Identification of a Pool of Non-pumping Na/K-ATPase*

Recent studies have ascribed many non-pumping functions to the Na/K-ATPase. Here, we present experimental evidence demonstrating that over half of the plasma membrane Na/K-ATPase in LLC-PK1 cells is performing cellular functions other than ion pumping. This “non-pumping” pool of Na/K-ATPase, like the pumping pump, binds ouabain. Depletion of either cholesterol or caveolin-1 moves some of the “non-pumping” Na/K-ATPase into the pumping pool. Graded knock-down of the α1 subunit of the Na/K-ATPase eventually results in loss of this “non-pumping” pool while preserving the pumping pool. Our prior studies indicate that a loss of the non-pumping pool is associated with a loss of receptor function as evidenced by the failure of ouabain administration to induce the activation of Src and/or ERK. Therefore, our new findings suggest that a substantial amount of surface-expressed Na/K-ATPase, at least in some types of cells, may function as non-canonical ouabain-binding receptors.

the interaction between the Na/K-ATPase and Src keeps Src in an inactive state. Binding of ouabain to the Na/K-ATPase/Src complex changes the interaction, resulting in activation of the associated Src (4). In cultured cells, the receptor Na/K-ATPase resides with Src in caveolae (11). Once the Na/K-ATPase/Src complex is activated by ouabain, it stimulates tyrosine phosphorylation of downstream effectors, which in turn recruits additional kinases and adaptor proteins to the activated receptors, resulting in the activation of protein kinase cascades and the generation of second messengers (4).
To further prove the above Na/K-ATPase-specific signaling mechanism, we have generated several ␣1 knock-down cell lines using a siRNA-based approach. These cell lines (e.g. A4-11 and PY-17) were cloned from LLC-PK1 cells that were transfected with an ␣1-specific siRNA-expressing vector (12). While A4-11 cells express about 44% of ␣1 subunit in comparison to the control P-11 cells, PY-17 cells contain only 8% of the ␣1 subunit. Functional studies indicated that graded knock-down of ␣1 decreased the interaction between the Na/K-ATPase and Src, and resulted in an increase in basal Src activity. Concomitantly, it also abolished ouabain-induced activation of Src and the downstream protein kinase cascade (12). These findings clearly demonstrate that there is a pool of Src-interacting Na/K-ATPase, thus providing further support to the above newly discovered cellular signaling mechanism. They also suggest that there may be two functionally separable pools of Na/K-ATPase in the plasma membrane, one performing the canonical (pumping) function, the other carrying out non-canonical functions such as signal transduction and scaffolding. To test this hypothesis, we have further characterized the ionpumping activity of Na/K-ATPase in the established ␣1 knockdown cells. The data presented here clearly demonstrate the existence of a pool of non-pumping Na/K-ATPase in the plasma membrane.

EXPERIMENTAL PROCEDURES
Materials-Image-iT FX signal enhancer, Antifade kit, Alexa Fluor 488-conjugated anti-mouse/rabbit IgG and Alexa Fluor 546-conjugated anti-rabbit IgG antibodies were obtained from Molecular Probes (Eugene, OR). Anti-Na/K-ATPase ␤1 (clone C464.8) antibody was from Upstate (Lake Placid, NY); anti-Giantin antibody was from Covance (Berkeley, CA). The monoclonal anti-␣1 antibody (␣6F) was obtained from the Developmental Studies Hybridoma Bank at the University of Iowa. All the secondary horseradish peroxidase-conjugated antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Radioactive 86 Rb ϩ and [ 3 H]ouabain were from PerkinElmer Life Science Products (Boston, MA).
Cell Culture-The ␣1 knock-down stable cell lines were generated from LLC-PK1 cells and cultured in Dulbecco's modification of Eagle's medium (DMEM) 3 as described (12). Most of the experiments were performed after the cells reached 90% confluence. To test whether polarity affects the conclusion of our studies, cells were also cultured on Transwell inserts and used when the monolayer was formed (13). Because serumstarved cells were used for our prior studies of non-pumping function of the Na/K-ATPase (12), to be consistent, cells were serum-starved for 12 h and then used for studies of pumping functions.
Cell Surface Biotinylation, Streptavidin Precipitation, and Immunoblot-Labeling with biotin of plasmalemma proteins was performed with modifications according to the protocol described by Gottardi et al. (14). Specifically, cells grown on 60-mm Petri dishes were washed once with a nominally Ca 2ϩfree DMEM (15), and placed on ice. Afterward, the cells were rinsed twice with ice-cold PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.4) and incubated with 2 ml of NHS-ss-biotin freshly diluted into biotinylation buffer (10 mM triethanolamine, pH 9.0, 150 mM NaCl) for 25 min at 4°C with very gentle horizontal motion to ensure mixing. Cells were rinsed twice with PBS containing 100 mM glycine, and washed in this buffer for 20 min at 4°C to make certain that all of the un-reacted biotin was quenched. Cells were then rinsed twice with PBS to wash away the quenched biotin, and solubilized in 900 l of lysis buffer (1% Triton X-100, 150 mM NaCl, 5 mM EDTA, 50 mM Tris, pH 7.5) for 60 min on ice. Cell lysates were clarified by centrifugation at 14,000 ϫ g for 10 min at 4°C. Control experiments showed that nearly all of the cellular Na/K-ATPase was solubilized, which is consistent with prior observations (16). To label the cells grown on Transwell inserts, the same protocol was followed after the cells were first incubated in PBS solution containing 1 mM EDTA for 5 min (14). Biotinylation buffer was then added to the apical side of the Transwell. According to Gottardi et al. (14), pretreatment of polarized cell cultures with PBS-EDTA solution completely opens the tight junction for the apically added biotinylation buffer to access the basolateral Na/K-ATPase (14). The cleared cell lysate (750 l) was collected and incubated overnight with 100 l of packed streptavidin-agarose beads at 4°C with endover-end rotation. The supernatant was then collected, and 1:10 of the total volume was loaded on the polyacrylamide gel as unbound fraction (U), together with 1:10 of the cleared lysate as total (T). Protein signal was detected using an ECL kit and quantified by Bio-Rad GS-670 imaging densitometer.
Confocal Fluorescence Microscopy-The subcellular localization of Na/K-ATPase ␤1 subunit and Giantin was analyzed by immunofluorescence staining. Cells cultured on cover slips were washed twice with PBS and fixed with methanol prechilled at Ϫ20°C. The fixed cells were then rinsed with PBS for three times and blocked with 200 l of Image-iT FX signal enhancer for 30 min at room temperature. Cells were then incubated with primary antibodies (anti-Na/K-ATPase ␤1 and anti-Giantin) in PBS containing 1% bovine serum albumin for 1 h, followed by incubation with secondary Alexa Fluor-conjugated antibodies. Image visualization was performed by Leica TCS-SP2 laser scanning microscope (Leica, Mannheim, Germany).
Ouabain-sensitive 86 Rb ϩ Uptake-Ouabain-sensitive 86 Rb ϩ uptake was employed to measure the transport function of Na/K-ATPase (17). Cells were cultured in 12-well plates to 90% confluence, washed, and serum-starved overnight. After washing once with the fresh medium, the cells were incubated in the same medium at 37°C for 10 min in the presence of different concentrations of ouabain. Monensin (20 M) was added to clamp intracellular Na ϩ to ensure maximal capacity of active uptake (18). Then, 86 Rb ϩ as the tracer for K ϩ (1 Ci/well) was added to start the uptake experiment. After 10 min, 3 ml of ice-cold 100 mM MgCl 2 was added to stop the 86 Rb ϩ uptake. Cells were then washed three times with the same solution. Trichloroacetic acid-soluble 86 Rb ϩ was measured by a liquid scintillation counter. Trichloroacetic acid-precipitated protein was dissolved in 0.1 N NaOH, 0.2% SDS solution for protein assay. Preliminary experiments showed that the uptake was a linear function of time for up to 20 min. Ouabain-sensitive 86 Rb ϩ uptake was about 85% of the total in control P-11 and LLC-PK1 cells, and about 70% in PY-17 cells. To test whether tight junction affects the measurement of 86 Rb ϩ uptake in different cell lines, cells were incubated in nominally Ca 2ϩ -free DMEM medium containing 0.1 mM EGTA (15). Afterward, the 86 Rb ϩ uptake was conducted in this Ca 2ϩ -free medium as described in the above paragraphs. Prior studies have demonstrated that removal of extracellular Ca 2ϩ was sufficient to open up tight junction (14,19,20). For experiments using cells cultured on Transwell inserts, 86 Rb ϩ solution was added to both apical and basolateral sides to measure the total pumping activity. Control experiments showed that the basolateral activity accounted for about 93% of total ouabain-sensitive 86 Rb ϩ uptake. To measure Na/K/2Cl cotransporter activity, bumetanide-sensitive 86 Rb ϩ uptake was conducted as previously described (21). This activity in P-11 and PY-17 cells accounted for less than 10 and 16% of the total uptake activity, respectively.
[ 3 H]Ouabain Binding-The number of ouabain binding sites in cultured cells was estimated by the protocol that had been previously used (with modifications) and discussed (22)(23)(24). Briefly, cells were seeded into 12-well plates, cultured to 90% confluence, and then serum-starved overnight. Afterward, cells were rinsed, incubated in K ϩ -free Kreb's solution (NaCl 142.4 mM; CaCl 2 2.8 mM; NaH 2 PO 4 0.6 mM; MgSO 4 1.2 mM; dextrose 10 mM; Tris 15 mM; pH 7.4) in the presence of 20 M monensin for 10 min, and then exposed to 2 M [ 3 H]ouabain for 15 min at 37°C. Monensin was added to clamp intracellular Na ϩ and to prevent recycling of the Na/K-ATPase from the intracellular pool to the plasma membrane (25,26). Control experiments showed that maximal binding was reached at 2 M ouabain after 15 min of exposure in both P-11 and PY-17 cells. At the end of incubation, cells were washed four times with ice-cold K ϩ -free Kreb's solution, solubilized in 0.1 N NaOH/0.2% SDS, and measured by a scintillation counter. Nonspecific binding, measured in the presence of 5 mM unlabeled ouabain, was less than 3% of total binding, and was subtracted from the total binding. Control experiments showed that PY-17 cells contained 0.21 Ϯ 0.03 ng of protein/cell whereas P-11 had 0.19 Ϯ 0.02 ng of protein/cell. The sizes of these cells are also similar: 15.0 Ϯ 2.6 m in diameter for P-11 cells and 16.0 Ϯ 2.9 m for PY-17 cells. Thus, the binding data were normalized by the cell numbers counted in paralleled dishes.
To test whether tight junction affects the measurement of ouabain binding, the ouabain binding assay was performed in a nominally Ca 2ϩ -and K ϩ -free Krebs solution in the presence of 0.1 mM EGTA as described above. The ouabain binding sites in P-11 cells under this condition were 8 ϫ 10 5 /cell, which is comparable to that (8.5 ϫ 10 5 /cell) in Ca 2ϩ -containing solution. For binding experiments in cells cultured on Transwell inserts, [ 3 H]ouabain solution was added to both apical and basolateral sides to measure the total binding. The binding activity was normalized per mg total cellular proteins (3.6 ϫ 10 12 /mg of protein in P-11 cells).
[Na ϩ ] i Measurement-The relative intracellular Na ϩ content was measured according to Kim et al. (27) with minor modifications (28). Briefly, cells cultured in 12-well plates were incubated at 37°C in DMEM containing trace amounts of 22 Na ϩ (0.5 Ci/ml) for 60 min. Control experiments showed that 60 min of incubation was sufficient to fully equilibrate the exchangeable intracellular Na ϩ with 22 Na ϩ in both control and ␣1 knock-down cells. Cells were then washed three times with ice-cold 100 mM MgCl 2 , and precipitated by trichloroacetic acid. The trichloroacetic acid-extractable 22 Na ϩ was measured by a scintillation counter and the data were normalized by the cell number.
[Na ϩ ] i Measurement with Fluoroprobe Sodium-binding Benzofuran Isophthalate (SBFI)-Intracellular Na ϩ concentration was measured as previously described (15). Briefly, P-11 cells and PY-17 cells were loaded with 5 M SBFI-AM in the presence of 0.075% pluronic for 30 min at 37°C and then washed to remove the excess fluorescent dye. The cells were alternately illuminated by 340 nm and 380 nm UV light and the emission light at 520 nm was recorded. The emission ratio of 340/380 was used to calculate intracellular Na ϩ concentration based on the calibration curve constructed by varying extracellular Na ϩ in the presence of 10 M gramicidin. The basal [Na ϩ ] i concentration in P-11 cells was 20 Ϯ 2 mM (n ϭ 15).
Data Analysis-Data presented are mean Ϯ S.E. of at least three independent experiments, and statistical analysis was performed using the Student's t test.

RESULTS AND DISCUSSION
Pumping and Non-pumping Pools of Na/K-ATPase-We have recently demonstrated that knock-down of the cellular Na/K-ATPase reduces the size of a pool of Src-interacting Na/K-ATPase, resulting in an increase in basal Src activity and an inhibition of ouabain-induced Src activation (12). These new findings led us to propose that there may be two functionally separable pools of Na/K-ATPase in cultured LLC-PK1 cells, one being the canonical ion pumping pool and the other being the non-pumping pool. Because cells need the pumping Na/K-ATPase to survive, we reasoned that knock-down of the cellular ␣1 subunit might preferentially deplete the non-pumping pool of Na/K-ATPase to preserve the pumping pool if these two different pools of Na/K-ATPase co-exist in the plasma membrane. Indeed, when total Na/K-ATPase activity was measured in the control and the ␣1 knock-down cells, we observed that while PY-17 cells expressed only 8% of the ␣1 protein, they exhibited 21% of ouabain-sensitive ATPase activity in comparison to the control P-11 cells (12). These data support the notion that there is a pool of non-pumping Na/K-ATPase in the control P-11 cells. They also indicate that depletion of the ␣1 preferentially reduces the size of this non-pumping pool, resulting in a mismatch between the ouabain-sensitive ATPase activity and the ␣1 protein amount in the knock-down PY-17 cells.
Because the ATPase activity was measured in the presence of alamethicin, the above assay determined the total Na/K-ATPase activity including the enzyme in the plasma membrane and other intracellular compartments (29), which could overestimate the actual pumping pool of Na/K-ATPase in the plasma membrane. Thus, to further assess how ␣1 knock-down affects the size of the pumping pool of Na/K-ATPase in the plasma membrane, we measured the ouabain-sensitive 86 Rb ϩ uptake. Control experiments showed that ouabain-sensitive 86 Rb ϩ uptake accounted for more than 85 and 70% of total uptake in the control and PY-17 cells, respectively. As shown in Table 1, the reduction of the ␣1 subunit content caused a roughly proportional decrease of the maximum pumping activity in A4-11 cells when compared with that of control P-11 cells (46 Ϯ 2% of control). However, PY-17 cells that express 8% of the ␣1 subunit still exhibited about 42% of pumping activity. These findings provide further support to the notion that the control P-11 cells contain a large pool of non-pumping Na/K-ATPase.
It is important to note that the above ouabain-sensitive 86 Rb ϩ uptakes were done in the presence of 20 M monensin. Monensin increases the intracellular Na ϩ to a degree sufficient to maximally stimulate Na/K-ATPase activity (18). In the absence of monensin, PY-17 cells exhibited about 66% of pump activity compared with P-11 cells (Table 2). When intracellular Na ϩ was measured by 22 Na ϩ equilibrium assay, we observed that the ␣1 knock-down significantly increased intracellular Na ϩ per cell. For example, intracellular Na ϩ in PY-17 cells was elevated to about 194% of that in P-11 cells (Table 2). To confirm these observations, we also measured intracellular Na ϩ concentration using a fluorescence probe SBFI in both P-11 and PY-17 cells, and observed a similar relative change as measured by 22 Na ϩ equilibrium assay (Table 2). Importantly, while addition of monensin caused an 86% increase in ouabain-sensitive 86 Rb ϩ uptake because of the increased turnover rate (or reserved pump capacity) in P-11 cells (18), it failed to do so in PY-17 cells (Table 2). These data indicate that the pumps in PY-17 cells operated at the maximal pumping capacity, which is consistent with the fact that the Na ϩ content was almost doubled in these cells.

Non-pumping Na/K-ATPase
Because LLC-PK1 cells can form tight junction when they are cultured at confluent density on Transwell porous inserts, it is prudent to rule out the possibility that the potential difference in tight junction formation among different cell lines may cause the observed mismatch between the pumping activity and the amount of Na/K-ATPase. Because the formation of tight junction requires the presence of extracellular Ca 2ϩ (14,19), we repeated the above functional assays after the cultures were incubated in a nominally Ca 2ϩ -free DMEM medium. As shown in Table 3, the pumping activity in PY-17 cells was 46% in comparison to that in control P-11 cells, which is essentially the same as that measured in the presence of extracellular Ca 2ϩ .
Because the above studies were done on cells cultured in Petri dishes, to further test whether fully polarized LLC-PK1 cells also contain a large pool of non-pumping Na/K-ATPase, we measured the pumping activity in cells cultured at confluent density on Transwell inserts. As shown in Table 3, PY-17 cells had about 32% of pumping activity in comparison to the control P-11 cells. Taken together, our findings clearly show that there is a large pool of non-pumping Na/K-ATPase in LLC-PK1 cells, cultured either in Petri dishes or on Transwell inserts.
To test whether the ␣1 knock-down changes other K ϩ uptake mechanism, we measured Na/K/2Cl cotransporter mediated-86 Rb ϩ uptake, and observed that bumetanide-sensitive uptake activity in PY-17 cells was comparable to that in P-11 cells (91 Ϯ 3% of PY-17 cells, n ϭ 5, p Ͼ 0.05). However, since intracellular Na ϩ was much higher in PY-17 cells, these data suggest that ␣1 knock-down probably up-regulated Na/K/ 2Cl cotransporter to maintain the similar transporter activity as in control P-11 cells. Because PY-17 cells exhibited a similar cell size as control P-11 cells, the above findings also suggest that ␣1 knock-down cells are likely adapted to the decreased Na/K-ATPase activity.
Surface Distribution of Na/K-ATPase-The above data indicate that there may be a large pool of non-pumping Na/K-ATPase in the control P-11 cells. The formation of a functional pumping Na/K-ATPase requires the assembly of the ␣␤ complex and the subsequent delivery of this complex to the plasma membrane. Although it is less likely, if most Na/K-ATPase in P-11 cells resided in cellular compartments other than the plasma membrane, this could account for the observed nonpumping pool. Thus, to rule this out, we performed the following two sets of experiments.
Because the ␤ subunit is essential for targeting Na/K-ATPase to the plasma membrane (30), in the first set of experiments we determined the effects of ␣1 knock-down on the expression and cellular distribution of the ␤1 subunit. As depicted in Fig. 1A, control P-11 cells presented at least three glycosylated ␤1 bands. The expression of these glycosylated ␤1 was significantly reduced in both A4-11 and PY-17 cells, and the reduction was correlated to the degrees of the ␣1 knock-down (Fig. 1A). When the density of the ␤1 band was normalized against all three glycosylated ␤1 bands in P-11 cells, we found that the ␤1 in PY-17 cells was about 25% of that in P-11 cells. This was further confirmed by Western blot analysis of deglycosylated ␤1 prepared from these cell lysates (data not shown). Interestingly, knock-down of the ␣1 caused almost complete depletion of the two ␤1 bands that are highly glycosylated. As depicted in Fig.  1A, under normal exposure conditions, these two species were   barely detectable in PY-17 cells. However, when the film was overexposed, we were able to detect the expression of these two species. The quantitative data indicated that PY-17 cells expressed less than 5% of these two species in comparison to P-11 cells. These findings were consistent with the immunostaining results depicted in Fig. 1B. While the majority of the Na/K-ATPase ␤1 resided in the plasma membrane in control P-11 cells, it was largely retained in the intracellular compartments in TCN23-19 cells, another ␣1 knock-down cell line that behaves like PY-17 (12). In addition, co-immunostaining indicated that a portion of the retained Na/K-ATPase ␤1 in TCN23-19 cells was clearly co-localized with the Golgi marker protein, giantin. Taken together, the data showed that knockdown of the ␣1 caused a significant reduction of the plasma membrane ␤1.
To determine surface distribution of the Na/K-ATPase and to calculate how much surface Na/K-ATPase is required for cellular pumping activity, we performed surface biotinylation assay. As depicted in Fig. 2A, about 30% of ␣1 in the control P-11 cells remained in the unbound fraction, indicating that the other 70% was accessible to biotinylation, thus distributed in the plasma membrane. Using the same protocol, we detected about 19% ␣1 in the unbound fraction in A4-11 cells. However, we failed to detect any ␣1 subunit in the unbound fraction in PY-17 cells even when the amount of cell lysate used for the biotinylation assay was doubled (Fig. 2A). These data suggest that most of the ␣1 in PY-17 cells must be targeted to the plasma membrane. Thus, if we assume that all of the ␣1 (8% of control) in PY-17 cells resided in the plasma membrane and functioned as ion pumps, about 27% [Y ϭ (8 ϫ 100/70)/42 ϭ 0.27] of the surface ␣1 in control P-11 cells would be required for maintaining the maximal pumping activity. Apparently, knock-down of the ␣1 initially reduces both pumping and nonpumping pools as demonstrated in A4-11 cells. Once the maximal pumping activity is reduced to half, further reduction in ␣1 causes a preferential depletion of the non-pumping pool of Na/K-ATPase, which is consistent with our prior observations (12).
To further confirm this, we repeated the same measurements in cells cultured on Transwell inserts. Interestingly, we observed that only 13% of ␣1 in control P-11 cells was in the unbound fraction (Fig. 2B), indicating that the other 87% was distributed in the plasma membrane. As in cells cultured in Petri dishes, we failed to detect any ␣1 in the unbound fraction in PY-17 cells (data not shown). Because high density cell cultures were used, we re-measured the relative amount of ␣1 in both control and PY-17 cells. As shown in Fig. 2C, the ␣1 in PY-17 cells was about 11 Ϯ 3% of that in P-11 cells, which is similar to that measured in cells cultured in Petri dishes (Table  1). Thus, if we assume that all of the ␣1 (11% of control) in PY-17 cells resided in the plasma membrane, about 40% of surface Na/K-ATPase in control P-11 cells would be involved in pumping. This calculation suggests that a higher percentage of surface Na/K-ATPase may pump in the fully polarized cells than that in cells cultured in Petri dishes. Although it is unlikely, we cannot exclude the possibility that the activity assay conducted in cells cultured in Petri dishes might underestimate the pumping activity caused by impaired substrate accessibility.
Ouabain Binding and Turnover Rates-The above biotinylation data indicate that there is a large pool of non-pumping ␣1 in the plasma membrane of P-11 cells. Although the ␣1 subunit in the plasma membrane of control P-11 cells are likely to be paired with the ␤1 subunit based on the prior studies (31,32), it is prudent to conduct additional experiments to ensure that the non-pumping pool we have detected is constituted by the Na/K-ATPase (i.e. the ␣␤ complex). To do so we measured [ 3 H]ouabain binding in both control and knock-down cells because prior studies have shown that only the ␣␤ complex binds ouabain (33).
As depicted in Table 1, P-11 cells contained about 850,000 ouabain binding sites per cell, which is similar to the reported value in cultured Hela cells (34). However, this number is lower than the value previously reported for LLC-PK1 cells (35). Because serum can increase the expression of Na/K-ATPase in cultured cells (36,37), we believe that 12 h of serum starvation used in our experimental condition might significantly reduce the number of ouabain binding sites. When ouabain binding was measured in PY-17 cells, we found that these cells contained about 115,000 sites per cell, thus 14% of the control P-11 cells. A4-11, on the other hand, expressed about 381,000 sites per cell, 45% of that in control P-11 cells. Quantitatively, with 8% of the ␣1 content, PY-17 cells manifest 14% of the ouabain binding sites and carry out 42% of the ion transport activity. We can analyze this if we make certain assumptions. First, we assume (as supported in Ref. 12) that there is close to zero signaling (non-pumping) pool in PY-17 cells. Second, we assume that the reason that they manifest 14% of the ouabain binding sites with 8% of the ␣1 subunit means that a proportion of the ␣1 is inaccessible to ouabain (e.g. the Na/K-ATPases in recycling vesicles and the unpaired ␣1 in the ER and later endosomes) and thus not contributing to either 86 Rb ϩ uptake or [ 3 H]ouabain binding in P-11 cells, but they are all accessible in PY-17. Thus, 42% of the transport activity can be carried out by 14% of the ouabain binding sites, giving a ratio of 3 arbitrary units of pump activity per site in PY-17 cells. For P-11 cells, the corresponding ratio is obviously 1.0. This indicates that in P-11 cells, the maximum activity per site, even in the presence of monensin where Na ϩ is not rate-limiting, is 1/3 ϭ 33%. This implies that 67% of the ouabain-binding sites in P-11 cells are non-pumping pumps, but may perform cellular functions other than ion pumping. The only alternative interpretation is that all of the sites are pumping, but the turnover number (pumping per site) is increased 3-fold as a result of ␣1 knockdown, which seems very unlikely. Moreover, this estimation is essentially similar to that (73%) calculated based on surface ␣1 biotinylation and the activity data. The similar ratio (46% activity versus 14% ouabain binding site) was also observed in cells cultured in nominally calcium-free DMEM as shown in Table 3. In cells cultured on Transwell inserts, PY-17 cells contained about 14% of ouabain binding sites and had 32% of pumping activity in comparison to the control P-11 cells. Thus, about 44% of pumps in fully polarized P-11 cells are involved in pumping. This is similar to that estimated with biotinylation data. Taken together, both estimations indicate that a large number of surface Na/K-ATPase is likely engaged in cellular activities other than ion pumping.
It is known that purified renal Na/K-ATPase has a turnover rate of 8,000 -10,000 cycles per min for ATP hydrolysis. However, when the same turnover rate was measured in cultured cells or isolated renal tubules, the number was much lower, and ranged from 1,500 to 5,000 cycles per min depending on cell type (22,34,38,39). Using the above ouabain-binding and 86 Rb ϩ uptake data, we calculated turnover rates of the Na/K-ATPase-mediated ATP hydrolysis (two K ϩ per pump cycle per ATP hydrolysis) in the different cell lines. These calculations revealed that the Na/K-ATPase in P-11 cells operated at about 1,600 cycles per min while the pump in PY-17 cells ran at 5,100 cycles per min. These findings are consistent with the reported values in the literature, and with the notion that these control cells contain a large pool of non-pumping Na/K-ATPase. They also suggest that interaction of the Na/K-ATPase with other cellular proteins is a key to the generation of a pool of nonpumping Na/K-ATPase. Removal of the interacting proteins from the Na/K-ATPase during the purification may disinhibit the enzyme, converting the non-pumping pump into pumping pump.
In short, the above data indicate that a large pool of nonpumping Na/K-ATPase exists in cultured cells. Importantly, this pool of Na/K-ATPase, like the pumping pump, is capable of binding ouabain. If this non-pumping pool is involved in transmitting the extracellular ouabain signal, it is estimated that ouabain at physiological concentration (e.g. 0.1 nM) can activate about 2000 receptor sites per cell in LLC-PK1 cells. This activation should generate enough second messengers to change cellular function, especially when the initial signal can be amplified via the protein kinase cascades.
A Role for Caveolae-We have shown that caveolae contain a pool of Na/K-ATPase that interact with caveolin-1, Src, and other signaling proteins (11,40). Based on the structural information from SERCA and the Na/K-ATPase (41), both A (actuator) and N (nucleotide binding) domains of the ␣1 may undergo a large and rapid conformational change during each pump cycle. Because these domains are involved in interaction with the signaling and structural proteins (41,42), these interactions, in principle, might either slow down or inhibit the movement of these functional domains; thus, the pumping activity of the Na/K-ATPase. Conceptually, this could generate a pool of non-pumping Na/K-ATPase. Of course, not all interacting proteins will inhibit the pumping activity of the Na/K-ATPase. For example, when adducin was added to the purified Na/K-ATPase, it stimulated the ATPase activity by accelerating the E2 to E1 conformational change (43). Nevertheless, by concentrating Na/K-ATPase and its signaling partners, caveolae could facilitate the interactions between the Na/K-ATPase and inhibitory proteins, thus playing a role in the generation of a pool of non-pumping Na/K-ATPase.
We previously demonstrated that the disruption of caveolae structure by either cholesterol depletion or siRNA-mediated knock-down of caveolin-1 could redistribute the Na/K-ATPase from caveolar fraction, and inhibit ouabain-activated signal transduction in LLC-PK1 cells (11). If caveolae are involved in assembly of the non-pumping pool of Na/K-ATPase, we would expect that depletion of cholesterol by methyl-␤-cyclodextrin (M␤CD) should increase Na/K-ATPase-mediated 86 Rb ϩ uptake. Indeed, as illustrated in Fig. 3A, pretreatment of P-11 cells with 10 mM M␤CD resulted in a significant increase (40% over control) in the Na/K-ATPase-mediated 86 Rb ϩ uptake. The same effect was observed when the parent LLC-PK1 cells were exposed to 10 mM M␤CD (data not shown). It is important to note that the M␤CD treatment did not change the maximal ouabain binding sites in these cells (data not shown). Significantly, when the same treatment was applied to the knockdown cells, we found no change in the pumping activity in PY-17 cells. On the other hand, it caused a significant, but much smaller increase in A4-11 cells. These data are consistent with the notion that most of the Na/K-ATPase in PY-17 cells operate as a pump whereas A4-11 cells contain reduced pools of both pumping and non-pumping Na/K-ATPase. To further confirm the above findings, we also measured 86 Rb ϩ uptake and [ 3 H]ouabain binding in both control P-11 cells and caveolin-1 knock-down C2-9 cells that were derived, like P-11, from LLC-PK1 cells (40). As depicted in Fig. 3B, depletion of caveolin-1 also resulted in a comparable increase in the pumping activity per ouabain binding site. Clearly, disruption of caveolae can convert a portion of non-pumping Na/K-ATPase into pumping pumps. It is also clear that caveolae in LLC-PK1 cells only contain less than half of cellular non-pumping Na/K-ATPase assuming that these treatments release all of caveolar Na/K-ATPase from interactions with the inhibitory proteins. Furthermore, these data indicate that the caveolar signaling Na/K-ATPase is capable of pumping ions once the interacting proteins (or factors) are removed by depletion of cholesterol or caveolin-1, which is consistent with the findings that the partially purified caveolar Na/K-ATPase has the normal ouabainsensitive ATPase activity (31).
To test whether cells other than LLC-PK1 contain a pool of non-pumping Na/K-ATPase, we treated cultured fibroblasts derived from the rat heart with M␤CD, and then measured for the Na/K-ATPase-mediated 86 Rb ϩ uptake. We found that the depletion of cholesterol doubled the pumping activity in cultured fibroblasts (232 Ϯ 16% of control, n ϭ 6). These results suggest that an even greater majority of plasma membrane Na/K-ATPase belongs to the non-pumping pool in fibroblasts than that in the cells derived from renal epithelium. This is not surprising as renal epithelial cells express a large number of Na ϩ and K ϩ channels and transporters, thus requiring more "pumping pumps" to keep up with the inward sodium and outward potassium "leaks." Fibroblasts, on the other hand, are far less "leaky." Therefore, their pump population would contain relatively less "pumping pumps." Interestingly, reduction of cellular cholesterol has also been reported to increase the Na/K-ATPase activity in several other types of cells (44,45). For example, depletion of cholesterol by 5-25% could increase the pumping function by 15% in human red blood cells (44).
To assess whether the presence of a large pool of non-pumping Na/K-ATPases affect ouabain-induced inhibition of the pumping activity, we measured the dose-dependent effects of ouabain on 86 Rb ϩ uptake in both P-11 and PY-17 cells in the presence of 20 M monensin. As shown in Fig. 4, we found that depletion of the cellular Na/K-ATPase increased the potency of ouabain on cellular pumping activity. Specifically, while 357 nM ouabain was sufficient to cause 50% inhibition of ouabain-sensitive 86 Rb ϩ uptake in PY-17 cells, the same inhibition in control P-11 cells required 2131 nM ouabain (Fig. 3). Consistently, the apparent K1 ⁄ 2 for A4-11 cells was between P-11 and PY-17 cells. Although we do not know how depletion of the cellular Na/K-ATPase sensitizes ouabain-induced inhibition of the pumping activity, several possibilities are worthy of discussion. First, the existence of a large pool of non-pumping Na/K-ATPase may make the access of ouabain to the pumping pump more restricted in P-11 cells. Because the expression of ␤1 was significantly reduced in PY-17 cells, it could alter the formation of tight junction (46), thus making PY-17 cells much leakier than P-11 cells. However, this appears to be an unlikely explanation because if this is true, we would detect a large difference in ouabain binding, and total pump activity among these cells when the formation of tight junction was disrupted by removal of extracellular Ca 2ϩ (Table 3). Alternatively, the on-rate of ouabain binding might be faster in PY-17 cells because of increased intracellular Na ϩ . However, because the assays were conducted in the presence of monensin, it is less likely that differences in intracellular Na ϩ among these cells can fully account for the observed ouabain sensitivity. Finally, an increase in ouabain sensitivity could be caused by decreased endocytosis of the Na/K-ATPase in PY-17 cells (24,47). It is known that activation of the Na/K-ATPase/Src complex by ouabain can stimulate the endocytosis of the Na/K-ATPase in  P-11 cells (40). Because depletion of the non-pumping pool of Na/K-ATPases blocks ouabain-induced activation of Src (12), it is conceivable that ouabain would be less effective in stimulation of the endocytosis of Na/K-ATPase. Clearly, these issues remain to be resolved experimentally.
In short, the above data indicate that caveolae play an important role in the organization of non-pumping pool of Na/K-ATPase. Moreover, the size of the non-pumping pool may vary depending on the function of different cells as well as culture conditions (Table 3). This is important because the size could determine the mode of ouabain action (Fig. 4). For example, if the size of the non-pumping pool is large enough, the physiological effects of ouabain are most likely mediated by the activation of protein kinase cascades as we demonstrated (4,12) or by stimulation/inhibition of other unknown protein complexes. On the other hand, if the non-pumping pool is small or depleted, the ouabain effect is most likely because of inhibition of the pumping activity as depicted in Fig. 4. Because Na/K-ATPase is very important in the economy of cells, the existence of a pool of non-pumping Na/K-ATPase could also serve as a reservoir for cells to rapidly and appropriately adjust to different conditions.
In summary, the evidence provided in the present study strongly suggests the existence of a large pool of non-pumping Na/K-ATPase in cultured cells. However, it is important to point out an alternative possibility that this so-called nonpumping pool is constituted by a number of slow pumping pumps that are in complex with signaling, structural and other unknown proteins. Apparently, this non-pumping pool of Na/K-ATPase resides in the plasma membrane. Like the pumping pumps, they are capable of binding ouabain. These findings are significant. First, they are consistent with recent studies from many laboratories, showing that the cellular Na/K-ATPase can transmit an extracellular ouabain signal independent of its pumping function. Second, they suggest that over half of plasma membrane Na/K-ATPase, at least in some cell types, is engaged in activities other than ion pumping. It is important to note that these activities include, but are not limited to, the well-characterized signal transducing functions of Na/K-ATPase. The Na/K-ATPase is known to interact with many protein partners. For example, Na/K-ATPase binds phospholemman and agrin, and these interactions can either slow the pump or keep the pump in an inactive state (48 -51). Recent work has also revealed the interaction between the Na/K-ATPase and cofilin. Interestingly, this interaction appears to play a role in the regulation of cellular metabolic activity (52). Third, the new findings bring about many new and important issues regarding the Na/K-ATPase. For example, how are these two pools of Na/K-ATPase assembled, delivered, and disassembled? Furthermore, are the sizes of these pools dynamically regulated? Finally, they call for the engagement of more investigators to unravel many unknown non-canonical functions of the Na/K-ATPase, and delineate the roles of these functions in cell biology and animal physiology.