press, NY (1991)). Protein bands excised from PVDF blots were also subjected to amino-terminal sequence analysis using an Applied Biosystems model 4770 protein sequenator.

The effect of preincubation of mercurial reagents on WCV P _f was quantitated by monitoring the fluorescence quenching of entrapped carboxyfluorescein (CF) aε described previously (Verkman, et al.. Nature 333:268- 269, (1988), (Harris, et al.. Am. J. Phvsiol. 253:F237- F243, (1990)). Briefly, WCV were loaded with IOIΓIM CF in a fashion identical to that described for horseradish peroxidase. A partially purified membrane fraction (intermediate pellet) containing CF-loaded WCV waε uεed for all studies. All P _f measurements were performed using an Applied Photophysicε SF017 εtopped flow fluorimeter with a measured dead time of 0.7 ms. Mean values of P _f are expressed aε cm/ε +/- standard error. The statistical significance of differenceε between groupε of P _f meaεurements were tested using the Bonferroni t test (Glantz, Primer of Bioεtatiεticε. McGraw Hill Book Co. New York, (1988)).

Inhibition and labeling studies were performed on

veεicles resupended in 500 μl of homogenization buffer (2.5 ΠLM HEPES, 7.5 M KC1, and 0.1 mM EDTA, pH 8.0) and preincubated for 30 min with either 2 mM p-CMBS, 0.25 Π_M FMA, or 0.5 mM NEM prepared as 4, 0.5, and 5 M stocks, respectively, homogenization buffer immediately prior to use. In selected experiments, 2-Mercaptoethanol waε added, to a final concentration of 2 mM, 5 min prior to FMA addition. After these incubations were completed, the vesicles were collected by centrifugation at 10,000 X g for 10 min, washed twice in homogenization buffer and 100 μl of anti-fluorescein antisera added to quench all extravesicular fluorescence. WCV P _f waε then meaεured using stopped flow fluorimetry.

Horseradish peroxidase-loaded WCV were incubated with FMA as described above,- then purified by density shifting. WCV proteins were fractionated by SDS-PAGE, substituting NEM in place of mercaptoethanol in SDS-PAGE denaturation buffer to prevent detachment of mercurials from proteins. Proteins were transferred to nitro- cellulose membranes and either stained with 0.1% Amino Black and destained with 50% methanol, 45% water, 5% acetic acid mixture or uεed for immunoblotting aε described below. Since the presence of DAB interferes with standard protein asεayε, the protein load on each lane waε εtandardized by firεt staining a separate lane loaded with an identical quantity of density-εhifted protein. To quantitate the distribution of blot proteinε, filterε were rendered clear by incubation in 100% Triton X-100 followed by quantitated using a Zeineh soft laser densitometer (Biomed Instrument: Co.,

Fullerton, CA) as described previously (Harris, et al., Am. J. Phvsiol. 261-.C143-C153. (1991)).

For identification of FMA proteinε, filters containing identical quantities of vesicle protein were

εubjected to immunoblot analysis using affinity-purified rabbit anti-fluorescein antisera. This was prepared by repeated intradermal immunization of fluorescein- conjugated keyhole limpet hemocyanin in rabbitε. The resulting antisera was affinity-purified using a fluorescein-albumin column.

Membranes were incubated at 37° C in a 3% gelatin solution for 1 h. After 3 rinses in phosphate-buffered saline with Tween (PBS-T) (100 mM sodium chloride, 50 mM sodium phosphate, pH 7.5, 0.1% Tween 20), the blots were incubated for 1 h with a 1:500 dilution of rabbit anti- fluorescein antisera. After 3 rinses with PBS-T, bound antibody was detected using an alkaline phosphatase- conjugated affinity-purified goat anti-rabbit secondary antibody and exposure to color development substrateε.

Results Previous data produced from LPO-mediated apical membrane and intraendosomal ¹²⁵ I-labeling of ADH- stimulated granular cells aε well aε quantitation of proteins from purified WCV revealed that candidate ADH water channel proteins of 55 and 53 kDa appear aε distinct sharp bands, while the 17-kDa band is quite broad, stains poorly with Coomasεie Blue, and extends from 20 to 15.5 kDa. To obtain εufficient quantitieε of purified material to examine the structure and composition of theεe proteins, WCV were isolated, pooled, and vesicle fractions were subjected to preparative SDS- PAGE to purify these protein bandε. Optimal εeparation of the 55- and 53-kDa proteinε waε obtained using 9% acrylamide gels under conditions where the bromophenol blue dye front waε permitted to run off and electrophoreεiε continued until proteinε poεεessing molecular maεεeε of 28 kDa were preεent in the dye front

gel. The 17-kDa protein waε purified uεing 12% SDS-PAGE gels. As reported previously, the 55- and 53-kDa bands were present in a 2:1 ratio. In contrast to the 55- and 53-kDa protein bands, the 17-kDa protein band stained faintly with Coomassie Blue.

Three hundred micrograms of 55-and 53-kDa protein bands were prepared representing approximately 5 nmol of each protein. Eighty micrograms of 17-kDa protein band (approximately 4.7 nmol) was prepared in an identical fashion. To prevent possible cross-contamination of the 55- or 53-kDa protein by its neighboring band, only the upper (55 kDa) or lower (53 kDa) half of the stained protein band was used for peptide mapping, compositional, or amino-terminal analyses deεcribed below. It iε possible that the 55/53-kDa doublet or the broad 17-kDa band may be due in part to post- tranεlational modificationε such as phosphorylation or covalent carbohydrate addition. In previous studies, it has been demonεtrated that the 17-kDa protein iε not ³² P- labeled after incubation of intact toad bladderε with

[ ³ P] orthophosphate. To establish whether the 55- or 53- kDa proteins contain significant quantitieε of exchangeable phoεphate, intact toad bladderε were incubated with [ ³² P] orthophoεphate under identical conditionε and ³² P content of the 55-and 53-kDa bandε aεεesεed by autoradiography of gels identical to those described above. These gelε showed that neither the 55- nor the 53-kDa band iε ³² P-labeled.

To assess whether the 55-, 53-, or 17-kDa protein species contained significant quantities of covalent carbohydrate, bandε on nitrocelluloεe were subjected to mild periodate treatment with subsequent attachment of digoxigenin via a hydrazide group. Digoxigenin-labeled glycoconjugateε were then detected uεing alkaline

phosphatase-conjugated sheep antidigoxigenin Fab fragments as described above. None of the three protein bands demonstrated significant reactivity under conditions where covalent carbohydrate was readily detected in 1 μg of transferrin. On the basis of these data, it appears that the 55-, 53-, 17-kDa protein bands do not contain significant quantities of exchangeable phosphate or carbohydrate.

Two-dimensional mapping of chymotryptic peptides derived from fixed, Coomassie Blue-stained protein bandε was used to determine the structural relationships between the 55-, 53-, and 17-kDa protein bandε. Thiε method has been widely used to determine relationships between purified SDS-PAGE protein bands. The peptide maps of the 55- and 53-kDa bandε were similar but not identical. Detailed comparisonε of theεe mapε reveal the preεence of many common peptideε in addition to the preεence of a limited number of peptides that are unique to either the 55- or 53-kDa band. All of the major chymotryptic peptideε of the 17-kDa protein have counterpartε in the mapε of the 55-and 53-kDa proteinε. Together, these data suggest that the 55- and 53-kDa proteins poεsess similar structual characteriεtics and the 17-kDa protein band is possibly a fragment or subunit derived from the 55- and 53-kDa proteins.

Table I liεtε the average of three co poεitional analyεeε for each of the three proteins expressed as mol %. Comparison of the 55- and 53-kDa proteins reveal they have very similar compositionε. There are, however, notable differences between the 55- and 53-kDa species. These include the 53-kDa protein'ε higher content of glutamine/glutamic acid (Gl(x); 12.1 mol %) aε compared with that of the 55-kDa protein (2.1 mol %) . The 55-kDa protein contains nearly twice as much tyrosine aε does the 53-kDa band. Interestingly, both proteinε are highly

enriched in cysteine and contain little proline.

The composition of the 17-kDa protein band is notable for its lack of histidine, threonine, tyrosine, lysine, or proline. It contains 3.5 mol % cysteine and is enriched for leucine, glycine, alanine, and Gl(x). These compositional data provide additional support for the hypothesis that the 17-kDa band may derive from either the 55- or 53- kDa proteins. The 17-kDa protein contains 11.6 mol % approximately 17 residues of Gl(x). Since the Gl(x) content of the 17-kDa protein exceeds the Gl(x) content of the 55-kDa band (2.1 mol % or 10 Gl(x) residues) , the 17- kDa protein may actually derive from the 53-rather than the 55-kDa protein. The larger Gl(x) content (12.1 mol %) of the 53-kDa protein could allow the 17-kDa protein to derive from it by proteolysis. If the 17-kDa protein iε indeed a fragment of the 53-kDa protein, theεe data may suggest that the bulk of the Gl(x) residues are clustered within one region of the 53- kDa polypeptide.

TABLE I

Mol %

17 kDa 5.8

11.6 3.3 0.0

11.2 0.0

10.0 7.4 0.0 3.5 9.5

Met He Phe Leu Lys Pro

The data contained in Table II show that the amino termini of the 55-and 53-kDa proteins appear to be blocked. In contrast, identical analysis of the 17-kDa protein yields multiple amino-terminal amino acids including glycine, asparagine, and valine. Previously, it was observed that exclusion of the protease inhibitors phenylmethylsulfonylfluoride, leupeptine, and EDTA results in lower yields of the 55- and 53-kDa protein bands in conjunction with increased staining of the 17- kDa band. Thus, these data are consistent with the possibility that the multiple amino termini displayed by the 17-kDa protein band may reεult from proteolyεiε.

TABLE II

55 kDA None recovered

53 kDA None recovered

17 kDa'

Cycle 2 40 p ol Gly, 20 pmol Aεn, 14 p ol Val

Cycle 3 31 pmol Glu, 11 pmol Leu, 8 pmol Ser

* Yields from representative sequence analysis cycles 2 and 3 are displayed. Cycle l was not informative.

Although the 55- and 53-kDa proteins are abundant in WCV and membrane labeling studies suggest strongly that

these integral membrane proteins may be water channel components, there has been no evidence linking directly the modification of these proteins to functional changeε in ADH water flow. Since affinity-purified rabbit anti- fluorescein anti-sera binds to the fluorescein moiety of fluorescein mercuric acetate (FMA) , an inhibitor of ADH water flow in anuran bladders, an attempted to link FMA induced inhibition of water channels in WCV with the specific covalent labeling of the 55-, 53-, or 17-kDa proteins was made. The effects of preincubation with either p-CMBS or FMA on water channels present in WCV were assessed. WCV were exposed to either p-CMBS or FMA in a series of paired experiments. Control WCV were highly water-permeable (P _f =0.115 + 0.017 cm/s; n = 23) aε deεcribed previously. Preincubation with l m p-CMBS resulted in a 93% reduction of WCV P _f to a value of 0.008 + 0.0025 cm/s (n = 9; p- θ.001). In a similar fashion, preincubation with 0.25 mM FMA reduced WCV P _f by 82% to a value of 0.021 + 0.005 cm/s (n = 9; p<0.00l). Inhibition of WCV P _f by p-CMBS and FMA appeared to be specific for mercurials and mediated through binding to εulfhydryl groupε. Expoεure to 0.5 mM n-ethylmaleimide (NEM) did not effect WCV P _r (0.12 + 0.03; n = 6; not εignificant (NS)) . Identical results have been obtained in intact ADH-stimulated toad bladderε. FMA'ε inhibition of WCV P _f waε prevented by prior addition of 5 mM 2-mercaptoethanol (0.11 + 0.05; n = 3;NS) .

SDS-PAGE-fractionated proteinε from control (lane A) and FMA-treated WCV were analyzed. To aεs,esε whether FMA produced alterations in specific WCV proteins, 55- and 53- kDa WCV proteins were quantitated by laser densitometry in a paired fashion. In previous εtudieε, it has been εhown that these 55- and 53-kDa proteins account for 49.5 +/- 4% of WCV protein. The amount of

55- and 53-kDa protein waε not changed by FMA exposure, since gels from control (50+5%) and FMA-treated (48.5 +6%) WCV were indistinguishable.

WCV proteins labeled by FMA were detected by binding of affinity-purified anti-fluorescein antisera in a Western blot. Four protein bands of 92, 55, 53 and 29 kDa are recognized specifically by this antisera under conditions that cause a 82% inhibition of WCV Pf. Affinity-purified antisera do not bind to any proteins from control WCV. Although the number of FMA molecules binding to a particular protein band have not been quantitated, it haε been conεistently observed that the 55-kDa protein was the moεt intenεely labeled whereaε the other bandε of 92, 53, and 29 kDa were labeled with leεε and approximately equal intensity. Prior addition of 2- mercaptoethanol that reεultε in partial protection of FMA'ε inhibitory effect caused a marked diminution in the FMA labeling of all four protein bands.

Example 8

Measurements of osmotic water permeability have suggested that water channels are present in plasma membranes of human red cell (RBC) and certain epithelial cells in toad bladder (TB) and mouse kidney (MK) . Further, it haε been hypotheεized previouεly that proteinε of 55 and 53 kD are componentε of antidiuretic hormone (ADH) water channelε in TB. Polyclonal rabbit antiserum was raiεed againεt theεe proteinε and its

functional and structural specificity against water channels in TB, MK, and RBC was tested. To quantitate inhibition of TB P _f , anti-55/53 KDa as preimmune antiserum were added to solutions bathing either the apical surface of paired ADH stimulated toad bladders fixed by exposure to 1% glutaraldehyde or to the cytoplasmic surface of water channel containing endosomes prepared form TB homogenates. Apical addition of anti- 55/53 kDa antisera maximally inhibited TB Pr by 78 + 25% (n-9) after 20 minutes but had no effect on water channel endosome P _f (n-3) . Western blots demonstrated this anti¬ serum recognized specifically proteins of 55 and 53 kDa in TB, 46, 38, and 30 kDa in MK and 55 and 29 kDa in RBC. It waε concluded that anti-55/53 kDa anti-εerum 1) provides strong evidence that the 55 and 53 kDa proteins are ADH water channel components, 2) inhibitε ADH εtimulated P _f by TB by binding to the apical membrane domainε of these proteins and 3) recognize MK and RBC proteins that may also be water channel componentε.

Example 9

Purification and Characterization of Water Channel Containing Vesicles from Rat Inner Medulla

A major obstacle to the isolation and characterization of mammalian ADH water channelε haε been the lack of methods to purity WCV from mammalian IM. Rat renal IM endosomes were loaded with fluorescein dextran and fractionated by denεity gradient centrifugation. Two populationε of endoεomeε of different denεity were identified: light endoεomeε and heavy endoεomeε. Flow cytometry analyεiε showed that <98% of each population was positive for endosomes fluorescein marker. Stopped-

flow measurements of osmotic water permeability (P _f ) by self-quenching of entrapped fluorescein revealed that the heavy endosome population consisted nearly entirely of vesicles of high P _f of 0.28 ± 0.05 (SE) cm/sec (n=10) . In these vesicles, the activation energy for water flow was 3.1 + 0.7 kCal/mole and P _f was reduced by the organic mercurial reagent, parachloromercuribenzene sulfonate (pCMBS) to 0.019 ± 0.001 cm/sec (n=3). By contrast, light endosomes exhibited low P _f (0.0016 ± 0.0001 cm/sec) . SDS-PAGE of heavy endosomes revealed a pattern of protein bands distinct from that of the initial homogenate, with a prominent band at 40 kDa. Immunoblotε were performed uεing antiεerum that both recognizeε putative 55/53 kDa ADH water channel proteinε in toad bladder granular cells and selectively labels the apical membraneε of mammalian collecting ductε. Immunoblotε of heavy endosomeε revealed labeling of the 48 kDa band with thiε antiserum. It was concluded that the heavy endosome fraction iε compoεed predominantly of veεicleε containing functional water channelε. The identification of a prominent protein band of 48 kDa which εhareε antigenicity with candidate toad bladder ADH water channel proteins will help in the purification and characterization of mammalian ADH water channels.

Eσuivalents

Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments described herein. Such equivalents are intended to be encompassed by the following claims.