Activation of the Na-K-Cl Cotransporter NKCC1 Detected with a Phospho-specific Antibody*

The Na-K-Cl cotransporter NKCC1 is activated by phosphorylation of a regulatory domain in its N terminus. In the accompanying paper (Darman, R. B., and Forbush, B. (2002)J. Biol. Chem. 277, 37542–37550), we identify three phosphothreonines important in this process. Using a phospho-specific antibody (anti-phospho-NKCC1 antibody R5) raised against a diphosphopeptide containing Thr212 and Thr217of human NKCC, we were readily able to monitor the cotransporter activation state. In 32P phosphorylation experiments with rectal gland tubules, we show that the R5 antibody signal is proportional to the amount of 32P incorporated into NKCC1; and in experiments with NKCC1-transfected HEK-293 cells, we demonstrate that R5-detected phosphorylation directly mirrors functional activation. Immunofluorescence analysis of shark rectal gland shows activation-dependent R5 antibody staining along the basolateral membrane. In perfused rat parotid glands, isoproterenol induced staining of both acinar and ductal cells along the basolateral membrane. Isoproterenol also induced basolateral staining of the epithelial cells in rat trachea, whereas basal cells in the subepithelial tissue displayed heavy, non-polarized staining of the cell membrane. In rat colon, agonist stimulation induced staining along the basolateral membrane of crypt cells. These data provide direct evidence of NKCC1 regulation in these tissues, and they further link phosphorylation of NKCC1 with its activation in transfected cells and native tissue. The high conservation of the regulatory threonine residues among NKCC1, NKCC2, and NCC family members, together with the fact that tissues from divergent vertebrate species exhibit similar R5-binding profiles, lends further support to the role of this regulatory locus in vivo.

membrane, activated in response to secretagogues, and paramount for the transepithelial secretion of Cl Ϫ and water (2). NKCC1-mediated Cl Ϫ secretion has been well documented in rat parotid gland (3,4), shark rectal gland (5), rat colon (6), and dog trachea (7). At least in shark rectal gland, the evidence is consistent with an indirect activation of NKCC1 upon agonist stimulation: secretagogues cause a drop in intracellular [Cl Ϫ ] and volume via protein kinase A-mediated phosphorylation of apical chloride channels; in turn, low intracellular [Cl Ϫ ] and low cell volume provide activation stimuli for the currently unidentified NKCC1 kinase(s) (2,8).
Our laboratory (9) and others (10) have linked the phosphorylation of the intracellular N-terminal domain with NKCC1 activation. In the accompanying paper, Darman and Forbush (1) describe the phosphorylation of three residues in this regulatory domain, of which Thr 184 and Thr 189 are necessary for maximal sNKCC1 1 activation. Thr 189 in particular is demonstrated as being essential for sNKCC1 up-regulation. The three phosphoacceptor sites reside in a 30-amino acid region of the N terminus that exhibits 80% homology between sodium-coupled cation chloride cotransporter isoforms and 97% homology within NKCC1 proteins in species ranging from shark to human (shark Thr 184 /Thr 189 correspond to human Thr 212 /Thr 217 ). This region of the N terminus also contains a protein phosphatase-1-binding site (RVXF), which is conserved across divergent species and thought to mediate regulated dephosphorylation (11).
To date, most studies of NKCC regulation have been conducted in isolated tissue preparations, cell cultures, or cell-free systems utilizing [ 3 H]benzmetanide binding, 32 P incorporation, or isotopic uptake. Although these studies have provided a wealth of information at the cellular and molecular levels, little is known about the phosphorylation state of NKCC in native tissue. To address this issue, we have developed a phosphospecific polyclonal antibody (anti-phospho-NKCC1 antibody R5) raised against the in vitro phosphorylated peptide corresponding to the regulatory domain.
In this report, we demonstrate R5 specificity and sensitivity in recognizing the phosphorylated conserved threonines. We determine a positive correlation between NKCC1 activation and Thr 184 /Thr 189 phosphorylation in sNKCC1-transfected HEK-293 cells using the R5 antibody. We also investigate in vivo NKCC1 activity and address the universality of this regulatory domain by examining R5 immunohistographs of shark rectal gland and several rat secretory tissues. These data provide a critical link between molecular regulation studies and the activation profile of NKCC1 in vivo.

EXPERIMENTAL PROCEDURES
Antibodies-A 16-amino acid peptide, Tyr h208 -Arg h223 -Lys (YYLRT*FGHNT*MDAVPRK) with an additional C-terminal lysine residue for coupling, was synthesized by the Keck Peptide Synthesis Facility at Yale University; the regulatory phosphothreonines Thr 212 and Thr 217 (in human NKCC1) were incorporated directly during synthesis. A rabbit antibody (anti-phospho-NKCC1 antibody R5; Pocono Rabbit Farms, Canadensis, PA) was raised against this peptide coupled to maleimidobenzoic acid-N-hydroxysuccinimide-activated keyhole limpet hemocyanin (Sigma) using standard procedures. The anti-phospho-NKCC1 antibody will be referred to as R5 for the remainder of the paper. For immunofluorescence studies, R5 was subjected to positive and negative affinity purification using the phosphorylated and nonphosphorylated kindred peptides coupled to N-hydroxysuccinimide-activated Sepharose beads (Amersham Biosciences) following the manufacturer's instructions.
The specificity of R5 antisera for the immunizing peptide was measured using an ELISA format. As illustrated in Fig. 1 (upper panel), R5 serum displayed Ͼ20-fold selectivity for the diphosphopeptide compared with the non-phosphorylated peptide. This high degree of selectivity enables a sensitive analysis of the phosphorylation state of the conserved threonines in situ and also Western blot analysis; for most purposes, the antibody can be effectively employed as diluted serum. Positive and negative affinity purification of the antibody further increased the phospho-specificity to Ͼ100-fold (Fig. 1, lower panel); the purified antibody was used for immunofluorescence analyses.
Unnoticed until after the initial batch of R5 antibody had been produced, the second aspartate in this peptide was found to have been modified to an aspartimide during synthetic procedures; the unmodified peptide was subsequently synthesized for analytic and purification procedures. Fig. 1 shows that the R5 antibody had a slightly higher affinity for the modified peptide compared with the unmodified peptide. We speculate that the modification may have increased the immunogenicity of the Tyr h208 -Arg h223 -Lys peptide. In any case, this antigen was quite successful; the second of two rabbits (R4, not discussed further here) also produced a high titer phospho-specific serum, with greater phospho-specificity, but less NKCC specificity, compared with R5.
Several other antibodies were used to detect NKCC1 independent of its phosphorylation state. The T4 monoclonal antibody (12) and the N1c polyclonal antibody (gift of Chris Lytle, University of California, Riverside, CA) (13) both recognize the C terminus of human NKCC1 in a wide range of species. J3, J4, and J7 are anti-NKCC1 monoclonal antibodies that are highly specific for the shark cotransporter (14); these were used to detect sNKCC1 in ex vivo perfused fixed gland sections, isolated shark rectal gland tubules, and sNKCC1-transfected HEK-293 cells.
ELISA and Western Blotting-For determination of antibody affinity and phospho-specificity by ELISA, phosphorylated and non-phosphorylated peptides were covalently coupled to N-oxysuccinimide binding plates (DNA-Bind, Costar Corp.). Plates were blocked with 20 mM ethanolamine, pH 8.2; washed twice with wash buffer (PBS, 1% bovine serum albumin, and 0.1% Triton X-100); and blocked with 7% milk in PBS in 0.1% Triton X-100. After washing once with wash buffer and sequential incubations with serially diluted sera and horseradish peroxidase-conjugated anti-rabbit IgG antibody, the optical density was measured with a spectrophotometer using o-phenylenediamine as a substrate.
For Western blotting, samples were subjected to Tricine/SDS gel electrophoresis (7.5 or 10%) and transferred to polyvinylidene difluoride membranes (Immobilon, Millipore Corp., Bedford, MA). Blots were sequentially probed with primary and horseradish peroxidase-conjugated secondary antibodies. Chemiluminescent substrate was detected using a cooled CCD camera.
NKCC1 in HEK-293 Cells-Lines of stably transfected HEK-293 cells were handled as described in the accompanying paper (1). Following appropriate preincubations, cells were solubilized by one of two procedures. (a) Cells were lysed either in 20 mM HEPES, pH 7.2 (see Fig. 2), or in phosphatase inhibitory buffer (300 mM NaCl, 60 mM NaF, 10 mM EDTA, 30 mM Na 2 HPO 4 , 30 mM sodium pyrophosphate, 40 mM HEPES, 0.2 mM NaVO 4 , and 0.5 M calyculin A) (see Fig. 3), with both buffers containing 1% Triton X-100 and protease inhibitors (1). Lysate was centrifuged to remove insoluble material and diluted into SDS sample buffer. (b) Alternatively, cells were lysed in 1% SDS and 1 M H 3 PO 4 and subsequently diluted into sample buffer for Western blotting or dot blotting. For reasons that are yet unclear, we consistently saw a higher level of background R5 signal on dot blots (see Figs. 5 and 6) and gels (data not shown) of samples stopped with H 3 PO 4 /SDS stop medium compared with that seen when cells were harvested in 1% Triton X-100 (see Figs. 3

and 4).
NKCC1 32 P Incorporation in Shark Rectal Gland Tubules-Shark rectal gland tubule isolation and 32 P incorporation were conducted as previously described (1,15). At specific time points following appropriate incubations, 200-l aliquots of tubules were pelleted; resuspended in 100 l of phosphatase inhibitory buffer containing 3% Triton X-100, 1 M calyculin A, and a mixture of protease inhibitors; snap-frozen in liquid nitrogen; and stored at Ϫ80°C. For NKCC1 immunoprecipitation, samples were thawed and centrifuged, and J4 antibody-Sepharose beads (15) were added to the supernatant. After an overnight incubation at 4°C with subsequent washing, samples were subjected to SDS gel electrophoresis with Western blotting as described above. 32 P incorporation into proteins was analyzed on gels and blots using a Phospho-rImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Tissue Perfusion and Fixation-Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were anesthetized via intraperitoneal injection of sodium pentobarbital (50 mg/kg of body weight). Each animal was kept warm using a heated plate, and body temperature was monitored with a rectal thermometer. In most experiments, a polyvinyl catheter was introduced into the appropriate artery (common carotid for the parotid gland, thoracic for the trachea, and mesenteric for the colon), and the animals were infused with prewarmed Krebs-Henseleit bicarbonate solution (140 mM Na ϩ , 5.4 mM K ϩ , 1.2 mM Mg 2ϩ , 1.2 mM Ca 2ϩ , 124 mM Cl Ϫ , 21 mM HCO 3 Ϫ , 2.4 mM HPO 4 2Ϫ , 0.6 mM H 2 PO 4 Ϫ , pH 7.4; 300 mosmol/liter adjusted with mannitol). Agonists were added to the perfusate to stimulate cotransporter activity as explained in the figure legends for each particular tissue. For Western blot analysis of parotid tissue, the gland was removed and snap-frozen in liquid nitrogen. Frozen, pestle-ground tissue was placed in boiling SDS sample buffer.
For organ fixation, periodate/lysine/paraformaldehyde fixative was perfused through the same catheter for 5-15 min (16). Tissues were removed from the animal and further fixed for 2-4 h at 4°C. To achieve best fixation results, the trachea was also filled with fixative at 20 cm of H 2 O. Shark rectal glands were isolated from the animals, and each FIG. 1. ELISA analysis of R5 antibodies. Phosphorylated (q) and non-phosphorylated () Tyr h208 -Arg h223 -Lys peptides, the modified phosphorylated Tyr 208 -Arg 223 -Lys peptide (E), and the water control (f) were covalently coupled to N-oxysuccinimide plates and probed with various dilutions of R5 serum and affinity-purified R5 antibody as shown. Values represent means and range of duplicate determinations of the optical density of the ELISA reaction product.
gland was perfused through its unique artery and treated as described for rat tissues.
Immunofluorescence-Small sections of fixed tissues were cryoprotected with 50% polyvinylpyrrolidone in 2.3 M sucrose overnight at 4°C. Tissue sections were then mounted on aluminum nails and snap-frozen in liquid nitrogen. Semi-thin 0.5-1-m sections were cut using a Reichert Ultracut E ultramicrotome fitted with a FC-4E cryoattachment. Sections were placed on coated slides (Superfrost Plus, Erie Scientific, Portsmouth, NH) and washed with PBS prior to antigen retrieval by incubating the tissues with 1% SDS for 5 min. After quenching with NH 4 Cl for 15 min, sections were blocked with 0.1% bovine serum albumin and 10% goat serum in PBS, incubated for 1-2 h with the primary antibody, washed, and incubated with the appropriate secondary antibody conjugated to fluorescein isothiocyanate or Alexa dyes (Molecular Probes, Inc.). Sections were washed three times with PBS and mounted with Vectashield (Vector Laboratories, Inc., Burlingame, CA), and micrographs were taken with a Zeiss Axiophot mi-croscope on Kodak Tmax 100 film (Eastman Kodak Co.) or using an Olympus Fluoview confocal microscope.

RESULTS AND DISCUSSION
In the accompanying paper (1), Darman and Forbush identify three phosphoacceptor threonines in an activation domain in the N terminus of shark NKCC1. In particular, phosphorylation of Thr 184 and Thr 189 is highlighted as a key element in NKCC1 cotransport activation. Because this region is well conserved both across species and between isoforms, it is an excellent candidate as a universal regulatory motif for NKCC. Thus, we prepared a diphosphorylated peptide corresponding to this region of the human NKCC1 sequence (Tyr h208 -Arg h223 -Lys) and raised a polyclonal antibody (R5) using the keyhole limpet hemocyanin-conjugated peptide as antigen.
Analysis of the Phosphorylation State of the Regulatory Threonines in sNKCC1-and hNKCC1-transfected HEK-293 Cells-The specificity of R5 for phosphorylated NKCC is illustrated by the Western blot in Fig. 2. As shown here, R5 recognized a single protein of ϳ180 kDa, corresponding to the Na-K-Cl cotransporter (see Fig. 3). HEK-293 cells transfected with shark or human NKCC showed a strong signal compared with native NKCC in HEK-293 cells transfected with vector alone. Cross-reactivity of the R5 antibody with other proteins was not detected in HEK-293 cells and is rarely seen in lysates of eukaryotic cells.
To investigate the activation dependence of the R5 signal, we FIG. 2. Detection of NKCC1 phosphorylation using the R5 antibody. sNKCC1-, vector-, and hNKCC1-transfected HEK-293 cells were incubated for 1 h in low Cl Ϫ hypotonic medium (ϩ) or high K ϩ medium (Ϫ) to activate or deactivate NKCC1. After solubilization of the cells in Triton X-100 with protease inhibitors, samples were incubated for an additional 90 min at room temperature in the presence (ϩ) or absence (Ϫ) of 0.5 M calyculin A (cal. A), prior to the addition of SDS sample buffer, gel electrophoresis, and Western blotting.
FIG. 3. Recognition of Thr 184 and Thr 189 mutants by the R5 antibody. sNKCC1-and vector-transfected HEK-293 cells were preincubated in low Cl Ϫ hypotonic medium (ϩ) or high K ϩ medium (Ϫ) for 50 min prior to lysis in Triton X-100 with protease and phosphatase inhibitors. After centrifugation and dilution into SDS sample buffer, identical volumes of solubilized protein representing 6 mm 2 of tissue culture monolayer were loaded. Coomassie Blue staining of blots showed ϳ20% variation in protein levels among these samples; to correct for this variation, individual lanes in all three panels have been digitally corrected accordingly. The sNKCC1 mutants are described in the accompanying paper (1)  incubated NKCC1-transfected HEK-293 cells in low Cl Ϫ hypotonic medium (3 mM Cl Ϫ solution diluted 2-fold with water) or in high K ϩ medium (135 mM NaCl, 5 mM RbCl, 1 mM CaCl 2 / MgCl 2 , 1 mM Na 2 HPO 4 /Na 2 SO4, 15 mM NaHEPES, pH 7.4, and 10 mM K ϩ ) to activate or deactivate the transporter, respectively. As illustrated in Fig. 2, the antibody signal was severalfold greater in sNKCC1-and hNKCC1-transfected HEK-293 cells under activating conditions. As a further confirmation that the R5 signal is dependent upon phosphorylation of the cotransporter, the samples in this experiment were incubated at room temperature for 90 min following lysis of the cells to allow endogenous protein phosphatases to dephosphorylate NKCC1. As shown here, the R5 signal was almost completely eliminated by incubation in lysis buffer devoid of the protein phosphatase-1 inhibitor calyculin A.
We examined various NKCC phosphorylation site mutants to determine whether R5 recognizes the monophosphorylated as well as the diphosphorylated regulatory domains. Fig. 3 presents Western blots of transfected HEK-293 cells from such an experiment comparing vector alone, wild-type sNKCC1, and Thr 184 /Thr 189 mutants under the activation and deactivation conditions described above. Total shark NKCC1 was indicated by J3 antibody binding (Fig. 3, middle panel); as shown here, the expression was similar for each of the constructs. For both Thr 184 and Thr 189 single mutants, the R5 signal was found to increase with activation (Fig. 3, upper panel), and although the level was clearly greater than in control HEK-293 cells, there was much less signal than for wild-type sNKCC1. The simplest explanation for this result is that the R5 antibody, although having the greatest affinity for the diphosphorylated regulatory domain, also has significant although lower affinity for the monophosphorylated forms.
The J3 antibody consistently recognized two bands in transfected cells (Fig. 3, middle panel), and we presume that this is due to different degrees of glycosylation, possibly reflecting different subcellular compartmentalization. Only the upper of these two bands was phosphorylated upon activation, as demonstrated by R5 recognition (the T189D mutant may exhibit a small amount of another lower band). The identity of the R5positive band of lesser mobility in deactivated cells is a puzzle. Our interpretation is that this signal is from the endogenous HEK-293 cell cotransporter, based on two observations. (a) There was little or no J3 signal at this position (this R5 signal was from the region between the two strong J3 bands, better seen in superimposition of reblotting experiments (data not shown)); and (b) the upper J3 band exhibited only a very small upward mobility shift upon activation of the transporter. Surprisingly, and somewhat at odds with this interpretation, the native HEK-293 cell cotransporter appeared to undergo a mobility shift upon phosphorylation, despite the fact that the amount of HEK-293 cell R5 signal did not increase much upon activation (Figs. 2 and 3 and discussion below).
Relationship between the R5 Antibody Signal and 32 P Incorporation in NKCC1-To analyze the phosphorylation state of sNKCC1 residues Thr 184 /Thr 189 in their native cellular environment, we examined 32 P incorporation and R5 binding in sNKCC1-, T202E-, and vector-transfected cells were preincubated for two sequential 30-min preincubation periods as follows: black bars, basic/basic medium; dark gray bars, basic/0Na-0K-130Cl hypertonic medium; light gray bars, low Cl Ϫ hypotonic/0Na-0K-130Cl hypertonic medium; white bars, basic/low Cl Ϫ hypotonic medium. Following preincubation, 86 Rb influx was performed on one-fourth of the plate (upper panel, n ϭ 6 on one plate); and the other wells were sucked dry, and H 3 PO 4 /SDS was added for dot blotting with the R5 antibody (lower panel, n ϭ 18 on the same plate). A constant background value was subtracted from all dot-blot values; this represented 22, 50, and 54% of the maximum R5 dot signal in sNKCC1, T202E, and the vector, respectively. Data are plotted as means Ϯ S.E. relative to the value with the last of the four preincubation protocols. isolated shark rectal gland tubules preloaded with 32 P i (1). Tubules were activated by incubation either in hypertonic media or in isotonic media containing forskolin (to activate apical Cl Ϫ channels and to lower intracellular [Cl Ϫ ] (8)) and calyculin A. Under both conditions, NKCC1 displayed a time-dependent increase of 32 P incorporation in the 195-kDa band of NKCC1 as detected by 32 P phosphorimaging of samples subjected to gel electrophoresis and transferred to polyvinylidene difluoride membranes (Fig. 4A); immunoprecipitation with the J4 antibody highlighted 32 P phosphorylation of NKCC1 (Fig. 4B). Us-ing the R5 antibody, NKCC1 phosphorylation was readily detected even in the crude lysates (Fig. 4C). Fig. 4 (D-F) illustrate the quantitative relationship between the R5 signal and 32 P incorporation. It is clear from these data that R5 binding varies linearly with 32 P incorporation within our measurements, both in crude lysates (Fig. 4, D and E) and in immunoprecipitates (Fig. 4F). Thus, Thr 184 and Thr 189 are indicators of overall sNKCC1 phosphorylation under these conditions, providing rational support for the use of R5 as a quantitative tool.
To further establish the relationship between phosphorylation and NKCC1 activation, we compared the effects of several activation media on the R5 signal and 86 Rb influx. Fig. 5 illustrates the results of an experiment, similar to that of Fig.  8 of the accompanying paper (1), in which the time course of activation and deactivation was determined in various media. Following preincubation of cells in a single 96-well plate for each cell line, alternate rows were either lysed in H 3 PO 4 /SDS for dot-blot analysis (possible because only a single band was detected by R5; Fig. 1) or subjected to a 1-min 86 Rb influx assay. It is readily seen in Fig. 5 that function mirrors phosphorylation, i.e. activating and deactivating conditions had similar effects on both measurements (a small discrepancy between phosphorylation and flux for hNKCC1 cells in low Cl Ϫ hypotonic medium transferred to 0Na-0K-130Cl hypertonic medium was seen in two of four such experiments). Human and shark NKCC1 exhibited similar behavior, except that hNKCC1 was less activated by the hypertonic sucrose medium.
We have used similar experiments to answer a question regarding the mechanism by which the T202E mutation produces a profound change in the pattern of cotransporter acti- Values represent the ratio of phosphorylated NKCC1 to total NKCC1 given by R5 signal/T4 signal. Data are represented as means Ϯ S.D. and were tested by paired Student's t test for significance at the 5% level. d, immunohistographs of semi-thin sections of rat parotid gland perfused with shark Ringer's solution in the absence (left and right panels) or presence (middle panel) of isoproterenol (5 M). Tissues were probed with the R5 (left and middle panels) and N1c (right panel) antibodies, respectively. Ctrl, control. vation, abolishing the stimulation by hypertonic media (1). Fig.  6 presents the results of an experiment involving four activation conditions, each evaluated by R5 dot-blot analysis and by 86 Rb influx assays of samples from the same plate. For sNKCC1-transfected cells, the cotransporter was activated both by low Cl Ϫ hypotonic medium and by 0Na-0K hypertonic medium; the results are similar to those in Fig. 5 in that the changes in 86 Rb influx mirrored the changes in the R5 signal. For the T202E mutant, 86 Rb influx was slightly increased over basal levels in 0Na-0K-130Cl hypertonic medium compared with low Cl Ϫ hypotonic medium, and again this pattern was reproduced by the changes in the observed R5 signal. This result demonstrates that the effect of the Thr 202 mutation on cotransporter activation is the result of a change in the degree to which Thr 184 /Thr 189 are phosphorylated by various stimuli. Fig. 6 also illustrates an anomalous result with regard to the R5 signal of the endogenous HEK-293 cell cotransporter. Following preincubation in the normal flux medium, the R5 signal was high, although the flux was only ϳ20% activated (we suspect that a tendency of T202E in the same direction may be due to contamination by the endogenous HEK-293 cell signal). It is important to note that there was a relatively high level of R5 signal in HEK-293 cells (see the legend to Fig. 6 and "Experimental Procedures") and that there was only a small fractional change on top of this. This phenomenon is also seen in the Western blots of Figs. 2 and 3, where the change in the HEK-293 cell R5 signal was very small upon activation. The nature of the HEK-293 cell cotransporter remains an enigma: although reverse transcription-PCR results are more consistent with its identity as NKCC1, its regulatory behavior is distinctly different from that of overexpressed human NKCC1 in several regards (1,17).
Activation of NKCC1 in Intact Shark Rectal Gland-Previously, in vivo analysis of NKCC1 activation in native tissue has been difficult. Earlier studies by ourselves and others have used measurements of [ 3 H]benzmetanide binding, 32 P incorporation, or 86 Rb uptake to measure NKCC1 activation in vitro, but it is usually impractical to apply these techniques in an intact tissue or animal model. To carry out these analyses, the tissue must be extracted and digested to isolate epithelial cells, possibly altering the NKCC1 activation state prior to analysis. To date, there have been no studies analyzing the phosphorylation state of NKCC1 in native tissue. We investigated whether the R5 antibody can be used for measuring sNKCC1 activity in situ, utilizing tissue fixation to arrest enzymatic reactions that could alter phosphorylation states. In Fig. 7, immunohistographs of shark rectal gland perfused with shark Ringer's solution (240 mM NaCl, 4 mM KCl, 2.5 mM CaCl 2 , 1.25 mM MgSO 4 , 20 mM HEPES, 70 mM trimethylamine N-oxide, 350 mM urea, 1.0 mM Na 2 HPO 4 , and 5.6 mM D-glucose, adjusted to pH 7.8 at room temperature and bubbled with O 2 ) in the presence or absence of forskolin show a dramatic increase in R5 staining along the basolateral membrane of the secretagogueperfused tubular epithelium. Probing with the J4 antibody verified the subcellular localization of NKCC1 to this region under stimulatory and basal conditions. The strong correlation confirms the specificity of R5 as sufficient for in situ analysis and reveals the importance of the two conserved threonines in the regulation of sNKCC1 in its native tissue. In addition, although the R5 antibody was raised against the human sequence, its affinity and specificity in elasmobranches as well as teleosts (18) are consistent with conservation of the regulatory motif across evolutionarily distant species.
NKCC1 in the Parotid Gland-Numerous reports of divergent stimuli causing activation of NKCC1 in the parotid gland support a complex pattern of regulation in this tissue (3). In this study, we perfused the parotid gland in vivo in the absence or presence of the ␤-adrenergic receptor agonist isoproterenol; the subsequent rise in intracellular cAMP has been previously shown to increase NKCC1 activity in this tissue (19,20). In Fig.  8 (a and b), tissue homogenate subjected to gel electrophoresis and probed with the R5 antibody displayed a 170-kDa band whose identity as NKCC1 was confirmed by the T4 antibody. Densitometric band analysis revealed a 2-fold increase in the R5/T4 ratio under stimulatory conditions (Fig. 8c). In addition to the 170-kDa protein, the C-terminal T4 antibody detected a second band with an apparent molecular mass of 140 kDa. This band was not strongly recognized by the R5 antibody, and we presume it to represent immature, non-phosphorylated NKCC1 retained in intracellular compartments.
In situ analysis of rat parotid gland with the R5 antibody revealed pronounced staining along the basolateral membrane in ductal and acinar cells of glands perfused with isoproterenol (Fig. 8d). Basolateral staining with the N1c antibody confirmed that NKCC1 was expressed in both cell types. The finding that NKCC1 is abundantly expressed in the duct epithelium of rat parotid gland contrasts markedly with our previous finding of rabbit parotid gland NKCC1 expression only in the acini (12). This expression profile was confirmed (data not shown) and thus suggests a striking species-specific difference in the function of the duct epithelium.
NKCC1 in the Trachea-In the trachea, as in many other secretory tissues, NKCC1 works in concert with the cystic fibrosis transmembrane conductance regulator to aid in the vectorial secretion of fluid into the lumen (2). Haas et al. (7) have previously described NKCC1 activation in isolated dog epithelial cell preparations in response to ␤-adrenergic receptor agonists. This activation appears, like that in the rectal gland, to occur primarily via a transient reduction in intracellular [Cl Ϫ ] resulting from the apical exit of chloride. Fig. 9a demonstrates that NKCC1 expression is localized to the basolateral membrane of the epithelial cell layer of rat trachea. Interestingly, we also found heavy, non-polarized staining of the basal cells directly beneath this epithelial cell monolayer, as well as heavy N1c staining of the subepithelial gland (Fig. 9b, arrow). Fig. 9 (c and d) illustrates that the R5 antibody selectively stained along the basolateral membrane of the epithelial layer and the cell membrane of non-polarized basal cells in rats perfused with isoproterenol. This result demonstrates the activation of NKCC1 in the intact trachea in response to ␤-adrenergic receptor agonists, supporting the previous analysis with isolated cells. The high level of NKCC1 expression and activity in basal cells were unexpected. However, because these basal cells demarcate a proliferation zone (21) and NKCC1 activity has been linked to cell cycle regulation (reviewed in Ref. 22) and cell proliferation (23), we speculate that active NKCC1 could play an important role in mediating cell growth in this tissue.
NKCC1 in the Distal Colon-In this secretory tissue, NKCC1 activity has been linked to both K ϩ and Cl Ϫ secretion (24,25). This NKCC1-mediated secretory activity has been linked to both ␤and ␣ 2 -adrenergic receptors, respectively, through increases in intracellular [cAMP], with a resultant fall in intracellular chloride, and through increases in intracellular [Ca 2ϩ ], with the activation of basolateral K ϩ recycling channels (26).
In Fig. 10, the R5 antibody was used to address the activation state of NKCC1 in response to ␣and ␤-adrenergic receptor agonists (epinephrine and isoproterenol, respectively). In tissue perfused with isoproterenol, strong staining was observed in the base of the crypts and decreased markedly along the crypt-villus axis. NKCC1 was expressed throughout the crypt, as demonstrated by T4 antibody binding and similar to the expression profile of the gastric gland. This segregation of activity further supports the hypothesis that the base of the crypts constitutes a region of cell proliferation and secretion (27,28).
R5 staining was negative in epinephrine-perfused tissue (Fig. 10) and was similar to levels observed in control animals (data not shown). We believe that the absence of NKCC1 stim-ulation by epinephrine may be due to stimulation of G␣ i via ␣ 1 -adrenergic receptors in this colonic segment. This would correlate the response of NKCC1 to adrenergic stimuli with the local expression of receptor subtypes.
Limitations on the Use of the R5 Antibody-Although the R5 antibody has proven to be an outstanding tool for the investigation of cotransporter regulation, we have identified a number of limitations in its utilization. (a) It must be noted that, although R5 recognizes two phosphoacceptor sites in NKCC, available data imply that at least five phosphoacceptors are utilized in regulatory phosphorylation (see "Discussion" in the accompanying paper (1)). Also, although most experiments have supported the concept of phosphorylation by a single kinase, recent studies have raised the possibility of alternate modes of regulation (1,3). (b) In the work presented here, we examined regulation of NKCC1 in secretory tissues in which it is very highly expressed. We have found that R5 is not suitable for immunofluorescence studies in tissues with lower levels of cotransporter expression due to background staining of cytoplasm and particularly of nuclei. (c) We are puzzled by a high level of R5 background signal when phosphorylation and dephosphorylation are "stopped" by acid and SDS compared with when they are stopped by Triton X-100, EDTA, and phosphatase inhibitors (see "Experimental Procedures"). (d) Although the R5 signal mirrors transport function both in shark rectal gland cells and in HEK-293 cells overexpressing shark or human NKCC1, in wild-type HEK-293 cells, there is an anomalously high R5 signal under basal conditions (Figs. 2, 3, and 6). We currently have no explanation for this observation, but appreciate that it may complicate the interpretation of future experiments in similar cell types.
Conclusions-As demonstrated above, the R5 antibody enables us to directly investigate the phosphorylation state of the two conserved threonines in cells and tissues and to discern the subcellular localization of activated NKCC1 via immunofluorescence. We show the first reported data of NKCC1 activation in vivo using fixed sections of secretory tissue. Analyzing these data, we conclude that Thr(P) 184 and/or Thr(P) 189 in shark and the homologous mammalian residues parallel NKCC1 activation in vivo and that this phosphorylation profile is independent of the tissue type and vertebrate species tested. We propose that these regulatory threonines are part of a universal regulatory locus necessary for NKCC1 activation.