Activation of the Avian Erythrocyte Na-K-Cl Cotransport Protein by Cell Shrinkage, cAMP, Fluoride, and Calyculin-A Involves Phosphorylation at Common Sites*

Na-K-Cl cotransport activity in duck erythrocytes increases ∼10-fold in response to osmotic cell shrinkage, norepinephrine, fluoride, or calyculin-A (an inhibitor of type-1 and -2a phosphatases). To assess whether all four stimuli promote phosphorylation of the cotransport protein and whether this phosphorylation is catalyzed by the same kinase, the cotransporter was isolated from erythrocytes by immunoprecipitation and its pattern of phosphorylation was evaluated. Each stimulus evoked proportionate increases in cotransporter activity and phosphorylation. No two stimuli in combination evoked greater activation and phosphorylation than did the more potent of the two stimuli acting alone. Phosphoamino acid analysis of the cotransport protein indicated that phosphorylation occurs at serine and threonine residues. Phosphopeptide mapping revealed a distinctive pattern of 8 major tryptic phosphopeptides, none of which were significantly phosphorylated in the unstimulated state. Maps of cotransporters activated by the four different stimuli were indistinguishable. Measurements of phosphorylation stoichiometry indicated that each cotransporter acquires ∼5 phosphates on going from an inactive state in swollen cells to an active state in shrunken cells. Staurosporine, a kinase inhibitor with broad selectivity, inhibited each stimulus equipotently (IC50 ∼ 0.7 μm). Staurosporine promptly reversed cotransporter activity and phosphorylation when added to shrinkage-stimulated but not to calyculin-stimulated cells, indicating that it enters the cell rapidly and blocks phosphorylation. These results suggest that cell shrinkage, cAMP, fluoride, and calyculin-A promote the phosphorylation of the Na-K-Cl cotransport protein at a similar constellation of serine and threonine residues. It is proposed that all modes of stimulation ultimately involve the same protein kinase.

Na-K-Cl cotransport is regulated by numerous first and second messengers through a complex and cell-specific interplay of stimulatory and inhibitory signals (1). The molecular mechanisms by which cell surface receptors, cell volume, cytosolic chloride, cytoskeletal architecture, and proliferative status modulate cotransport activity remain unknown. Early recognition that ion movement by the Na-K-Cl cotransporter, although energetically passive (2,3), requires cytosolic ATP and Mg 2ϩ (2, 4 -7) prompted speculation that acute regulation might involve reversible phosphorylation of the cotransport protein, regulatory subunits, or upstream signal transducers (8,9). Circumstantial support came from demonstrations that cotransport activity is increased by agents that inhibit protein phosphatases (10,11) and decreased by agents that inhibit protein kinases (11,12). Recent studies have established that the Na-K-Cl cotransporter itself is a phosphoprotein (11,13,14) whose phosphorylation state parallels its activation state (13)(14)(15)(16)(17)(18). While it is generally assumed that cotransporter phosphorylation is both necessary and sufficient for transport activity, recent research suggests that additional factors, including affiliated proteins (19), cytoskeletal interactions (16, 20 -22), and mechanical changes in the cell membrane (23) might influence cotransport activity.
Duck erythrocytes have long served as a premier model of electroneutral ion transport by virtue of their simplicity, uniformity, and ease of experimental manipulation. These cells manifest robust Na-K-Cl cotransport in response to four types of stimuli: osmotic cell shrinkage, elevated cytosolic cAMP (norepinephrine), Ser-Thr phosphatase inhibitors (calyculin-A, okadaic acid, endothall thioanhydride, fluoride), and deoxygenation (24,25). Unlike other modes of stimulation, the norepinephrine response is associated with increases in cytoplasmic cAMP (24) and cAMP-dependent protein kinase activity (8,11), and can be blocked by kinase inhibitors like K-252a and H-9 at doses that disable cAMP-dependent protein kinase in intact avian erythrocytes (11). The same kinase inhibitors also block activation of cotransport by cell shrinkage, fluoride, and okadaic acid, but only at concentrations an order of magnitude higher (11). These observations suggest that the avian erythrocyte Na-K-Cl cotransporter is regulated by at least two kinases, one of which is cAMP-dependent protein kinase. The fact that K252a blocks activation by cell shrinkage, fluoride, and okadaic acid at a similar high dose raises the possibility that non-cAMP-dependent stimuli are transduced by the same kinase (25). While cotransport activity appears to determined by a dynamic competition between ongoing protein kinase and phosphatase activities, a key question is whether all modes of stimulation involve phosphorylation of the cotransport protein itself and whether different stimuli involve different kinases.
The purpose of the present study was to test the hypothesis that four different modes of stimulation (cell shrinkage, cAMP, fluoride, and calyculin-A) involve phosphorylation of the cotransport protein at common sites. The recent advent of monoclonal antibodies capable of immunoprecipitating the Na-K-Cl cotransport protein from detergent extracts of 32 P-labeled duck erythrocytes with high efficiency (26) now makes it possible to quantitatively compare the phosphorylation induced by different stimuli and to assess the physical disposition of phosphorylation sites. The results of this analysis suggest that all forms of activation promote phosphorylation of the cotransport protein at a common set of Ser/Thr sites.
A monoclonal antibody (T14) directed against the carboxyl-terminal 310 amino acids of the human colonic Na-K-Cl cotransporter (hNKCC1) was developed as described previously (26). The T14 antibody detects 0.5 ng of duck erythrocyte cotransporter on Western blots using enhanced chemiluminescence (26) and immunoprecipitates the SDS-solubilized protein with Ͼ80% efficiency. Recognition of this protein by monoclonal antibodies T4 and T14 but not by T9 suggests that avian erythrocytes possess a homologue of the ubiquitous NKCC1 isoform. Hybridoma cells were grown in the ascites of pristane-primed severe combined immunodeficient mice (Taconic CB-17 Fox-Chase SCID). After clarification by centrifugation, the ascitic fluid was preserved by the addition of sodium azide (0.025%) and protease inhibitors (1.5 M pepstatin A, 3.5 M chymostatin, 10 units/ml aprotinin, 2.5 M leupeptin, and 200 M AEBSF) and stored at 4°C.
Preparation of Erythrocytes-Blood was drawn from the brachial vein of female Pekin ducks (Anas platyrynchos) into heparinized syringes. After removal of plasma and buffy coat, the erythrocytes were stored in ice-cold DFS (146 mM NaCl, 6 mM KCl, 0.1 mM Na 2 PO 4 , 10 mM glucose, 20 mg/liter penicillin, 45 mg/liter streptomycin, and 20 mM Na-TES, pH ϭ 7.40 at 41°C, 320 mosM) for up to 4 days with no apparent detrimental effect. Before use, the erythrocytes were incubated (2.5% hematocrit) for 45 min in fresh DFS at 41°C to achieve a steady state with respect to ion and water contents.
Cell Volume-Intracellular water content, an index of cell volume, was measured by a gravimetric method. Duplicate aliquots of cell suspension (800 l of 5% hematocrit) were added to baked microcentrifuge tubes of predetermined weight and centrifuged for 2 min at 5000 ϫ g. After removal of the supernatant, the pellets were centrifuged again for 5 min at 12,000 ϫ g at 4°C. Fluid was removed from the pellet surface by capillary action using a small pointed swab. Each tube was then reweighed, dried in an oven at 85°C for Ͼ18 h, and weighed again. Wet cell weight was corrected for an extracellular fluid weight of 2.5%.
In Vivo Phosphorylation-Duck erythrocytes (22 l/sample) were incubated with gentle agitation in a siliconized flask (7% hematocrit, 41°C) for 3 h in DFS containing 150 Ci/ml [ 32 P]orthophosphate. Labeled cells were washed twice in ice-cold DFS and incubated (2.5% hematocrit, 41°C) for 12 min in fresh DFS containing 50 M ouabain and an activator of Na-K-Cl cotransport. After stimulation, the cells were pelleted by centrifugation (10 s, 6000 ϫ g) and frozen by immersion in liquid nitrogen.
Measurement of Transport-Cotransporter activity was assayed as the unidirectional influx of 86 Rb, an ion that quantitatively substitutes for K ϩ in the cotransport process (27). Cells were treated exactly as those employed for 32 P labeling to allow comparison between cotransporter activation and phosphorylation. Erythrocytes were then incubated in DFS (5% hematocrit, 41°C) containing 50 M ouabain and an activator of cotransport (10 M norepinephrine, 100 mM sucrose, 10 mM sodium fluoride, or 200 nM calyculin-A) for 10 min, a period during which the effect of each activator becomes maximal and invariant. 86 Rb entry was initiated by the addition of isotope and terminated 1-4 min later by dilution with ice-cold "stop solution" (DFS containing 250 M bumetanide). Extracellular 86 Rb was removed by washing the cells three times in stop solution. Intracellular and extracellular 86 Rb was quantified by ␥ spectroscopy (Beckman). Influx assays were confined to an early period during which 86 Rb accumulation was proportional to time. The portion of 86 Rb influx attributable to cotransport was determined by subtraction of the component resistant to 100 M bumetanide; this residual component averaged Ͻ5% of the stimulated rate and was unaffected by activators of cotransport.
Immunoprecipitation-Frozen 32 P-labeled cells were thawed in icecold AP buffer (150 mM NaCl, 30 mM NaF, 5 mM Na 4 EDTA, 15 mM Na 2 HPO 4 , 15 mM Na 4 -pyrophosphate, 20 mM Hepes, pH ϭ 7.2). The lysed cells were washed in ice-cold AP buffer until pink in color and then solubilized in 200 l of warm AP buffer containing 1% SDS by probe sonication (Fisher Sonic Dismembranator 50, setting 2, 20 s). After heating to 55°C for 20 min, the SDS extract was diluted with 600 l of ice-cold AP buffer containing 2.5% CHAPS, protease inhibitors, and 1 mM orthovanadate. After 90 min on ice, the extract was clarified by centrifugation (12,000 ϫ g for 5 min at 4°C) and incubated overnight on ice with 2 g of monoclonal antibody T14. Immune complexes were collected on Protein G-Sepharose beads and rinsed four times with ice-cold AP buffer containing 1% Triton X-100, followed by one rinse with PBS. The immunoprecipitate was extracted into 70 l of SDSsample buffer (4% SDS, 50 mM dithiothreitol, 50 mM Tris-HCl, pH 6.8, 12% glycerol, and 0.01% Serva Blue G) and electrophoretically separated on 7.5% Tricine-SDS-polyacrylamide gels (28). Autoradiography was performed using x-ray film or a storage phosphor screen (Phospho-rImager, Molecular Dynamics). Western blot analysis indicated that the efficiency of immunoprecipitation from extracts of both resting and stimulated cells exceeded 80%.
Specific Activity of [␥-32 P]ATP-The 32 P content of ATP in KCO 3neutralized perchloric acid extracts of 32 P-labeled cells was determined by ascending thin layer chromatography on polyethylamine-cellulose plates in 0.85 M KH 2 PO 4 at pH 3.4 (29). To quantify [ 32 P]ATP, discrete spots of radioactivity comigrating with [ 32 P]ATP standards were scraped from the plate and analyzed by liquid scintillation spectroscopy, correcting for a counting efficiency of 70% for [ 32 P]ATP bound to cellulose (8).
The fraction of [␥-32 P]ATP in extract [ 32 P]ATP was measured as described by Mayer and Krebs (30). A neutralized perchloric acid extract of labeled cells (20 mg) was applied to a column (5 ϫ 40 mm) of Dowex 1-formate. The column was washed with 5 ml of 4 N formic acid, and ATP was eluted with 0.4 M ammonium formate in 4 N formic acid. The solution was lyophilized, dissolved in 300 l of water, and lyophilized again. The dried material was dissolved in 450 l of 100 mM imidazole (pH ϭ 7.4) containing 1 g of phenol red to confirm neutral pH. ATP was converted to glucose 6-phosphate by adding 10 mM glucose, 0.05% bovine serum albumin, 50 M cold ATP, and 5 units of yeast hexokinase, and incubated at 30°C for 1 h. The solution was then applied to a Dowex 1-formate column, and the column was washed with 5 ml of water. Glucose 6-phosphate was eluted with 2 N formic acid, after which residual [ 32 P]ATP was eluted with 0.4 M ammonium formate in 4 N formic acid. Both eluates were analyzed for 32 P by liquid scintillation spectroscopy. Parallel assays using known quantities of authentic [␥-32 P]ATP provided correction factors for the yield of cellular ATP (ϳ85%) and for the efficiency of enzymatic formation and recovery of [ 32 P]glucose 6-phosphate (ϳ77%). The ratio of [␥-32 P]ATP to total [ 32 P]ATP, i.e. the fraction of cellular [ 32 P]ATP converted into [ 32 P]glucose 6-phosphate by hexokinase, was determined to be 0.82 Ϯ 0.02 (mean Ϯ S.D., n ϭ 3).
The specific activity of cellular [␥-32 P]ATP was calculated from the radioactivity of ATP in the extract, the [␥- 32  Phosphoamino Acid Analysis-Cotransporter protein was isolated from 32 P-labeled cells by immunoprecipitation, separated by SDS-PAGE, and electrophoretically transferred to PVDF membrane. Regions of PVDF containing [ 32 P]cotransporter were located by autoradiography, excised, and hydrolyzed in 6 N hydrochloric acid at 100°C for 3 h. The hydrolysate was lyophilized, reconstituted in 20 l of water containing 10 g of unlabeled phosphoamino acids, and separated by one-dimensional thin layer electrophoresis (600 V for 5 h at 4°C) on a silica gel plate (Whatman 4410 -221) in a buffer consisting of formic acid, acetic acid, and water (25:78:897). Phosphoamino acids were visualized by ninhydrin staining and autoradiography using a storage phosphor screen.
One-dimensional Phosphopeptide Analysis-Gel slices containing [ 32 P]cotransporter were rinsed thoroughly with water, then treated with 15 mM N-chlorosuccinimide for 20 min at 23°C to selectively cleave tryptophanyl peptide bonds (32). Proteolytic fragments were separated on a 7.5% Tricine SDS-polyacrylamide gel, and those containing 32 P were detected by autoradiography using a storage phosphor screen.
Two-dimensional Phosphopeptide Analysis-Gel slices containing [ 32 P]cotransporter were rinsed thoroughly with water and equilibrated with 200 mM NH 4 HCO 3 . The gel pieces were then rotated with 500 l of 200 mM NH 4 HCO 3 containing 100 g of L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin for 24 h at 37°C, with the addition of 50 g of freshly prepared trypsin after 17 h. The digest was then subjected to three cycles of lyophilization and reconstitution in water (500 l). The final residue was dissolved in 20 l of electrophoresis buffer (10% acetic acid, 1% pyridine, pH 3.5) and spotted onto thin layer cellulose sheets along with marker dyes (xylene cyanol FF and phenol red, 2 g each). Phosphopeptide maps were generated by electrophoresis at 500 V for ϳ3 h followed by crossed ascending chromatography in 1-butanol:pyridine:acetic acid:water (60:40:12:48). To ensure uniformity between different samples, electrophoresis and chromatography were allowed to continue until each marker dye had migrated a fixed distance. Phosphopeptides were detected by autoradiography using a storage phosphor screen.

Time Course of 32 P Incorporation into Cellular ATP-Incu-
bation of duck erythrocytes with [ 32 P]orthophosphate resulted in a slow equilibration of 32 P into cellular ATP (Fig. 1). Halfmaximal incorporation required ϳ3 h, consistent with previous data on turkey erythrocytes (8). Analysis of extracts of cells labeled for 3 h revealed that 82 Ϯ 2% (mean Ϯ S.D., n ϭ 3) of the radioactivity in cellular [ 32 P]ATP could be transferred in vitro to glucose by hexokinase, indicating that the predominant form of radioactive ATP at this point in the labeling process is [␥-32 P]ATP. All experiments described hereafter were performed on erythrocytes that had been prelabeled with 32 P for 3 h and then exposed for 9 -12 min to an activator of cotransport and ouabain (to block the Na/K pump). During the activation period, changes in cellular [ 32 P]ATP (measured chromatographically), water content (measured gravimetrically), and pH were negligible, in agreement with previous studies (7,8,31).
Activators of Cotransport Promote Cotransport Protein Phosphorylation-Four factors known to stimulate cotransport activity (hypertonicity, norepinephrine, fluoride, and calyculin-A) promoted incorporation of 32 P into the cotransport protein (Fig.  2). Optimal phosphorylation was obtained with doses that evoke maximal cotransport activity: 10 M norepinephrine, 10 mM fluoride, 100 nM calyculin-A, and 100 mM sucrose (25). When added with the stimulus, 100 M bumetanide abolished cotransporter activity but did not alter the rate or the extent of phosphorylation (data not shown); this observation supports the idea that cotransporter phosphorylation is the cause rather than the consequence of cotransport activity and excludes the possibility that bumetanide inhibits by blocking phosphorylation. The stimulatory effect of fluoride was not enhanced by the addition of 10 M Al 3ϩ nor diminished by the chelation of Al 3ϩ with 1 mM deferoxamine mesylate (data not shown), discounting the possibility that it is due to aluminofluoride (AlF 4 Ϫ ) as proposed for rat parotid acinar cells (33).
Cotransporter Activity and Phosphorylation Co-vary with Changes in Cell Volume-The effect of osmotically induced changes in cell volume on cotransporter activity and phosphorylation was measured on paired suspensions of duck erythrocytes. The conditions used for 32 P labeling (preincubation for 3 h at 10% hematocrit) had no significant effect on cell volume or on the cotransporter's subsequent responsiveness to various stimuli. In unstimulated cells, cotransporter activity remained low, averaging only 3-7% of maximal levels. Slight osmotic cell swelling, from 1.5 to 1.63 liters/kg of cell solid, caused a virtual cessation of cotransport activity and a slight reduction in cotransport protein phosphorylation (Figs. [3][4][5]. Greater degrees of swelling, even to extreme prelytic dimensions, caused no further decrease in cotransporter phosphorylation. This residual phosphorylation persisted when the swollen erythrocytes were exposed for 12 min to 30 M staurosporine, a broad spectrum protein kinase inhibitor that rapidly abolishes cotransport activity (e.g. Figs. 6 -8). Together these results suggest that the cotransport protein possesses a minor subset of nonregulatory or inhibitory phosphorylation sites whose turnover is volume-independent and either staurosporine-insensitive or relatively slow.
Cell shrinkage evoked parallel increases in cotransporter activity and phosphorylation (Fig. 3). A 33% reduction in cell water, from 1.5 to 1.0 liter/kg of cell solid, evoked near-maximal transport and 32 P incorporation. The effect of cell shrinkage on phosphorylation was somewhat selective for the cotransporter as none of the major membrane phosphoproteins, of which ϳ28 could be resolved on a 7.5% Tricine-SDS gel, were significantly affected by osmotic perturbation (data not shown).
Stoichiometry of Phosphorylation-If the phosphorylation observed here is the direct cause of transport function, each cotransport unit recruited into activity should acquire at least one phosphate group. To estimate the stoichiometry of phos-phorylation, aliquots of 32 P-labeled erythrocytes were analyzed for cell number, [␥-32 P]ATP specific activity, and cotransport protein phosphorylation (i.e. the 32 P content of the cotransport protein after isolation by immunoprecipitation and gel electrophoresis). The number of cotransporters in the immunoprecipitate was estimated from the average number of specific [ 3 H]bumetanide binding sites on a maximally stimulated duck erythrocyte (3750), the actual number of erythrocytes subjected to immunoprecipitation (ϳ1.4 ϫ 10 8 ), and the average efficiency of immunoprecipitation (ϳ80%). 2 The number of phosphate groups associated with the cotransporter was calculated from the radioactivity of the 146-kDa cotransporter band and the measured specific activity of the phosphate source, i.e.
[␥-32 P]ATP. The results of four experiments in which the relationship between cell water content and phosphorylation stoichiometry was measured is shown in Fig. 4. In cells of normal volume, the nominally active cotransporter contained 2.3 Ϯ 0.9 phosphate groups. Osmotic swelling reduced this ratio to 1.0 Ϯ 0.4, whereas shrinkage increased it to 5.8 Ϯ 1.2 (mean Ϯ S.E., n ϭ 4). The analysis therefore indicates that each cotransporter acquires 4.8 Ϯ 0.9 phosphates on going from an inactive state in swollen cells to an active state in shrunken cells. 3 2 C. Lytle, unpublished results. 3 This analysis assumes that each cotransporter binds a single molecule of [ 3 H]bumetanide avidly upon activation (42), and therefore ignores transporters that fail to respond to a given stimulus. The existence of "reserve cotransporters" is not unprecedented. In the shark   FIG. 3. Changes in cell volume coordinately affect Na-K-Cl cotransport activity and Na-K-Cl cotransport protein phosphorylation. 32 P-Labeled erythrocytes were exposed for 12 min to a range of osmolality (220 -420 mosM). Cotransport protein was isolated by consecutive immunoprecipitation and SDS-PAGE, and its 32 P content (right ordinate) was determined by Cerenkov analysis. Cell volume and cotransport activity were measured as cell water content (abscissa) and bumetanide-sensitive 86 Rb influx rate (left ordinate) on companion cells lacking 32 P over the interval between 11 and 13 min. The water content of unstimulated 32 P-labeled erythrocytes (1.55 liters/kg of cell solid) is indicated by the shaded bar. Similar results were obtained in three additional experiments.
FIG. 4. Stoichiometry of Na-K-Cl cotransport protein phosphorylation as a function of cell volume. Erythrocytes were labeled with 32 P for 3 h, then exposed to an isotonic (323 mosM), hypotonic (223 mosM), or hypertonic (410 mosM) medium. After 9 min, samples from each cell suspension were obtained for analysis of wet cell mass, dry cell mass, [ 32 P]ATP, and [ 32 P]cotransport protein as described under "Experimental Procedures." Data are expressed as moles of 32 P incorporated into each mole of cotransport protein versus cell water content (volume). Shaded bar denotes normal (isosmotic) cell water content. Different symbols represent results from four independent experiments.
FIG. 5. Coordinate effects of four stimuli on Na-K-Cl cotransporter activity and phosphorylation: lack of additivity. Aliquots of duck erythrocytes (one labeled with 32 P) were exposed for 12 min to stimuli, either separately or in combination, at their maximally effective doses: 10 M norepinephrine (NE), 10 mM sodium fluoride (F Ϫ ), 100 mM sucrose (hypert), or 0.2 M calyculin-A (cal). hypot, hypotonic. Cotransport activity (open bars) was measured as bumetanide-sensitive uptake of 86 Rb between 11 and 13 min. Phosphorylation (shaded bars) was measured as 32 P content of cotransport protein at 12 min. Values were normalized to those with calyculin-A (designated as 100% maximal). Each bar represents the mean Ϯ S.E. obtained in four to six experiments.
Magnitude and Additivity of Stimuli-Paired measurements of cotransporter activity and phosphorylation indicated that norepinephrine, sucrose, fluoride, and calyculin-A, when applied individually at optimally effective concentrations, evoked similar levels of cotransport activity and cotransport protein phosphorylation (Fig. 5). With norepinephrine, activity increased 13-fold, and this was associated with a 5.7-fold increase in phosphorylation. The response was mimicked by 50 M 8-chlorophenylthio-cAMP, a permeant analog of cAMP (data not shown), consistent with previous evidence that norepinephrine acts by stimulating PKA (11,24). Coordinate increases in activity and phosphorylation were also observed in response to stimuli that do not involve cAMP (cell shrinkage, fluoride, and calyculin-A). Fluoride raised activity and phosphorylation ( Fig.  5) with an effect half-maximal at ϳ3 mM and maximal at 10 mM (data not shown). The highest levels of activity and phosphorylation were evoked by calyculin-A. Importantly, the effects of the four stimuli were not additive; application of two stimuli in combination evoked no more activity or phosphorylation than did the more potent of the two stimuli acting alone. This suggests that the four stimuli activate the cotransport protein by promoting its phosphorylation at common regulatory sites.
Changes in Cotransporter Activity and Phosphorylation Are Temporally Correlated-Cell shrinkage increased cotransporter activity and phosphorylation to plateau levels within about 2.5 min (Fig. 6). A rapid reversal of activity and phosphorylation occurred on addition of staurosporine, a protein kinase inhibitor with broad specificity (34).
Better temporal resolution of the activation and deactivation processes was obtained by cooling the cells to 30°C. Under these conditions, activation in response to cell shrinkage or calyculin-A commenced without a discernible delay and conformed to a single exponential function (Fig. 7A). Activation by calyculin-A was faster (t1 ⁄2 ϳ 0.8 min) than that by cell shrinkage (t1 ⁄2 ϳ 5 min). When applied to resting cells, staurosporine (15 M) rendered the cotransporter refractory to cell shrinkage and calyculin-A (Fig. 7A) and to fluoride and norepinephrine (data not shown). When added to cells already stimulated by osmotic shrinkage, staurosporine caused cotransporter activity and phosphorylation to subside rapidly (Figs. 6 and 7B). This reversal was not observed in cells prestimulated with calyculin-A (Fig. 7B), indicating that staurosporine deactivates the cotransporter by inhibiting its phosphorylation rather than by stimulating its dephosphorylation. Thus, each of the four stimuli appears to be transduced by a kinase that is inhibited, either directly or indirectly, by staurosporine.
Staurosporine Blocks Each Stimulus Equipotently-If all modes of activation involve the same staurosporine-sensitive step, each stimulus should be inhibited by staurosporine equipotently. To test this hypothesis, erythrocytes were preincubated for 10 min with various doses of staurosporine before stimulation by norepinephrine, fluoride, or hypertonicity. As shown in Fig. 8, all three stimuli were inhibited to a halfmaximal extent by a similar concentration of staurosporine (ϳ0.7 M).
One-dimensional Phosphopeptide Analysis-To assess the distribution of phosphorylation sites, the cotransport protein was chemically fragmented with N-chlorosuccinimide, an agent that selectively cleaves tryptophanyl peptide bonds (32). After treatment of 32 P-labeled cells with various stimuli, the cotransporter was isolated by immunoprecipitation and SDS-gel electrophoresis. Gel bands containing the 147-kDa cotransporter were then treated with N-chlorosuccinimide, and 32 P-labeled cleavage products were analyzed by SDS-gel electrophoresis and autoradiography. Chemical cleavage for 20 min produced two major 32 P-labeled fragments of 82 and 41 kDa (Fig. 9). The fragments appear to be different domains since (i) fragments of identical size were obtained with cleavage times half and twice as long, (ii) treatment with fresh N-chlorosuccinimide for an additional 20 min failed to convert the isolated 82-kDa fragment into the 41-kDa fragment, and (iii) a monoclonal antibody that recognizes the carboxyl terminus of the cotransport protein (26) recognized only the 41-kDa fragment on Western blots of N-chlorosuccinimide-treated protein (data not shown). With all four stimuli, each of the two fragments was phosphorylated to roughly similar extents. Fragments obtained from calyculinstimulated cotransporters exhibited greater phosphorylation and slower electrophoretic mobility (88 and 45 kDa), which may reflect changes in the folding, net charge, or SDS-binding properties associated with the higher degree of phosphorylation. These results indicate each stimulus promotes phosphorylation of two large domains, one of which comprises part of the carboxyl terminus.
Regulatory Phosphoacceptors-Phosphoamino acid analysis of the cotransport protein (Fig. 10) indicated that each of the four stimuli promote phosphorylation at serine and threonine residues. Phosphotyrosine was not detected in either resting or stimulated cotransporters.
Different Stimuli Promote Phosphorylation of Common Sites-Two-dimensional phosphopeptide maps of cotransporters isolated from stimulated erythrocytes revealed a distinctive pattern of eight prominent tryptic phosphopeptides (designated 1-8 in Figs. 11 and 12). Maps of cotransporters phosphoryl-rectal gland, for example, only a small fraction of potentially functional cotransport units respond to a single application of a stimulus (43). In the avian erythrocyte, however, most cotransporters are presumed to be responsive since similar rates of cotransport are evoked if stimuli are applied separately, in combination, or in succession. The analysis also assumes that all cotransporters are structurally equivalent and that 32 P is distributed among them uniformly; this assumption is supported by the observation that deglycosylated duck cotransporters colocalize on Western blots and autoradiographs as single coherent bands. ated in response to cell shrinkage, fluoride, and norepinephrine were qualitatively indistinguishable (Fig. 11). None of the eight spots were detected in maps of unstimulated cotransporters (Fig. 11, control).
Because calyculin-A evoked more cotransporter phosphoryl-ation and activity than did other stimuli (Figs. 2, 5, and 9), it was important to determine whether the phosphatase inhibitor promotes phosphorylation of different sites or more complete phosphorylation of the same sites. To distinguish between these possibilities, phosphopeptide maps of cotransporters stimulated by cell shrinkage and/or calyculin-A were compared. The patterns of phosphopeptides obtained with cell shrinkage and calyculin-A were similar (Fig. 12) and resembled those obtained with norepinephrine and fluoride (Fig. 11). No additional phosphopeptide spots were observed in maps of cotransporters stimulated by cell shrinkage and calyculin-A si- FIG. 8. Staurosporine equipotently blocks activation of Na-K-Cl cotransport by norepinephrine, cell shrinkage, and fluoride. Duck erythrocytes were preincubated in a medium containing staurosporine at the concentrations indicated. After 10 -12 min, an activator of cotransport (10 M norepinephrine, 10 mM fluoride, or 100 mM sucrose) was added, and 7 min later the initial rate of 86 Rb uptake was measured over the ensuing 3-min interval. The bumetanide-inhibitable component of influx was normalized to that evoked by the stimulus in the absence of inhibitor (designated as 100%). Each point represents the mean Ϯ S.E. of three experiments in which influxes were measured in duplicate. The effect of each stimulus was half-maximally inhibited by ϳ0.7 M staurosporine.
FIG. 9. All four stimuli promote phosphorylation of the Na-K-Cl cotransporter within 82-and 41-kDa domains generated by N-chlorosuccinimide cleavage. Duck erythrocytes were labeled with 32 P, then stimulated for 12 min with hypertonicity (ϩ100 mM sucrose), 10 M norepinephrine (norepi), 10 mM sodium fluoride (F Ϫ ), or 0.2 M calyculin-A. After isolation by immunoprecipitation and SDS-PAGE, cotransport protein was treated with N-chlorosuccinimide for 20 min. Cleavage products were separated by SDS-PAGE and detected by autoradiography. 32 P-labeled fragments of 82 and 41 kDa, or 88 and 45 kDa in calyculin-treated cells, were identified. multaneously (Fig. 12). Hence, the greater degree of cotransporter phosphorylation and activity observed with calyculin-A appears to reflect a more complete phosphorylation of the same sites phosphorylated with other stimuli.
Whether each spot represents a unique phosphorylation site is uncertain. Some spots might represent precursors of others, or contain mixtures of different peptides, or contain a single peptide with multiple phosphorylation sites. Incomplete trypsinolysis is unlikely, however, since labeled cotransporters were digested overnight twice with fresh trypsin in great excess, and since different digests yielded consistent phosphopeptide patterns. Given the uniformity of each major spot, inadequate separation of peptide mixtures in two dimensions is also unlikely. It is apparent that upon activation a heterogeneous array of tryptic peptides are phosphorylated and that a similar array is observed with each stimulus.

DISCUSSION
The experiments described here suggest that activators of Na-K-Cl cotransport in the avian erythrocyte (cell shrinkage, cAMP, fluoride, and calyculin-A) promote phosphorylation of the cotransport protein at a common constellation of serine and threonine residues. These results substantiate the concept that the cotransport protein is regulated by direct phosphorylation (11,13,14) and suggest that different stimuli act through the same kinase. The single kinase theory is supported by three lines of evidence. First, application of any two stimuli in combination evokes no greater cotransport activity or phosphorylation than does the more potent stimulus alone (Fig. 5). If the phosphorylation evoked the four stimuli was catalyzed by different kinases acting on separate sites, application of the stimuli in pairs should yield the sum of the phosphorylation produced by the stimuli individually. The data indicate, rather, that phosphorylation by one stimulus precludes further phosphorylation by a different stimulus, and therefore suggest that the four stimuli act on common sites. Second, phosphopeptide maps of cotransporters activated by the four different stimuli are qualitatively indistinguishable (Figs. 11 and 12). Third, stauroporine equipotently blocks activation of the cotransporter by cell shrinkage, cAMP, and fluoride (Fig. 8), consistent with the hypothesis that each signal is transduced to the same sites by the same kinase.
The molecular switch on the cotransport protein that controls ion translocation appears to involve several serine and threonine residues. Estimates of phosphorylation stoichiometry indicate that the cotransport protein acquires ϳ5 phosphates on going from an inactive state in swollen cells to an active state in shrunken cells, and phosphopeptide maps show incorporation into a heterogeneous array of tryptic peptides. Chemical cleavage studies using N-chlorosuccinimide suggested that the incorporated phosphate is distributed evenly between amino and carboxyl segments of the cotransport protein. This agent splits the cotransporter into two immunologically distinct domains (82 and 41 kDa) whose combined mass (121 kDa) approaches that of the intact protein (145 kDa). With all four stimuli, comparable quantities of phosphate are incorporated into the different domains. Although the molecular structure of the avian cotransporter is undefined, known mem- FIG. 10. Phosphoamino acid analysis of Na-K-Cl cotransport protein. 32 P-Labeled duck erythrocytes were stimulated for 12 min by exposure to 10 M norepinephrine (norepi), hypertonicity (ϩ100 mM sucrose), 10 mM fluoride, or 0.2 M calyculin-A. Cotransport protein was immunoprecipitated, separated by SDS-PAGE, and electrophoretically transferred to PVDF membrane. Acid hydrolysates containing equivalent amounts of 32 P were separated by thin layer electrophoresis and 32 P-labeled amino acids were detected by autoradiography. Migration of phosphoserine (P-Ser), phosphotyrosine (P-Tyr), and phosphothreonine (P-Thr) standards from the origin is indicated on the left. One of two similar experiments is shown.
FIG. 11. Two-dimensional phosphopeptide maps of Na-K-Cl cotransporters activated by hypertonicity, norepinephrine, and fluoride. Cotransport protein was isolated from 32 P-labeled duck erythrocytes before (control) or after stimulation with norepinephrine (norepi), hypertonicity, or sodium fluoride, then subjected to SDS-PAGE and digested with trypsin. Tryptic peptides were separated in two dimensions and detected by autoradiography using a storage phosphor screen. Images were obtained using equivalent exposure times, gains, and background cutoffs. Top panel (key) schematically depicts the positions of major phosphopeptides and directions of peptide migrations (arrows) away from origin (*). Spots containing lesser and/or variable quantities of 32 P are denoted as unfilled circles. Qualitatively similar maps were obtained in three other experiments.
bers of the Na-K-Cl cotransporter family possess only 16 tryptophan residues, all of which are conserved between diverse animal species (35). If the avian cotransporter contains the same conserved tryptophans, complete cleavage by N-chlorosuccinimide would produce 16 fragments, the largest being 29.3 kDa. It is likely, therefore, that the 41-kDa and 82-kDa phosphorylated fragments are generated by partial cleavage, i.e. at particular tryptophan residues. Since both large fragments are dynamically phosphorylated, each must contain segments that are cytoplasmically disposed in vivo. The finding of major phosphorylation sites in putative NH 2 -and COOH-terminal domains corroborates previous work on the chloride-secreting cells of the shark rectal gland. In these cells, as in duck erythrocytes, the Na-K-Cl cotransporter is stimulated and phosphorylated by cell shrinkage, cAMP, and calyculin-A at threonine and serine residues (13,18). Two of the threonine residues have been located, one (Thr-1114) in the carboxyl domain that responds to cell shrinkage and another (Thr-189) in the amino domain that responds to cAMP (13). Importantly, no region of the shark cotransporter, including that surrounding Thr-189, conforms to the consensus motif for PKA, substantiating the concept that the effect of cAMP is indirect.
Evidence that all stimuli drive the cotransporter into the same chemical and functional form suggests that the cotransporter exists in just two states: resting and active-phosphorylated. Interconversion between these states could reflect a competition between a single kinase and a single phosphatase. The rapidity at which resting state converts to active-phosphorylated state after addition of calyculin-A indicates that the kinase and phosphatase are active simultaneously and that the phosphatase outpaces the kinase in unstimulated cells (7,11). As this phosphatase is more susceptible to inhibition by calyculin-A than to okadaic acid (7), it appears to be type-1 (PP-1). Since PP-1 is known to be inhibited by cAMP-dependent protein kinase A via inhibitor-1 in vivo, and by fluoride in vitro (36), it is possible that norepinephrine and fluoride, like calyculin-A, activate the cotransporter by hindering its dephosphorylation.
How cells perceive changes in their volume and then activate transport processes that restore volume is poorly understood. There is little doubt that the volume signal controls cotransport protein phosphorylation, but whether the signal regulates the kinase, the phosphatase, or both remains obscure. Jennings and Al-Rohil (37) surmised from the kinetics of swelling-induced K-Cl cotransport in rabbit erythrocytes that cell volume must effect phosphorylation rather than dephosphorylation, and dubbed the putative volume-sensitive enzyme "V-kinase." A shrinkage-stimulated kinase would be consistent with preliminary studies on duck erythrocytes, which demonstrate that upon phosphatase inhibition with calyculin-A, Na-K-Cl cotransporters become active and phosphorylated more quickly if the cells are shrunken. 2 In like manner, activation of Na/H exchange in lymphocytes following phosphatase inhibition is hastened by cell shrinkage (38). Few characteristics of the putative V-kinase are known, other than its apparent stimulation by cell shrinkage and relative insensitivity to commonly used sulfonamide kinase inhibitors.
Some evidence suggests that the volume signal might also modulate dephosphorylation. Palfrey and Pewitt (7) noted that cotransporters activated by cell shrinkage, unlike those activated by cAMP, are resistant to general kinase inhibition (by addition of K252a or by depletion of cellular ATP or Mg 2ϩ ) and surmised that cell shrinkage might suppress dephosphorylation. However, this hypothesis was not borne out by the present study with staurosporine. When this kinase inhibitor was added to shrunken cells, cotransport activity and cotransport protein phosphorylation subsided rapidly (Fig. 6), suggesting that the deactivating phosphatase remains highly active after cell shrinkage.
Although still obscure, the transmission of the volume signal does not appear to require protein kinase C, cGMP-dependent protein kinase activity, or Ca 2ϩ /calmodulin-dependent protein kinase II, since modulators of these kinases (phorbol esters, dibutyryl cGMP, and cytosolic free Ca 2ϩ ) have negligible effects on the phosphorylation state, activity, or volume responsiveness of the cotransport protein in duck erythrocytes. 2 The recent identification of Ca 2ϩ -calmodulin-dependent myosin light chain kinase (MLCK) as a shrinkage-stimulated kinase (16,39) prompted Klein and O'Neill to suggest that MLCK conveys the volume signal to the cotransport protein. This concept was based on two observations. First, volume changes in endothelial cells evoked parallel alterations in Na-K-Cl cotransport activity and myosin light chain phosphorylation; and second, both responses were inhibited equipotently by the MLCK inhibitor ML-7 (16). However, ML-7 did not block phosphorylation of the cotransport protein in response to cell shrinkage, suggesting that the effect of MLCK on the cotransporter is indirect, possibly through alterations in cytoskeletal structure (16). Further evidence that volume signal transduction does not require MLCK is that duck erythrocytes depleted of calcium (by preincubation in EGTA plus ionophore A23187) respond normally to cell shrinkage. 2 A direct role is also unlikely for PKA, since (i) stimulation of cotransport by cell shrinkage and fluoride occurs without a significant increase in cytosolic [cAMP] (11,24), (ii) stimulation by phosphatase inhibition with okadaic acid is not associated with increased PKA activity (11), (iii) stimulation by cell shrinkage persists in the presence of the kinase inhibitor K252a at concentrations that abolishes PKA activity (11), and (iv) the Na-K-Cl cotransporter of the shark rectal gland can be activated and phosphorylated in response to cAMP (13) although it lacks a consensus motif for PKA (40).
Although calyculin-A evokes greater cotransporter phosphorylation and activity than the other stimuli, this appears to FIG. 12. Comparison of phosphopeptide maps from hypertonically stimulated and calyculin-stimulated Na-K-Cl cotransporters. Cotransport protein was isolated from 32 P-labeled cells that had been exposed for 12 min to hypertonicity (ϩ100 mM sucrose), or to calyculin-A (cal, 0.2 M), or to both stimuli together. Tryptic digests were analyzed as in Fig. 11. Panel labeled key depicts major phosphopeptides numbered according to the scheme shown in Fig. 11. Qualitatively similar maps were obtained in two experiments. reflect a more complete phosphorylation of the same sites that are phosphorylated with the other stimuli. This conclusion is based on the fact that no additional phosphorylation occurs when calyculin-stimulated cells are exposed to another stimulus, and that phosphopeptide maps of cotransporters phosphorylated by the various stimuli are similar. From these data it can also be concluded that each phosphorylated site is dephosphorylated by a calyculin-sensitive phosphatase, presumably PP-1. In support of this notion, calyculin-A prevents dephosphorylation completely in vivo as well as in vitro. 2 It is therefore not surprising that, after addition of calyculin-A, the kinase, now unopposed by the phosphatase, drives cotransporters into the fully phosphorylated state.
An unresolved question is whether phosphorylation of the cotransport protein is both necessary and sufficient for activity. The finding that staurosporine blocks cotransport protein phosphorylation and activation supports the notion that phosphorylation is in fact necessary. Nevertheless, other modes of regulation that do not involve transporter phosphorylation cannot be excluded. Indeed, a recent investigation of the rat parotid gland revealed that some stimulators of cotransport activity (cAMP, AlF 4 Ϫ , and carbachol) evoke prominent increases in cotransport protein phosphorylation, whereas others (cell shrinkage, thapsigargin, and calyculin-A) do not (41). If such modes of regulation exist, they are not apparent in shark rectal gland cells (13,18), human colonic T84 cells, 2 or duck erythrocytes (this study) where osmotic shrinkage and calyculin-A do promote cotransport protein phosphorylation.
In summary, these results suggest that the cotransport protein is phosphorylated at a common set of serine and threonine residues in response to cell shrinkage, cAMP, fluoride, and calyculin-A. If confirmed by phosphopeptide sequence analysis, these results would obviate the need for multiple kinases acting on disparate sites of the cotransport protein and allow for a more simple model in which cotransport activity depends on the relative rates of one kinase and one phosphatase.