Na:K:2Cl cotransporter (NKCC) of intestinal epithelial cells. Surface expression in response to cAMP.

During intestinal chloride secretion, epithelial uptake of salts is accomplished largely by a bumetanide-sensitive Na:K:2Cl cotransporter designated here as NKCC. Using monoclonal antibodies directed against NKCC from the human crypt epithelial cell line, T84, we define its surface localization as a function of cotransporter activation. Immunoelectron microscopy, confocal localization, and selective surface biotinylation studies revealed that the 195-kDa NKCC protein is polarized to the basolateral domain. Following immunoprecipitation, several polypeptides coprecipitated with the 195-kDa cotransporter including two prominent proteins of molecular mass 160 and 130 kDa. Immunoblotting with three distinct anti-NKCC monoclonal antibodies in conjunction with deglycosylation experiments suggested that the 160- and 130-kDa bands represented novel proteins unrelated to the cotransporter. Stimulation of T84 monolayers with cAMP agonists, a condition which elicits chloride secretion and leads to microfilament-dependent NKCC activation, did not significantly increase the number of bumetanide-binding sites and only marginally increased surface expression of the 195-kDa cotransporter available for surface biotinylation. In contrast, cAMP agonist stimulation increased the surface expression of the coprecipitating 160- and 130-kDa proteins ∼6-fold. The increase in surface 160- and 130-kDa proteins was attenuated by phalloidin preloading the cells, a condition which also prevents activation of NKCC without influencing the activity of other membrane transporters participating in chloride secretion. These studies define the polarized distribution of the NKCC protein on intestinal epithelia, indicate that NKCC may be associated with two other previously unidentified membrane proteins and such association is influenced by the F-actin cytoskeleton.

Mucosal surfaces, such as the lining of the intestine, maintain hydration in part as a result of the ion transport process known as electrogenic Cl Ϫ secretion. Important human diseases result when the ability to mount Cl Ϫ secretion is either impaired (i.e. cystic fibrosis) or enhanced (i.e. secretory diar-rhea). Cl Ϫ secretion represents the orchestrated activities of four membrane transport events. Basolateral ion uptake is largely accounted for by the bumetanide-sensitive Na:K:2Cl cotransporter, designated in this report as NKCC. Human intestinal NKCC relates to a recently cloned 195-kDa glycoprotein with 12 predicted transmembrane domains (1). When stimulated by hormones such as VIP or by cAMP agonists, NKCC activity is up-regulated allowing for the electroneutral influx of a 1:1:2 ratio of sodium, potassium, and chloride ions in an electroneutral fashion (for recent review, see Ref. 2). Relative homeostasis of intracellular Na ϩ and K ϩ is achieved by the presence on the basolateral membrane of both the Na-K-ATPase pump and a channel-like K ϩ efflux pathway. Regulated opening of Cl Ϫ channels on the apical membrane in the presence of the electrochemical gradient favoring Cl Ϫ exit permits vectoral secretory movement of this anion which is coupled to paracellular Na ϩ and water movement resulting in net secretion of isotonic fluid (3,4).
Several recent studies of cAMP-mediated Cl Ϫ secretion have focused on discerning mechanisms by which the apical Cl Ϫ channel is regulated. However, in order to sustain Cl Ϫ secretion, basolateral Cl Ϫ uptake through NKCC must increase in parallel with Cl Ϫ channel activation (5). NKCC is complexly regulated in that cotransport activity responds to additional influences other than chemical potential. For example, the activity of NKCC has been shown to positively correlate with serine/threonine phosphorylation of the corresponding protein in shark rectal gland (6). In many, but not all tissues, cAMPmediated activation of NKCC has been shown to be accompanied by increased binding for the radiolabeled ligand bumetanide. It is not clear whether such bumetanide binding increments simply reflect activation of existing cell surface cotransporters or actual recruitment of additional cotransporter units to the cell surface.
Using T84 cells, a human intestinal epithelial model, we have previously shown that activation of NKCC, which parallels cAMP-mediated Cl Ϫ secretion, also occurs in conjunction with reordering of basolateral F-actin microfilaments (7). Stabilization of F-actin with phalloidin prevented both microfilament reordering and activation of basolateral NKCC. This effect was shown to be specific for the cotransporter since phalloidin loading did not adversely affect either Na-K-ATPase activity or the regulated efflux of Cl Ϫ and K ϩ through their respective channels (8). Studies in HT29 cells (a human intestinal epithelial cell line that lacks a cAMP-mediated chloride efflux pathway) and its subclone, Cl.19A cells (possessing a functional cAMP-regulated chloride efflux pathway), have yielded further insights into NKCC activation. These studies indicated that NKCC activity was up-regulated in both cell lines by cAMP; however, an increase in bumetanide-binding sites on the cell surface occurs only in the subclone, Cl.19A (9). Surprisingly, however, the cAMP-elicited increase in bumetanide binding observed in clone Cl.19A cells is unaffected when F-actin is stabilized even though such prevention of cytoskeletal reordering results in a ϳ60% attenuation of functional measures of NKCC activation. The studies conducted on the HT29 cell line further indicated that exposure to cAMP agonists increases the activity of NKCC by a mechanism which is independent of Cl Ϫ efflux, only minimally reduced by phalloidin, and is not associated with an increase in bumetanide binding. Thus it appears that two regulatory responses may exist: (a) that associated with Cl Ϫ efflux and dependent on F-actin rearrangement and (b) that which is independent of chloride efflux and not associated with changes in F-actin organization. The relationship of cAMP elicited changes in bumetanide binding to cAMP-elicited cytoskeletal remodeling in these studies remains unsettled, although such observations unmask the complexity of regulatory influences on this key salt uptake process, and in doing so highlight the need for studies aimed more directly at assessing the level of surface expression of the cotransporter during stimulated secretion.
Here we employ monoclonal antibodies raised against NKCC derived from the human intestinal epithelial cell line T84 for studies of the cell surface expression of the corresponding ϳ195-kDa protein under conditions in which the activity of the cotransporter is stimulated by cAMP agonists. By confocal and electron microscopic immunolocalization as well as selective surface biotinylation, the 195-kDa protein is shown to be distinctly polarized to the basolateral domain of T84 cells. Surface biotinylation experiments identified two prominent, previously unidentified proteins of ϳ160 and ϳ130 kDa that consistently coprecipitate with the cotransporter. The coprecipitating proteins also represent integral membrane proteins since they possess external residues available for surface biotinylation. Basolateral surface biotinylation experiments suggest that expression of the 160-and 130-kDa proteins is increased severalfold upon stimulation with the cAMP agonist forskolin while expression of the 195-kDa cotransporter protein, examined by both surface biotinylation and bumetanide binding, is only modestly influenced. Last, phalloidin loading which prevents both the cAMP-elicited activation of the cotransporter as well as the concurrent reordering of the basolateral cytoskeletal F-actin, also prevents the forskolin-elicited increase in surface expression of the 160-and 130-kDa co-precipitating proteins. These data suggest that the cotransporter may associate with two distinct integral membrane proteins and that such associated proteins may participate in the regulation of this key salt uptake pathway.

Electrophysiology
Electrophysiological studies were carried out at 37°C on T84 cells grown to confluency on 24-well Costar inserts used 7-14 days postplating. The inserts consist of an outer well or chamber that corresponds to the basolateral surface of the monolayer and an inner chamber that is in contact with the apical surface. Inserts are washed with Hanks' balanced salt solution (HBSS, 1 Sigma, without phenol red or bicarbonate plus 10 mM HEPES, pH 7.4) warmed to 37°C, and transferred to a new 24-well tissue culture plate containing fresh HBSS. To determine currents, transepithelial potential, and resistance a commercial voltage clamp (Bioengineering Department, University of Iowa) was interfaced with an equilibrated pair of calomel electrodes submerged in saturated KCl and a pair of Ag-AgCl electrodes submerged in HBSS. For electrical determinations, one calomel and one Ag-AgCl electrode is placed on each side of the monolayer and a pulse of 25 A of current is passed across the monolayer. Using Ohm's Law (V ϭ IR), the resultant voltage deflection and the transepithelial resistance (R) can be calculated. The short circuit current is (I sc ) which represents the total amount of external current that is required to nullify the cell's active ion transport across the epithelium, can also be determined by Ohm's Law.

[ 3 H]Bumetanide Binding
Binding of [ 3 H]bumetanide was carried out as described previously (9). Briefly, [ 3 H]bumetanide (Amersham Corp.) was purified by high performance liquid chromatography using a Microsorb C-18 semipreparative column (Ranin, Woburn, MA) using a protocol developed by Dr. Mark Haas (University of Chicago) to yield a 10 M stock in ethanol with a specific activity of 81 Ci ϫ mmol Ϫ1 . Such purification was found to be essential in order to reduce nonspecific binding. T84 cells were grown on 1-cm 2 permeable filters as described above and incubated overnight in the presence or absence of 33 M phalloidin. Monolayers were incubated at 37°C on an orbital shaker for 30 -40 min in 0.25 M [ 3 H]bumetanide (basolateral) with and without a 200-fold excess of unlabeled bumetanide in a low chloride buffer containing 120 mM sodium gluconate, 25 mM potassium gluconate, 5 mM KCl, and 10 mM Tris-HCl (pH 7.4). Binding was terminated by 4 rapid washes in ice-cold 150 mM NaCl, 10 mM Tris-HCl (pH 7.4) followed by determination of radioactivity and protein content. Saturable binding was represented as the difference between total binding and binding measured in the presence and absence of excess unlabeled bumetanide.

Rb Uptake
Functional activation of the basolateral NKCC cotransporter was assessed as bumetanide-inhibitable 86 Rb uptake, with bumetanide (10 M) serving as a specific inhibitor of cotransport and 86 Rb as a tracer for K ϩ as described previously (8). Briefly, monolayers grown on 1-cm 2 supports were equilibrated in buffer for 20 min and then transferred to wells containing fresh buffer with or without bumetanide (10 M) and with or without forskolin for 5 min. Uptakes were initiated by transferring monolayers to an uptake buffer identical to the preincubation buffer also containing ϳ1.5 Ci/ml 86 Rb. After 3 min, uptakes were terminated by repeated rapid dunking in an ice-cold solution containing 100 mM MgCl 2 and Tris-Cl (pH 7.4). The filters were then sharply excised from the inserts and radioactivity measured by liquid scintillation. The bumetanide sensitive component of uptake, expressed as nmol of K ϩ mg protein Ϫ1 min Ϫ1 , was derived by subtracting individual values of uptake measured in the absence of bumetanide from the mean value of the uptakes measured in the presence of bumetanide. Proteins were determined from representative monolayers using a spectrophotometric bicinchonic acid assay from Pierce (Rockford, IL).

Antibodies Against T84 Na:K:2Cl Cotransporter
A battery of monoclonal antibodies were generated against the T84 cell NKCC cotransporter (12). In this study, we used monoclonal antibodies T9, T4, and T10 that were purified from ascites fluid by a 50% ammonium sulfate precipitation.

Biotinylation and Immunoprecipitation
Surface biotinylation of T84 membranes was carried out as described previously (13). Briefly, T84 cells grown on 5-cm 2 permeable supports were cooled to 4°C and washed with cold HBSS. Apical or basolateral surfaces were selectively biotinylated by adding 0.5 mg/ml biotin sulfo-N-hydroxysuccinimide ester (Pierce) in HBSS to the inner (apical) or outer (basolateral) chamber, and incubated for 20 min at 4°C. The reaction was quenched by incubating monolayers in 50 mM NH 4 Cl in HBSS for 20 min at 4°C followed by several washes with ice-cold HBSS. Monolayers were then briefly washed in Ca 2ϩ /Mg 2ϩ -free HBSS and scraped from filters into 10 mM HEPES (pH 7.4), 3.5 mM MgCl 2 , 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 10 M chymostatin, 10 M leupeptin, 1 M pepstatin, 1 M bestatin. Cells from three monolayers were pooled for each condition. Cells were pelleted by centrifugation at 1000 ϫ g for 10 min. All subsequent steps were carried out at 4°C. Pellets were resuspended in 600 l of the same buffer containing 2% Triton X-100 and rapidly vortexed. The extract (approximately 2.5 ϫ 10 7 cell equivalents/ml) was then centrifuged at 1000 ϫ g for 10 min to remove nuclei and large particulates. The resultant supernatant containing solubilized T84 cell biotinylated membranes was then incubated for 1 h with constant rocking in protein A-Sepharose (Sigma) to preclear the solution of any nonspecific binding. Protein A-Sepharose was pelleted out by centrifugation in a microcentrifuge for 10 s at 10,000 ϫ g. The resultant supernatant was incubated overnight with 2-10 g of T9 monoclonal antibody. Antibody-antigen complexes were isolated from the extract by incubation with protein A-Sepharose for 2.5 h. The antibody-antigen/protein A-Sepharose complex was microcentrifuged and repeatedly washed (5-6 times) with immunoprecipitation wash buffer (10 mM NaH2PO 4 , pH 7.4, 1% Nonidet P-40, 0.4 M NaCl, 2 mM EDTA, 0.1 M sodium fluoride, 1 mM benzamidine, 10 M chymostatin, 10 M leupeptin, 1 M pepstatin, 1 M bestatin). The antibody-antigen complex was separated from protein A-Sepharose by boiling for 3-5 min in reducing sample buffer (2.5% SDS, 5% ␤-mercaptoethanol, 25 mM Tris-HCl, pH 6.8, 20% glycerol, 0.01% bromphenol blue) followed by a 30-s microcentrifugation. The final supernatant containing the biotinylated, immunoprecipitated protein was run on SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose as described below. Nitrocellulose filters were blocked overnight at 4°C in 150 mM NaCl, 5 mM EDTA, 50 mM Tris (pH 7.4), 0.2% Triton X-100, 0.25% gelatin, followed by a 1-h incubation at room temperature in horseradish peroxidase-conjugated avidin (Pierce) diluted 1:10,000 in the same buffer. Nitrocellulose filters were washed 10 ϫ, 10 min each with TTBS (20 mM Tris, pH 7.5, 150 mM NaCl, 0.2% Tween 20, 0.02% NaN 3 ) followed by TBS (TTBS minus Tween 20). Detection of biotinylated proteins was accomplished by incubating nitrocellulose filters in ECL reagent (Amersham, UK) according to the manufacturers instructions for 1 min followed by exposure to Kodak X-ray film. ECL films were scanned using the Molecular Dynamics Scanning Densitometer 300A (Sunnyvale, CA). Band densities were analyzed using Image-Quant Software (Molecular Dynamics, Inc.) and statistical analyses were performed using StatView (Brainpower, Inc. Calabasas, CA).

Agonist Stimulation and Phalloidin Loading
In experiments requiring cAMP agonist stimulation, monolayers were washed three times with HBSS at 37°C and incubated apically and basolaterally for 45 min with 10 M forskolin in HBSS at 37°C. Monolayers were then rapidly placed at 4°C and washed with ice-cold HBSS.
Phalloidin loading was carried out according to the method of Shapiro et al. (7). Briefly, monolayers were incubated at 37°C with 30 M phalloidin (Sigma) in culture media for 12-15 h under sterile conditions. Phalloidin loaded monolayers were maintained at 37°C, washed with warm HBSS, and incubated either with or without 10 M forskolin for 45 min. Monolayers were then placed at 4°C and prepared for surface biotinylation. As a control for toxicity, companion monolayers (0.33 cm 2 inserts) were treated identically and measured electrically for an attenuation of forskolin induced short circuit current with lack of attenuation of carbachol elicited short circuit current serving as described previously (7).

Deglycosylation of the Na/K/2Cl Cotransporter
Triton X-100 solubilized T84 cell extracts (approximately 5 ϫ 10 7 cell equivalents/ml) were incubated at 37°C with 0.1 unit/ml recombinant N-glycanase enzyme (Genzyme Co., Cambridge, MA) for 90 min with constant turning. The reaction was terminated by placing extracts at 0 -4°C followed by immunoprecipitation with the T9 monoclonal antibody. Deglycosylated immunoprecipitated cotransporter was assayed by immunoblot analysis (see below).

Gel Electrophoresis and Western Blotting
Protein samples were electrophoresed on 7.5% polyacrylamide gels (14) according to the method of Laemmli (15). Polyacrylamide gels were either stained (Coomassie Brilliant Blue or silver) or transferred to nitrocellulose filters (100 volts for 2-4 h) according to the method of Towbin et al. (16). Nonspecific protein binding was blocked by incubation overnight at 4°C with 3% fetal calf serum, 1% bovine serum albumin, in TTBS. The filters were probed for 2 h at room temperature or overnight at 4°C with a 1:400 dilution of primary antibody (see below) followed by washing 5 times (10 min/wash) with TTBS and 5 times in TBS. The filters were incubated with peroxidase-conjugated goat anti-mouse IgG (Cappell) at a 1:1000 dilution for 2 h at room temperature. Following secondary antibody incubation, the filters are washed 5 times with TTBS followed by 5 times with TBS and developed by enhanced chemiluminescence (ECL).

Confocal and Electron Microscopy
Confocal Microscopy-T84 cells grown on permeable filters or glass coverslips were rinsed twice in PBS and fixed in methanol at Ϫ20°C for 25 min. Cells were rinsed 3 times for 3 min in PBS containing 0.08% saponin, 0.2% gelatin at room temperature. Fixed monolayers were then incubated in a 1:1000 dilution of the T9 monoclonal antibody for 1 h followed by 3-5 rinses in PBS. Monolayers were incubated at room temperature in a 1:100 dilution of fluorescein isothiocyanate-conjugated goat anti-mouse secondary antibody for 30 min followed by extensive washes in gelatin PBS. Monolayers were mounted in a phenylene diamine/glycerol/PBS medium on glass slides and viewed on a Zeiss Axiophot microscope equipped with epifluorescence and a Bio-Rad MRC 600 confocal imaging system.
Electron Microscopy-Ultrastructural localization of NKCC was performed using the T9 antibody and 10-nm colloidal gold-conjugated secondary antibodies. Filter-mounted monolayers were fixed using the periodate-lysine-paraformaldehyde technique, embedded in LR White, sectioned, and stained with antibody on grids as described previously (17). Sections, mounted on gold grids, were sequentially blocked for 5 min with 0.2% gelatin in PBS and then exposed for 2 h at room temperature to 5% non-fat dry milk in PBS with 0.1% gelatin. Sections were then exposed to primary antibody (diluted 1:10) for 2 h, washed four times with 0.2% Triton X-100 in PBS, and stained for 1 h with goat anti-mouse secondary antibody conjugated to 10-nm colloidal gold as described previously (17). Sections were evaluated at a magnification of 20,000 ϫ for orientation and images of oriented sections were obtained for further quantitation. Morphometry was used to determine: the number of gold particles within ϳ30 nm (3 particle diameters) of the lateral membrane; the area of this membrane associated compartment; the number of gold particles ϳ30 -150 nm from the lateral membrane; the area of this latter peribasal space compartment; the number of gold particles Ͼ150 nm from the lateral membrane, the area of this latter region. Since the distance between an epitope and a gold particle affixed to a secondary antibody is roughly 30 nm (5) the first distance was defined as lateral membrane associated, the second distance as submembranous, and the third as background (all normalized to the respective areas).

Polarization of the NKCC Cotransporter on the T84 Cell
Surface-The functionally defined NKCC cotransport process has been localized to apical membranes in some tissues (18), while to basolateral membranes in others including mammalian intestinal epithelia (19,20). To assess whether the basolaterally defined functional polarity in the intestine reflects specific protein targeting or alternatively, a domain specific activation of a uniformly distributed protein, we used the approaches of laser confocal microscopy, electron microscopy, and selective surface biotinylation. As shown in Fig. 1, immunofluorescence confocal microscopy monolayers revealed that immunolabeling T84 monolayers using either T9 or T4 anti-NKCC antibodies specifically labeled basolateral membranes of T84 cells when grown as polarized monolayers on permeable supports (Fig. 1A). Reconstruction of confocal images to obtain xz axis sections revealed that such staining was restricted to the basolateral domain (Fig. 1B). Immunogold localization (Fig. 2) suggested that the vast majority of NKCC was associated with the lateral membrane. Quantitative analyses of 1,252 particles was performed (see "Materials and Methods"), 81% of specific staining was within the distance from the lateral membrane directly accounted for by the distance between an epitope and a gold particle affixed to a secondary antibody. The number of particles 30 -150 nm removed from the membrane accounted for 19% of specific staining. Thus, the vast majority of specific staining of NKCC by immuno-EM localization was found to already exist on the lateral membrane in the unstimulated state.
Surface distribution of NKCC was analyzed at the protein level using selective apical or basolateral biotinylation (Fig. 3). Monolayers biotinylated apically lacked the presence of surface expressed 195-kDa protein when precipitated with the T9 anticotransporter antibody. In contrast, a biotinylated 195-kDa protein was consistently detected in the immunoprecipitate following selective basolateral biotinylation (Fig. 3, basolateral  lane).
Proteins of ϳ160 and ϳ130-kDa Coprecipitate with the NKCC Cotransporter-The surface biotinylation experiments described above resulted in several bands coprecipitating with the 195-kDa NKCC protein (Figs. 3 and 5). Heavily biotinylated bands migrating between 50 and 70 kDa represent im-munoglobulin heavy chain bands that become biotinylated during immunoprecipitation washes. Biotinylated antibody bands vary in intensity from preparation to preparation due to a nonspecific accumulation of free biotin during handling (data not shown). Two prominent bands of relative molecular mass ϳ160 and ϳ130 kDa (Fig. 3, basolateral lane; Fig. 5, lane 2) are biotinylated by surface labeling and appear to represent membrane proteins with ectodomains containing free amine groups. To determine whether the 160-and 130-kDa proteins contained the epitope recognized by T9, Western blotting with the T9 antibody was performed. As shown in Fig. 4, T9 recognized only the 195-kDa cotransporter and not the 160-and 130-kDa bands. Two additional monoclonal antibodies (T4 and T10) also raised against the 195-kDa cotransporter exclusively recognized the 195-kDa band on Western blots (Fig. 4). Likewise, immunoprecipitation of basolaterally biotinylated cotransporter with T4 also resulted in the appearance of the major coprecipitating 160-and 130-kDa bands (Fig. 5, lane 2) with only a detection of the 195-kDa band by immunoblotting (Fig.  5, lane 3). Several additional minor polypeptides at approximately 90 -100 kDa also appear in the immunoprecipitate. These bands do not appear consistently from preparation to preparation and may represent degradation products of the 160-and/or 130-kDa polypeptides (Fig. 5, lane 2). Distinctive antibodies raised against the 195-kDa cotransporter do not immunocross-react with the 160, 130, or lower molecular weight bands present in the immunoprecipitate. To test whether the 160-and 130-kDa proteins might represent deglycosylated precursors which were simply not recognized by the T9 antibody on Western blots, we deglyosylated T84 extracts with N-glyconase prior to performing immunoprecipitation with the T9 antibody. As shown in Fig. 6, the deglycosylated core peptide of the 195-kDa cotransporter of T84 cells runs, similar to the deglycosylated cotransporter of the shark (1), at ϳ130 kDa. This deglycosylation product, unlike the coprecipitating 160-and 130-kDa proteins, is readily blotted by the T9 antibody. In aggregate, the above data indicate that NKCC immunoprecipitates as a complex with several other membrane proteins including two major proteins of 160 and 130 kDa which do not appear to be precursor forms of the cotransporter. There is no detection of NKCC present on the apical membrane. Basolateral, biotin-avidin blot depicting immunoprecipitated protein from monolayers biotinylated on the basolateral surface. The 195-kDa NKCC band is selectively expressed on the basolateral membrane (arrow). In addition, several polypeptides coprecipitate with the 195-kDa NKCC protein, including two heavily biotinylated polypeptides of ϳ160 and ϳ130 kDa (arrowheads). The 160/130-kDa coprecipitating proteins are predominately expressed on the basolateral membrane. Immunoglobulin bands also detectable on the avidin blot since they become differentially biotinylated during the biotinylation procedure. Molecular weight markers were, myosin (205 kDa), ␤-galactosidase (116 kDa), phosphorylase b (97.5 kDa), and ovalbumin (45 kDa). Fig. 7, forskolin stimulation of T84 monolayers results in rapid induction of Cl Ϫ secretion as well as activation of the cotransporter as measured by bumetanide-sensitive 86 Rb uptake. These events, as well as the associated remodeling of basolateral F-actin microfilaments (7) are prevented by phalloidin loading (Fig. 7). Under these conditions there is no evidence of cellular toxicity and in fact other regulated ion transport events are unaffected (Fig. 7B) (7). We first examined whether forskolin-stimulated activation of the cotransporter resulted in increased specific binding of a radiolabeled drug, bumetanide, which binds specifically to the 195-kDa cotransporter. In data not shown, [ 3 H]bumetanide binding was found to be near-saturated at 0.25-0.5 M, consistent with tracheal epithelial cells (21) and HT 29 cells (9). Forskolin did not substantially increase specific binding or alter the apparent affinity for forskolin in preliminary experiments. This was also found to be true in binding studies performed using normal (135 mM) rather than low chloride (15 mM) buffer (data not shown). Thus, activation of cotransport by cAMP in T84 monolayers appears to occur largely by enhanced ion translocation per bumetanide-binding site rather than by an increase in the total number of binding sites. As shown in Fig. 7D, forskolin and/or phalloidin loading did not detectably increase specific binding of bumetanide to FIG. 4. Western blot analysis of NKCC immunoprecipitated with the T9 antibody. Immunoblots were probed with 20 g/ml T9, T4, or T10 antibodies. Lanes were equivalently loaded with immunoprecipitated protein from ϳ2.5 ϫ 10 7 cell equivalents/ml. Each of the three monoclonals recognize the 195-kDa band. Interestingly, the T10 monoclonal demonstrated a lower affinity for the 195-kDa polypeptide compared with T9 and T4. Immunoglobulin heavy chain bands (Antibody) present in the immunoprecipitate extract were readily recognized by the goat anti-mouse secondary antibody. B, absence of nonspecific cellular toxicity of phalloidin loading on T84 monolayers as evidenced by preservation of transepithelial Cl Ϫ secretory response (short-circuit current) to 0.1 mM carbachol, a Ca 2ϩ mediated secretagogue, in monolayers prestimulated with 10 M forskolin (left-side vertical axis). Additionally, transepithelial resistance (TER, right-side vertical axis), a sensitive index of junctional integrity, is not different between control monolayers and monolayers pre-loaded with 33 M phalloidin for 16 h. All n ϭ 3-4. C, NKCC cotransport functionally defined as bumetanide-sensitive K ϩ ( 86 Rb) uptake, using 86 Rb as a tracer for K ϩ and bumetanide (10 M) as a specific inhibitor of NKCC cotransporter activity. Control monolayers (light bars) display a 3-fold increase in NKCC cotransporter activity in response to 10 M forskolin. Both unstimulated and forskolin-stimulated cotransporter activity is markedly attenuated in monolayers pre-loaded with 33 M phalloidin (dark bars) T84 monolayers. Monolayers were treated with or without forskolin (10 M) for 40 min in buffer containing 0.25 M [ 3 H]bumetanide. D, specific binding represents the difference between total binding and binding measured in the presence of 50 M excess unlabeled bumetanide. Data points represent mean Ϯ S.E. for n ϩ 7 monolayers in each group. T84 cells. Selective surface biotinylation experiments (Fig. 8A) similarly suggested that movement of the 195-kDa cotransporter to the basolateral surface was modest at best following forskolin stimulation. Scanning densitometry of gels from seven independent experiments (Fig. 8B) showed that the average increase in density of the 195-kDa biotinylated band in response to forskolin stimulation was ϳ2.2-fold. In contrast, the major 160-and 130-kDa coprecipitating proteins were prominently up-regulated in forskolin-stimulated monolayers. Scanning densitometry of bands obtained from the seven experiments showed mean increases in biotinylation density of the 160-and 130-kDa bands of 5.8-and 6.1-fold, respectively (Fig. 8B), both p Ͻ 0.06 compared with the change in the 195-kDa band. Experiment to experiment variation in the forskolin-induced increase in the 160-and 130-kDa bands was examined in order to assess whether the proportionality of the increments in the 160-and 130-kDa bands relative to the amount of the 195-kDa band was fixed. Examined in this fashion, the correlation coefficient between increments in the 160-versus the 130-kDa bands was 0.86 suggesting that a proportional relationship in the forskolin stimulated increase in surface expression of these proteins exists.

Forskolin-stimulated Monolayers-As shown in
The apparent forskolin-stimulated surface expression of the 160-and 130-kDa proteins which coprecipitate with NKCC was examined in phalloidin loaded cells (Fig. 9), a condition which blocks functionally-defined activation of the cotransporter (see Fig. 7). Scanning densitometry of biotinylated immunoprecipitation experiments indicated approximately a 3-5-fold attenuation of the forskolin-induced up-regulation of the 160-and 130-kDa bands in cells preloaded with phalloidin ( Fig. 9). The slight up-regulation in surface expression of the 195-kDa NKCC band was only marginally affected by phalloidin preloading conditions. DISCUSSION We show that the distribution of the Na:K:2Cl cotransport protein (NKCC) is basolaterally polarized in intestinal epithelia. We have identified two prominent membrane proteins, of molecular mass 160 and 130 kDa, that coimmunoprecipitate with the 195-kDa NKCC protein from T84 extracts. The 160and 130-kDa coprecipitating polypeptides possess extracellular domains accessible to cell surface biotinylation thus indicating that they are transmembrane proteins. Based on Western blot analysis and deglycosylation with N-glycanase, the 160-and 130-kDa polypeptides appear to be distinct from the 195-kDa NKCC protein. Activation of NKCC, elicited by cAMP agonist stimulation, is accompanied by a modest increase in surface expression of the 195-kDa cotransporter protein (and no change in bumetanide binding) but is paralleled by an approximately 6-fold increase in the 160-and 130-kDa coprecipitating proteins. Phalloidin loading which prevents both the cAMPelicited activation of NKCC as well as the basolateral F-actin rearrangement also diminishes the observed forskolin induced increase the 160-and 130-kDa polypeptides.
Basolateral Polarization of NKCC in Intestinal Epithelia-Mammalian respiratory, intestinal, and shark rectal gland epithelia possess a basolaterally polarized bumetanide-sensitive transport process in which movement of sodium, potassium, and chloride is coupled in an approximate 1:1:2 stoichiometry. Conversely, mammalian renal (thick ascending limb) epithelia and flounder intestinal epithelia possess a similar transport activity on the apical membrane (for recent review, see Refs. 19 and 21). The divergence in polarization of the NKCC cotransport process may reflect differences in apical/basolateral targeting due to tissue specific functionality of cotransporter activity. The question arises, however, whether the identical or highly homologous protein representing cotransporter activity is differentially targeted in a tissue-specific fashion, or whether the respective cotransporter protein is not specifically targeted per se but differentially activated within selected membrane domains. Here we find by confocal microscopy, quantitative electron microscopic morphometry, and surface biotinylation that the ϳ195-kDa protein recognized by antibodies raised against the cotransporter protein selectively appears on the basolateral membranes of intestinal epithelia. Such studies suggest selective basolateral targeting of the cotransporter oc- curs in this tissue. How then might the differential targeting between epithelia be explained? The first hint that NKCC activities between tissue types might represent differences in proteins or protein isoforms arose from the recognition that subtle site-determined functional differences exist in cotransporter activity. For example, as recently reviewed by Haas (19,21), the affinity for bumetanide varies 10-fold between basolateral cotransporters of shark rectal gland, mammalian intestinal, and respiratory epithelial as compared to apical cotransporters from mammalian renal epithelia (thick ascending limb and medulla). It is of interest that the basolateral group are those in which the cotransporter participates in transepithelial salt and water secretion whereas the apical cotransporters participate in transepithelial absorption. Recent evidence suggests that different, although closely related, proteins may underlie such functional and membrane targeting specificities in cotransporters expressed in different epithelia. Using probes derived from the cloned shark rectal gland cotransporter, Payne and Forbush (2) have recently cloned a closely related but unique protein from renal cDNA libraries which corresponds to a smaller transcript which is seen in renal, but not intestinal tissues. Gamba et al. (22) have also recently cloned a cDNA which recognizes a similarly sized transcript, encoding a protein with functional characteristics of a Na:K:2Cl cotransporter, and, under high stringency conditions, blots with renal but not intestinal or respiratory tissues. In aggregate such data suggest that distinctive "secretory" and "absorptive" forms of the cotransporter may exist. If so it is likely that such differences include targeting sequences which ultimately determine the surface distribution of cotransporters between various epithelia.
The Intestinal Epithelial NKCC May Associate with Other Membrane Proteins-Immunoprecipitation of NKCC with three different monoclonal antibodies raised to this protein leads to coprecipitation of several polypeptides including two heavily biotinylated proteins of ϳ160 and 130 kDa. Interestingly, the intensity profile of the biotin signal of the 195-, 160-, and 130-kDa proteins identified the 130 kDa as the most heavily biotinylated, the 160 kDa less heavily biotinylated and the 195 kDa having the least accumulation of biotin signal. Since the biotin reaction involves a covalent coupling of an N-hydroxysuccinimide ester of biotin to free amine groups in the protein, this observation in biotin signal reflects differences in asparagine, lysine, and arginine amino acid residues present in the primary structure of the ectodomains of these proteins. This analysis is substantiated by the recent cloning of T84 cell NKCC which reports few such sites available for biotinylation. 2 Deglycosylation experiments demonstrated that the antibodies employed recognized epitopes which did not depend on sugar residues and that the antibodies blotted exclusively the 195-kDa cotransporter, not the coprecipitating proteins. We have also recently found that the 160-and 130-kDa polypeptides coprecipitate as a complex with the cotransporter in another intestinal epithelial cell line (HT29 subclone, Cl.19A). 3 Such observations raise the question as to whether membrane transport proteins, which in expression systems are shown to be clearly capable of exhibiting the major functions of a characterized transport process, might be additionally influenced by other associated membrane proteins. Recently, an example of a membrane protein which can influence the function of a well characterized transporter has been described. The transport protein responsible for Na-glucose cotransport in the brushborder of intestinal and renal epithelial cells has been well characterized and the expression of this protein in oocyte assays confers the transport characteristics of this SGLT transporter. Recently, however, Veyhl et al. (23) have cloned an unrelated 56-kDa protein (RS-1) from mammalian renal tissue which significantly modifies the activity of the SGLT-1 mediated Na-glucose cotransport when co-expressed. Of interest, the ability of RS-1 to enhance SGLT-1 "activity" is exceedingly sensitive to the stoichiometry of SGLT-1:RS-1 expression. Based on these and supplementary observations, it was proposed that RS-1 is a subunit of a complex in which the dominant function is defined by SGLT-1, however, the accompanying accessory proteins (RS-1) serve to fine tune the activity. Thus it is possible that individual transport proteins, whose function in expression assays satisfies stoichiometry of ion transport events and inhibitor recognition (such as has been shown for the Na:K:2Cl cotransporter; reviewed extensively in Ref. 19, might, when naturally expressed, display activities which are modified by associated proteins as well as by direct events such as phosphorylation or insertion of the primary protein into the membrane. Up-regulation of NKCC Function and Expression of 160-and 130-kDa Proteins Are Dependent on Reorganization of the Microfilament Cytoskeleton-In the presence of cAMP agonist stimulation, there is an up-regulation of basolateral NKCC activity, a process attenuated by stabilization of the actin cytoskeleton ( Fig. 6) (7,8). We have observed under forskolin stimulated conditions a substantial (ϳ6-fold) increase in coprecipitation of surface expressed 160-and 130-kDa polypeptides with only a slight increase in surface expression of the 195-kDa polypeptide. [ 3 H]Bumetanide binding studies also revealed no increase in bumetanide-binding sites upon forskolin stimulation, consistent with the view that surface expression of the cotransporter is not greatly altered in response to this agonist. These observations raise the possibility that regulation of NKCC may circuitously involve action on 195-kDa cotransporter protein by other proteins, such as the 160-and 130-kDa proteins or the cytoskeleton or both. It is thus possible, although speculative, that the mode of up-regulation of NKCC in T84 cells may involve recruitment of the 160-and 130-kDa polypeptides into the cotransporter complex, an event demonstrated to be blocked in the presence of phalloidin (see Fig. 9). Taken in concert, these data suggest an involvement of the actin cytoskeleton both in regulating functional cotransporter activity as well as the surface expression of the coprecipitating 160-and 130-kDa polypeptides. The causal link between the 160/130-kDa polypeptides and cotransporter activation is currently obscure. Further identification and analysis of the 160and 130-kDa polypeptides may provide valuable information on the events which occur during cAMP induced cotransporter activation.