ATP-dependent Activation of KCa and ROMK-type KATP Channels in Human Submandibular Gland Ductal Cells*

[Ca2+] i and membrane current were measured in human submandibular gland ductal (HSG) cells to determine the regulation of salivary cell function by ATP. 1–10 μm ATP activated internal Ca2+ release, outward Ca2+-dependent K+ channel (KCa), and inward store-operated Ca2+ current (I SOC). The subsequent addition of 100 μm ATP activated an inwardly rectifying K+current, without increasing [Ca2+] i . The K+ current was also stimulated by ATP in cells treated with thapsigargin in a Ca2+-free medium and was blocked by glibenclamide and tolbutamide, but not by charybdotoxin. This suggests the involvement of a Ca2+-independent, sulfonylurea-sensitive K+ channel (KATP). UTP mimicked the low [ATP] effects, while benzoyl-ATP activated internal Ca2+ release, a Ca2+ influx pathway, and KCa. Thus, ATP acts via P2U (P2Y2) and P2Z (P2X7) receptors to increase [Ca2+] i and activate KCa, but not KATP. Importantly, (i) ROMK1 and the cystic fibrosis transmembrane regulator protein (but not SUR1, SUR2A, or SUR2B) and (ii) cAMP-stimulated Cl− and K+ currents were detected in HSG cells. These data demonstrate for the first time that a ROMK-type KATP channel is present in salivary gland duct cells that is regulated by extracellular ATP and possibly by the cystic fibrosis transmembrane regulator. This reveals a potentially novel mechanism for K+ secretion in these cells.

Activation of the P2X subtype has been shown to be coupled to the activation of a nonspecific cation channel(s), resulting in increased permeability of the plasma membrane to Na ϩ , K ϩ , and Ca 2ϩ .
P2 receptors have been found in many epithelial cells, including those from various salivary glands. Four P2 receptors have been identified in salivary gland cells: P2Y 1 , P2Y 2 (P 2U ), P2X 4 , and P2X 7 (P 2Z ) (6). In rat parotid acinar cells, it has been reported that ATP primarily activates a Ca 2ϩ and Na ϩ -permeable cation channel via stimulation of P 2Z receptors. While transcripts for P2Y receptors have been detected in rat parotid acinar cells, ATP does not appear to strongly induce IP 3 generation or internal Ca 2ϩ release (6 -9). However, in rat submandibular ductal cells, evidence for both P 2U and P 2Z receptors have been presented (6, 10 -13). Interestingly, these P2purinergic receptors have distinct cellular localizations in these cells; P 2Z receptors were found in the luminal membrane, while P 2U receptors were detected in basolateral membrane (10 -13). Presently, the exact mechanisms and ion channels involved in P2 receptor-mediated regulation of salivary gland function are poorly understood (6). The human submandibular ductal cell line (HSG) has been widely used to study the mechanism(s) of Ca 2ϩ signaling in salivary gland cells (6, 14 -18). Previous studies with HSG cells have demonstrated that activation of the muscarinic receptor leads to the generation of IP 3 , which causes intracellular Ca 2ϩ release and activation of store-operated Ca 2ϩ influx (15)(16)(17)(18). HSG cells have also been reported to have P2 receptors (6,14). ATP, via the P 2U receptor, was shown to increase [Ca 2ϩ ] i and Ca 2ϩ -activated 86 Rb ϩ efflux (14). It was also reported that in HSG cells ATP-stimulated IP 3 formation is coupled to the P 2U activation, while ATP-stimulated 45 Ca 2ϩ influx is mediated via activation of P2X receptors (19). However, the types of ion channels involved in these ATP-induced ion fluxes have not yet been identified.
This study examines the mechanisms involved in the regulation of salivary gland cell function by ATP, by measuring [Ca 2ϩ ] i and the membrane conductance of HSG cells. The data show that ATP acts via multiple P 2 purinergic receptors, including P 2U (P2Y 2 ) and P 2Z (P2X 7 ), to activate the Ca 2ϩ -activated K ϩ channel (K Ca ), the store-operated Ca 2ϩ influx channel, and probably a Ca 2ϩ -permeable cation channel. Importantly, we report here that ATP also activates an inwardly rectifying, Ca 2ϩ -independent K ϩ current. The inhibition of this current by sulfonylurea compounds and the presence of ROMK1 and cystic fibrosis transmembrane regulator (CFTR) proteins in HSG cells, suggests for the first time that a ROMK type of epithelial K ATP channel is present in salivary gland epithelial cells. As has been suggested for kidney epithe-lial cells (20,21), the putative ROMK channel in HSG cells might also be regulated by a CFTR-dependent mechanism. These data reveal a potentially novel mechanism for the regulation of K ϩ secretion in salivary epithelial cells.

EXPERIMENTAL PROCEDURES
Cell Culture-HSG cells were a kind gift from Dr. Mitsunabo Sato (Tokushima University, Japan). The conditions for cell culture were similar to those described previously (16,17). Briefly, cells were grown in Eagle's minimum essential medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin (all from Biofluids, Rockville, MD) under 5% CO 2 at 37°C. Cells were passaged when confluent by detaching from the tissue culture dish with 0.25% trypsin, 1.0 mM EDTA (Biofluids). A single-cell suspension was reseeded onto coverslips and used after about 24 h.
Patch Clamp Measurements-Patch clamp in a whole cell configuration was performed on single HSG cells attached to coverslips using the standard patch-clamp technique described previously (16,17). The resistance of the pipette was about 3-6 megaohms. The chamber was connected with an Ag-AgCl pellet through a 150 mM NaCl-containing agar bridge. Membrane currents were measured with an Axopatch 200A amplifier in conjunction with pClamp 6.1 software and a Digidata 1200 A/D converter (Axon Instruments, Foster City, CA). The currents were filtered at 2 kHz (low pass bessel filter) and sampled with an interval of 10 ms. In the step protocol, the cell was held at 0 mV for 312 ms and stepped from Ϫ120 to Ϫ80 mV in 20-mV steps for 200 ms each step. The leakage subtraction function was used to minimize the leak currents. The currents were digitized and recorded directly onto the hard drive of a Dell Pentium computer. The I-V relationship was calculated using the peak amplitude of the current during the step protocol and exported to Origin 5 (Microcal Software, Inc., Northampton, MA) for further analysis.
[Ca 2ϩ ] i Measurements-Fluorescence was measured in fura2-loaded single HSG cells at excitation wavelengths of 340 and 380 nm, with emission at 510 nm, using a SLM 8000/DMX 100 spectrofluorimeter attached to an inverted Nikon Diaphot inverted microscope with a Fluor 40ϫ oil immersion objective. Images were acquired using an enhanced CCD camera (CCD-72, MTI) and the Image-1 software (Universal Imaging Corporation, PA). The 340/380-nm ratio of fura2 fluorescence has been used to represent [Ca 2ϩ ] i . Other details of the experiments are given in the text and figure legends. All experiments were performed at room temperature.
Western Blotting-Preparation of crude and plasma membrane fractions, SDS-PAGE, and Western blotting were performed as described previously (22). Reactivity of the samples against either a polyclonal anti-human CFTR antibody, 1:150 dilution (a generous gift from Dr. W. Guggino, Johns Hopkins University, Baltimore, MD) or polyclonal anti-ROMK antibody, 1:200 dilution (Alomone Laboratories, Jerusalem, Israel) was determined using a secondary antibody 1:5000 dilution and ECL reagent (Amersham Life Sciences). Plasma membrane fraction from HEK-293 cells overexpressing the CFTR protein was kindly provided by Dr. Nelson Arispe (Department of Cell Biology, Uniformed Services University of Health Sciences, Bethesda, MD). Control ROMK1 protein was supplied by Alomone Laboratories. Amido Black B was used to stain the proteins after transfer to the polyvinylidene difluoride membrane.

ATP-induced Increase in [Ca 2ϩ ] i and Membrane
Conductance in HSG Cells-ATP, at concentrations ranging from 0.1 to 100 M, was added sequentially to the cell as shown in Fig. 1. An outward current was detected in 95% (43/48) of the HSG cells tested at a holding potential of 0 mV in a whole cell patch clamp mode. Typically, the cells showed a response, i.e. an increase in the outward current, at 1 M ATP with a few cells responding at 0.1 or 10 M. A second response was seen at 100 M ATP. However, a subsequent addition of 1 mM or higher ATP failed to induce any further increase in the membrane current. The current was blocked by replacing intracellular K ϩ with Cs ϩ (data not shown), strongly indicating that it was due to an increase in K ϩ ion conductance. Further, this pattern of response was similar in the presence or absence of external Ca 2ϩ (Fig. 1, compare A and B). However, in the presence of external Ca 2ϩ , the outward current seen at 1 M ATP was more sustained, presumably due to the involvement of a Ca 2ϩ influx component in this response. Most cells showed a quick run down of the membrane conductance with a duration of about 2-3 min at any given ATP concentration. This is consistent with a previous report showing that ATP Fig. 1C. In the absence of external Ca 2ϩ , the initial peak increase in [Ca 2ϩ ] i was not changed, although the response was more transient in nature. This suggests that both intracellular Ca 2ϩ release and Ca 2ϩ influx account for the ATP-induced increase in [Ca 2ϩ ] i. Further, this pattern of [Ca 2ϩ ] i increase was not changed when Na ϩ was removed from the external medium (n ϭ 35, data not shown). However, when Mg 2ϩ was removed from the medium as shown in Fig. 1D, a small increase in [Ca 2ϩ ] i was induced with 100 M ATP in a Ca 2ϩcontaining medium. These data suggest that, at 100 M, ATP induces a Ca 2ϩ influx component that is inhibited by external Mg 2ϩ . This is also consistent with studies in pancreatic duct cells (23) and other salivary gland cells (6), showing that removal of external Mg 2ϩ increased the ATP-induced peak [Ca 2ϩ ] i . However, since the K ϩ current measurements (shown in Fig. 1, A and B) were made in the presence of external Mg 2ϩ , the K ϩ current induced at 100 M ATP is probably not associated with Ca 2ϩ influx.
UTP, a potent agonist of several P2Y receptors, stimulates IP 3 generation and release of Ca 2ϩ from intracellular Ca 2ϩ stores (1-4). Unlike with ATP, sequential additions of increasing [UTP] induced an increase in [Ca 2ϩ ] i and outward current only at 1 M (Fig. 2, A and B). This pattern was not changed in the absence of external Mg 2ϩ or with 10 mM external Mg 2ϩ (n ϭ 43, data not shown). However, prior stimulation of cells with UTP attenuated the [Ca 2ϩ ] i increase induced by a subsequent addition of ATP to the same cell, and vice versa (see Fig. 2, C and D). As seen with UTP, only low [2MeS-ATP] induced a response. However, the response was lower than that induced by ATP or UTP (data not shown).
We have recently reported that depletion of intracellular Ca 2ϩ stores in HSG cells by muscarinic receptor stimulation with CCh, introduction of IP 3 into the cell, or treatment with the Ca 2ϩ pump inhibitor Tg, activated I SOC, which was dependent on external [Ca 2ϩ ] and inhibited by La 3ϩ and Gd 3ϩ (17).
The data in Fig. 3 show that ATP (1 M) also induced an inward Ca 2ϩ current in HSG cells. This inward current was dependent on external [Ca 2ϩ ] and was eliminated by changing the external [Ca 2ϩ ] from 10 to 0 mM (Fig. 3A). ATP-induced I SOC had a larger amplitude (25 Ϯ 7 pA), but a shorter duration (3-4 min), compared with that induced by IP 3 or CCh (17) and was inhibited by La 3ϩ (Fig. 3B). Notably, ATP (1 M)-induced inward Ca 2ϩ and outward K ϩ currents displayed similar characteristics i.e. dependence on external Ca 2ϩ and inhibition by La 3ϩ (Fig. 3, compare A and B with C and D). Further, stimulation of cells with ATP in a Ca 2ϩ -free medium inhibited the response to a subsequent addition of CCh (data not shown). This is the first report of ATP-induced I soc in a salivary cell. Taken together, these data demonstrate that in HSG cells low [ATP] (Ͻ10 M) stimulates P 2U (P2Y 2 ) receptors to cause the release of Ca 2ϩ from internal Ca 2ϩ stores, which results in the activation of K Ca and I SOC .
ATP-induced Ca 2ϩ -independent K ϩ Current- Fig. 1 shows that high [ATP] activates a K ϩ current in HSG cells that is not accompanied by any changes [Ca 2ϩ ] i . To determine the nature of this K ϩ current, we examined the effect of depletion of the intracellular Ca 2ϩ store with Tg. [Ca 2ϩ ] i was monitored in the absence of external Ca 2ϩ , and data are shown in Fig. 4A. Tg induced an increase in [Ca 2ϩ ] i due to Ca 2ϩ "leakage" from the internal stores. The subsequent addition of ATP (10 -100 M) to the cells failed to induce any further increase in [Ca 2ϩ ] i (Fig.  4A). This suggests that the [Ca 2ϩ ] i increase induced by ATP is due to Ca 2ϩ release from Tg-sensitive stores. Similarly Tginduced increases in [Ca 2ϩ ] i or the K Ca current were diminished or greatly reduced after the cell was first exposed to ATP (n ϭ 56 or n ϭ 4 for [Ca 2ϩ ] i and the K Ca currents, respectively; data not shown). In contrast to [Ca 2ϩ ] i increases shown in Fig.  4A, 100 M ATP induced an increase in the outward current in Tg-treated cells (Fig. 4B). When external Ca 2ϩ was reintroduced, a further increase in the outward current was seen, probably due to the activation of K Ca as a result of Ca 2ϩ entry. These results clearly indicate that two kinds of outward K ϩ currents are stimulated by ATP in HSG cells. One is associated with [Ca 2ϩ ] i increase, and the other appears to be independent of [Ca 2ϩ ] i increase. Consistent with the [Ca 2ϩ ] i measurements shown in Fig. 4A, Tg induced a transient activation of the K ϩ current. In aggregate, the data presented above indicate that in HSG cells ATP (100 M) activates a K ϩ current that is not (i) due to activation of K Ca or (ii) accompanied by increases in To further characterize the apparent Ca 2ϩ -independent K ϩ current, the effect of charybdotoxin (ChTx) was examined. We reported earlier that ChTx, a large conductance Ca 2ϩ -dependent K ϩ channel antagonist, inhibits the carbachol-stimulated K Ca current in HSG cells (16). As seen in Fig. 5A, 25 nM ChTx also effectively blocked the K ϩ current induced by 1 and 10 M ATP but not that induced by 100 M ATP (Fig. 5A). The I-V curve of the outward current induced by 100 M ATP in ChTxtreated cells shows a weak inward rectification with a reversal potential of Ϫ75 mV, which is close to the K ϩ reversal potential (Fig. 5B). Furthermore, replacing intracellular K ϩ ion with Cs ϩ eliminated ATP-induced outward currents at all concentrations tested (from 1 M to 1 mM, n ϭ 5, data not shown). These data suggest that high [ATP] activates an inwardly rectifying, Ca 2ϩ -independent, K ϩ current in HSG cells.
Bz-ATP-induced Ca 2ϩ Influx and K ϩ Currents-In addition to the P 2U receptor, a P 2Z subtype of P 2 receptor (P2X 7 ) has also been found in the salivary gland cells (6). To determine whether the Ca 2ϩ -independent ChTx-insensitive K ϩ channel was stimulated by the activation of a P 2Z receptor, we examined the effects of Bz-ATP, a potent agonist of the P 2Z receptor, on [Ca 2ϩ ] i and whole cell current. The data are shown in Fig. 6. Fig. 6, A and B, shows whole cell currents following the sequential additions of 1, 10, 100, and 1000 M Bz-ATP to HSG cells in Ca 2ϩ -containing and Ca 2ϩ -free medium, respectively. In the presence of external Ca 2ϩ (Fig. 6A), Bz-ATP activated a K ϩ   FIG. 3. Inhibition of ATP-activated I SOC and outward K ؉ current by La 3؉ in HSG cells. Store-operated inward current (I soc , A and B) stimulated by 10 M ATP (perfused during the entire experiment) was measured by whole cell configuration at 0 mV holding potential as described under "Experimental Procedures." A, [Ca 2ϩ ] in the external medium was changed from 10 to 0 mM, shown by the corresponding bar. B, cells were exposed to external medium containing 10 mM Ca 2ϩ and 1 mM La 3ϩ , shown by the corresponding bar. These are the representative traces from five (A) or four (B) cells, respectively. K ϩ currents (C and D) were measured as described for Fig. 1  current at all concentrations tested, with the largest response, a somewhat sustained increase, obtained at 100 M. No significant response was seen at 1 mM Bz-ATP. When cells were stimulated in the absence of external Ca 2ϩ (Fig. 6B), the response at 1 M Bz-ATP was relatively more transient, while that at 10 M was greatly attenuated, and that at 100 M was barely detectable. Thus, Bz-ATP, at higher concentrations, appeared to stimulate K ϩ currents only in the presence of external Ca 2ϩ . Similar effects of Bz-ATP were seen on [Ca 2ϩ ] i (Fig.  6, C and D). P 2Z -associated cation channels in salivary gland cells are reportedly permeable to external Na ϩ and inhibited by external Mg 2ϩ (6). However, the Bz-ATP-induced [Ca 2ϩ ] i increase did not appear to be affected by replacing Na ϩ in the external medium with NMDG (Fig. 7A). Further, as shown in  , n ϭ 35). In aggregate, these data demonstrate the presence of a P 2Z (P2X 7 ) receptor in HSG cells, which can be activated by Bz-ATP to stimulate Ca 2ϩ influx, via an as yet unidentified ion channel. Low Bz-ATP also stimulates internal Ca 2ϩ release, probably via IP 3 generation or by an as yet unidentified mechanism (6,19). The resulting increase in [Ca 2ϩ ] i activates K Ca . However, the data are not consistent with the involvement of the P 2Z receptor in the activation of the ChTx-insensitive, Ca 2ϩ -independent, inwardly rectifying K ϩ channel by high [ATP].
Effects of Sulfonylurea Compounds on the ATP-induced Ca 2ϩ -independent K ϩ Current-To further characterize the ChTx-insensitive K ϩ current activated by higher [ATP], we examined the effects of the sulfonylurea compounds, which have been reported to block a family of inwardly rectifying K ϩ channels, Kir (20,21,25). Fig. 8 shows that glibenclamide (10 M) and tolbutamide (1 mM) inhibited the outward current induced by 100 M ATP. Importantly, the K ϩ current stimulated by 1 M ATP was not significantly affected by these compounds. Further, when ATP was substituted by Bz-ATP, 2MeS-ATP, UTP, or ADP no currents were detected in charybdotoxin-treated HSG cells (data not shown).
These data suggest that the ATP-activated ChTx-insensitive K ϩ current might belong to the K ATP family of Kir channels. However, since the pipette solution in these experiments contained 1 mM ATP, the K ATP channel in HSG cells appears to have a lower sensitivity for inhibition by intracellular ATP. Consistent with this, increasing internal [ATP] to 5 or 10 mM (in the presence of 1 mM Mg 2ϩ ) significantly reduced the K ϩ current (data not shown).
Presence of CFTR and ROMK1 Proteins in HSG Cells-It has been suggested that the sulfonylurea sensitivity of K ATP channels might be conferred via interaction with sulfonylurea-binding proteins such as SUR or CFTR protein (20, 21, 25). The association of the classical K ATP channel (Kir 6.2) in pancreatic ␤ cells with SUR has been well established. This K ϩ channel demonstrates high sensitivity for intracellular ATP and glibenclamide (25). In contrast, the ROMK family of weakly inwardly rectifying epithelial K ϩ channels have been reported to have a lower sensitivity for inhibition by intracellular ATP. These K ϩ channels also have a relatively lower sensitivity to glibenclamide (20,21). The K ATP channels in kidney epithelial cells (Kir 1.1a and 1.1b, ROMK1 and ROMK2, respectively) have been suggested to be regulated via interaction with the CFTR protein, although presently there are no data that conclusively demonstrate this molecular interaction. The presence of the CFTR protein was previously demonstrated, both functionally and by immunofluorescence, in the luminal membrane of salivary gland ductal cells (26 -28). Thus, we examined the presence of CFTR in HSG cells, a human submandibular gland ductal cell line, by Western blotting using an anti-human CFTR antibody. A strong reactivity was detected in the plasma membrane fraction of HSG cells (Fig. 9A, lane 2) associated with a protein of with an approximate molecular mass of 180 kDa, which co-migrated with the control CFTR protein (see lane 1, plasma membrane preparation of HEK cells infected with an adenovirus encoding the CFTR). Additionally, consistent with the reported intracellular localization of the CFTR protein (29,30), more reactivity was detected in a crude membrane fraction (lane 3), which probably also includes intracellular organelles. These data suggest that HSG cells, similar to rat submandibular gland ductal cells and kidney cells, have the CFTR protein.
Further, an antibody against the ROMK protein was used to examine the presence of this protein in HSG cells. The data are shown in Fig. 9B. Three prominent bands were detected in the plasma membrane fraction of HSG cells with estimated molecular masses of about 49, 47, and 42 kDa. We have not yet identified the higher molecular weight proteins that react with the antibody, which supposedly recognizes all isoforms of To confirm the presence of ROMK in HSG cells, RT-PCR was performed using ROMK-specific primers and HSG cell mRNA ( Fig. 9C; see "Experimental Procedures" for details). A ϳ1.2kilobase RT-PCR product was strongly detected (lane 1). Further, when RT-PCR was performed using SUR1-, SUR2A-, or SUR2B-specific primers, SUR-homologous sequences were not detected in HSG cell mRNA (lanes 2, 4, and 6). However, as reported earlier (31) One of the suggested criteria for the possible functional association between the CFTR and ROMK proteins in a cell has been the demonstration that cAMP can stimulate both Cl Ϫ and K ϩ currents in the cells. The data in Fig. 10 show that HSG cells fulfill this criterion. Inclusion of 10 M cAMP in the pipette solution stimulated a small inward Cl Ϫ current (250 Ϯ 34 pA, n ϭ 5, at Ϫ60 mV holding potential with 150 mM NMDGCl in the external and internal solutions). The current developed slowly, remained stable for a short time, and decreased to a lower level. It was blocked by glibenclamide (data not shown). These data provide evidence for the presence of a functional CFTR protein in HSG cells. Fig. 10B shows the activation of outward K ϩ current by 10 M cAMP. The outward current was of smaller magnitude, and relatively more sustained, than that seen with ATP. The current also showed a decreased sensitivity toward glibenclamide (data not shown), as has been previously reported for the ROMK channels in kidney cells (20).

DISCUSSION
The data presented above demonstrate that ATP activates multiple P2 receptors to increase [Ca 2ϩ ] i and activate distinct cation channels in HSG cells. Importantly, the data provide evidence that external ATP activates a novel glibenclamidesensitive, Ca 2ϩ -independent, inwardly rectifying K ϩ current that is mediated by a K ATP channel. Further, we have shown that a ROMK-like channel and CFTR, but not SUR, are present in HSG cells. Thus, we suggest that this K ATP channel belongs to the ROMK subfamily of K ATP channels, which might be regulated via an interaction with the CFTR. These data reveal a potentially new mechanism for K ϩ secretion in salivary epithelial cells. P2Y 1 , P2Y 2 (P 2U ), P2X 4 , and P2X 7 (P 2Z ) receptors are expressed in salivary glands (6). We have shown here that low [ATP] and [UTP] are equally potent in activating internal Ca 2ϩ release. Further, the response to low [ATP] was greater than the response to low [2MeS-ATP] (data not shown). These data suggest the presence of a P 2U receptor in HSG cells and are consistent with previous studies showing that in HSG cells the P 2U receptor has a nucleotide selectivity in the order ATP ϭ UTP Ͼ ADP (14). Additionally, P 2U -stimulated depletion of the internal Ca 2ϩ store resulted in the activation of the storeoperated Ca 2ϩ influx current (I SOC ) that was inhibited by La 3ϩ and Gd 3ϩ , but not by Zn 2ϩ . Thus, I SOC activated by ATP has similar characteristics as that activated by CCh or Tg (Fig. 3-5; also see Ref. 17). Low (0.1-10 M) ATP also activated an outward K ϩ current with characteristics similar to those previously reported for K Ca in HSG cells (15,16): (i) inhibition by charybdotoxin, but not by apamin, and (ii) dependence on [Ca 2ϩ ] i . Notably, this is the first report describing ATP-dependent activation of K Ca and I SOC in human submandibular gland cells. The present data demonstrate that P 2Z receptors are also present in HSG cells. Bz-ATP, a potent P 2Z receptor agonist, induced responses that were distinct from that induced by either UTP or ATP. Increasing concentrations of Bz-ATP produced dose-dependent increases in both [Ca 2ϩ ] i and the K ϩ current. At lower concentrations, Bz-ATP induced increase in [Ca 2ϩ ] i and K Ca in the absence of external Ca 2ϩ . As has been suggested earlier, this could be due to (i) release of Ca 2ϩ from intracellular stores by Na ϩ that might enter the cells when Ca 2ϩ is removed from the external medium or (ii) stimulation of IP 3 generation (6). However, the effects at higher Bz-ATP were acutely dependent on the presence of external Ca 2ϩ and were more sensitive to high external Mg 2ϩ . These data suggest that higher [Bz-ATP] induces a Ca 2ϩ influx component that is apparently not associated with internal Ca 2ϩ store depletion  Fig. 1, except that the pipette solution contained 10 M cAMP. (Fig. 5, compare A and B with C and D). 2MeS-ATP did not induce a similar Ca 2ϩ influx response, even at higher concentrations, while 100 M ATP induced a small increase in Ca 2ϩ influx, in the absence of external Mg 2ϩ . In aggregate, these data are consistent with the presence of a P 2Z receptor in HSG cells. However, it is unlikely that ATP induces any significant effects via this receptor in a normal Mg 2ϩ -containing medium.
The important, and novel, finding of this study was that high [ATP] (Ͼ10 M) induced an outward current that was not accompanied by an increase in [Ca 2ϩ ] i . We have shown that the current was detected in cells previously stimulated with low [ATP] (Fig. 1), Tg (Fig. 4), or CCh (not shown) and was not affected by the removal of Ca 2ϩ from the external medium ( Fig.  1) or treatment of cells with charybdotoxin. Furthermore, the current measured in the presence of high [ATP] and charybdotoxin (i) displayed a weak inward rectification with a reversal potential of Ϫ75 mV, a value close to the K ϩ equilibrium potential, and (ii) was blocked by replacing intracellular K ϩ with Cs ϩ . Taken together, these data provide strong evidence for the activation of a Ca 2ϩ -independent K ϩ channel by high [ATP] in HSG cells. Some previous studies with salivary gland cells have suggested that ATP induces a Ca 2ϩ -independent component of K ϩ efflux (24,32). It was suggested that this component of K ϩ release could be due to a direct action of ATP on nonspecific cation channels on the plasma membrane. We have not yet determined whether ATP activates the Ca 2ϩindependent K ϩ channel directly or via binding to an ATP receptor. However, our data rule out the possibility that the known P2X or P2Y receptors are involved in the ATP-dependent activation of the Ca 2ϩ -independent, inwardly rectifying, K ϩ channel. The charybdotoxin-insensitive K ϩ current was not activated by 2MeS-ATP, Bz-ATP, UTP, or low [ATP]. Thus, the Ca 2ϩ -independent K ϩ current appears to be selectively activated by high [ATP] via a mechanism that is distinct from that associated with the P 2Z or P 2U receptors. Significantly, the sulfonylurea compounds, glibenclamide and tolbutamide, inhibited the K ϩ current induced by high [ATP] but not that induced by low [ATP], in contrast to charybdotoxin, which inhibited only the current induced by low [ATP]. These data provide evidence for the presence of at least two pharmacologically distinct K ϩ channels in HSG cells: a Ca 2ϩ -dependent maxi-K ϩ channel that is sensitive to charybdotoxin and an inwardly rectifying K ϩ channel that is sensitive to sulfonylurea compounds, probably a K ATP channel (20,21,25).
Two classes of K ATP channels are modulated by sulfonyl urea compounds: the ROMK channels and the classical K ATP channels (20,21,25). Classical K ATP channels (Kir 6.0 subfamily) have been shown to be present in pancreatic ␤ cells, cardiac and skeletal myocytes, vascular smooth muscle, and neurons and are suggested to regulate the electrical activity of these cells (25). These K ATP channels have been shown to be associated with the sulfonylurea receptors (SUR1, SUR2A, or SUR2B), which confer the sensitivity to the sulfonylurea compounds (see Ref. 25 for a review). The ROMK family of K ATP channels (Kir 1.1 family) are expressed in the nephron, where they have been proposed to regulate K ϩ secretion (20,21,33,34). In the case of the ROMK channels, the sensitivity to sulfonylurea compounds has been suggested to depend on an interaction with the CFTR protein, although conclusive experimental data for a molecular interaction between these two proteins is yet lacking (20,21,35). However, several observations are consistent with this suggestion. For example both the CFTR and the ROMK proteins were shown to be localized in the apical membrane of the nephron (33, 34). Co-expression of the ROMK1 protein and CFTR in oocytes conferred glibenclamide sensitivity to the K ϩ current (35). Further, cAMP stimulated both the CFTR-associ-ated Cl Ϫ current and the inwardly rectifying, glibenclamidesensitive, K ϩ current, and cAMP plus ATP decreased the sensitivity of the K ϩ channel for glibenclamide (20).
The present data demonstrate that HSG cells have (i) a Ca 2ϩ -independent, sulfonylurea-sensitive, inwardly rectifying K ϩ channel (K ATP ) activity, (ii) the ROMK channel and CFTR, and (iii) cAMP-stimulated Cl Ϫ and K ϩ currents. Thus, the K ATP activity detected in HSG cells fulfills the criteria associated with the ROMK family of K ATP channels. Based on measurements of electrolytes in saliva and 86 Rb ϩ flux studies, it has been suggested that salivary gland ducts mediate K ϩ secretion into the lumen (36). Thus, the currently accepted model for ion fluxes in the salivary gland ductal cell depicts a pathway for K ϩ efflux via the apical membrane. However, presently there are no data to demonstrate the identity or function of this K ϩ conductance. Based on the data discussed above, we suggest that the ROMK-type K ATP channel detected in HSG cells might be a likely candidate for this K ϩ secretory function. Further, as has been proposed for the regulation of the ROMK channels in kidney cells, the present data suggest that the ROMK channel in HSG cells might also be regulated via interaction(s) with the CFTR protein. Importantly, we have clearly demonstrated that CFTR, but not SUR, is present in HSG cells. Thus, HSG cells provide an excellent experimental system for further studies to examine the putative mechanism by which CFTR regulates the ROMK channel. Notably, the CFTR protein has also been shown to be present in the luminal membrane of rat submandibular gland ductal cells (27,28), and studies reported by Muallem and co-workers have demonstrated cAMP-dependent activation of CFTR-associated Cl Ϫ current (i.e. Ca 2ϩ -independent, glibenclamide-sensitive) in rat submandibular gland ductal cells (7). However, the exact physiological role of CFTR in the ductal luminal membrane of salivary epithelial cells is still unclear.
In summary, ATP triggers two distinct Ca 2ϩ signaling pathways in HSG cells through the P 2U and P 2Z receptors, respectively. The Ca 2ϩ signaling pathway linked to the P 2U receptor, is independent of external Na ϩ and Mg 2ϩ and is associated with activation of K Ca and SOC. In contrast, the Ca 2ϩ signaling pathway via the P 2z receptor does not appear to be activated by ATP under normal conditions (i.e. with 1 mM external Mg 2ϩ ). However, it is associated with Ca 2ϩ influx, probably via an as yet unidentified Ca 2ϩ -permeable cation channel, and activation of K Ca when cells are exposed either to an agonist stronger than ATP or to ATP in low external [Mg 2ϩ ]. Importantly, we have shown here that ATP also activates a novel, Ca 2ϩ -independent, weakly inwardly rectifying K ϩ channel in HSG cells, which is blocked by sulfonylurea compounds but not by charybdotoxin. Together with our demonstration of the presence of CFTR, a ROMK channel, and the activation of both Cl Ϫ and K ϩ currents in HSG cells by cAMP, the present data strongly suggest that this K ϩ channel belongs to the ROMK family of K ATP channels. This putative salivary ROMK channel provides a potentially novel mechanism for the regulation of K ϩ secretion in salivary epithelial cells.