5-HT1A Receptor Activates Na+/H+ Exchange in CHO-K1 Cells through Giα2 and Giα3

5-HT1A receptors couple to many signaling pathways in CHO-K1 cells through pertussis toxin-sensitive G proteins. The purpose of this study was to determine which members of the Gi/o/z family mediate 5-HT1A receptor-activated Na+/H+ exchange as measured by microphysiometry of cell monolayers. The method was extensively validated, showing that proton efflux was sodium-dependent, inhibited by amiloride analogs, and activated by growth factors, phorbol ester, calcium ionophore, and hypertonic stress. 5-HT and the specific agonist (±)-8-hydroxy-2-(di-N-propylamino)tetralin hydrobromide rapidly stimulated proton efflux that was blocked by a specific receptor antagonist, amiloride analogs or pertussis toxin. The activation by 5-HT depended upon extracellular sodium and could be demonstrated under conditions of imposed intracellular acid load, as well as in the presence and absence of glycolytic substrate. Acceleration of proton efflux was not inhibited by sequestration of G protein βγ-subunits, a maneuver that blocked 5-HT1A receptor activation of mitogen-activated protein kinase. Transfection of Gzα and pertussis toxin-resistant mutants of Goα and Giα1 did not reverse the blockade induced by pertussis toxin. In contrast, pertussis toxin-resistant mutants of Giα2 and Giα3 “rescued” the ability of 5-HT to increase proton efflux after pertussis toxin treatment. These experiments demonstrate clearly that Giα2 and Giα3 can specifically mediate rapid agonist-induced acceleration of Na+/H+ exchange.

at the surface of all mammalian cells, subserving diverse functions including regulation of intracellular pH, cell volume, mitogenesis, and vectorial and nonvectorial transepithelial transport of Na ϩ and acid-base equivalents (1,2). The NHE-1 exchanger subtype is widely expressed and can be activated by a number of stimuli that include growth factors, hyperosmolarity, and ligands that act through G protein-coupled receptors (1)(2)(3)(4). Surprisingly little is known about the intermediary signals and effectors involved in stimulation of NHE-1 activity, particularly the pathways activated by receptors that act through heterotrimeric G proteins.
A number of receptors that are classically linked to the inhibition of adenylyl cyclase have been shown to inhibit or stimulate NHE activity through pathways that do not involve cAMP. Unexpectedly, in many cases those receptors couple to the regulation of NHE through pathways that are not sensitive to pertussis toxin. Hence, G proteins other than G i␣ or G o␣ are usually thought to convey the modulatory signal. For example, the D 2 dopamine receptors expressed in pituitary lactotroph cells cause intracellular acidification by inhibiting NHE activity (5). Likewise, somatostatin receptors expressed in pituitary lactotrophs, Ltk Ϫ fibroblasts and HEK293 cells also inhibit NHE in a non-pertussis toxin-sensitive manner (6,7). In contrast, D 2 dopamine receptors expressed in mouse L cells and C6 glioma cells activate NHE (8), as do ␣ 2 -adrenergic, muscarinic, and ␦-opiate receptors expressed in NG108-15 hybridoma cells (9), and all act through pertussis toxin-insensitive mechanisms. The common thread in all of those studies (5)(6)(7)(8)(9) is that receptors that inhibit adenylyl cyclase through pertussis toxinsensitive G proteins do not use those G proteins to modulate NHE activities. That notion has been supported by recent studies in which constitutively active G protein ␣-subunits were transiently expressed in mammalian host cells. Those studies showed clearly that G q␣ , G 12␣ , and G 13␣ can increase NHE activity, the former two through a protein kinase C-dependent (phorbol ester-sensitive) pathway, and the latter through a protein kinase C-independent (phorbol ester-insensitive) pathway (10 -12). However, in those studies, constitutively active G i proteins failed to increase the rate of recovery from an acid load. Thus, a number of careful studies have failed to show a role for G i proteins in regulating NHE activity (5)(6)(7)(8)(9)(10)(11)(12). Nevertheless, there has been mounting contrary evidence that supports some role for pertussis toxin-sensitive G proteins in * This work was supported in part by United States Public Health Service Grants NS30927, DK52448, and DK42486, a VA Merit Award, a shared equipment grant from the Department of Veterans Affairs, and monies from the Division of Nephrology Research Fund at Duke University. The laboratory of J. R. R. is supported by an endowment jointly administered by the MUSC Division of Nephrology and Dialysis Clinics, Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** Recipient of a postdoctoral award from the Alberta Heritage Foundation during the course of these studies.
For example, D 2 dopamine receptors, and the closely related D 3 and D 4 dopamine receptors, have been shown to stimulate NHE activity in CHO cells through pertussis toxin-sensitive G proteins (13,14). Indeed, angiotensin 1 receptors in opossum kidney cells (15) and endothelin receptors in renal cortical slices (16) were also shown to increase NHE activity through pertussis toxin-sensitive G proteins. Finally, supportive evidence for the involvement of G i or G o proteins in the regulation of NHE was provided by recent studies that showed an association between increased NHE activity and pertussis toxinsensitive second messenger generation in immortalized B lymphoblasts derived from hypertensive patients when compared with those derived from normotensive patients (17,18). Therefore, the role of pertussis toxin-sensitive G proteins in the regulation of NHE activity remains unresolved.
The purpose of the current study was to determine whether members of the G i/o/z family of G proteins regulate NHE-1 activity in CHO-K1 cells, and if so, which ones. We chose CHO-K1 fibroblast cells, because they have previously proven useful in elucidating some of the pathways involved in plateletderived growth factor receptor (4,19)-and dopamine receptor (13,14)-stimulated NHE-1 activity. To have a means by which pertussis toxin-sensitive G proteins could be selectively activated, we transfected CHO-K1 cells with the human 5-HT 1A receptor, a prototypical G i␣ -linked receptor. That receptor has been shown to modulate a large number of signaling pathways in CHO-K1 cells (20 -23) exclusively through pertussis toxinsensitive pathways. Therefore, the 5-HT 1A receptor serves as a very useful and specific tool to achieve short term activation of G i/o proteins. G i␣2 and G i␣3 are the major known pertussis toxin-sensitive G protein subtypes in CHO-K1 cells (23)(24)(25)(26), although a measurable but far smaller amount of G o␣ is also present (27). Importantly for the current study, the 5-HT 1A receptor has already been shown to efficiently activate all three G i␣ subtypes, G o␣ , and G z␣ (26, 28 -30). Because most cells endogenously express two or more subtypes of pertussis toxinsensitive G proteins, it has been very difficult to assign specific downstream regulatory functions to individual ␣-subunit types.
The basic strategy in the current study was to neutralize endogenous pertussis toxin-sensitive G proteins, then to attempt to rescue NHE activation by transfecting at high efficiency either G z␣ or various pertussis toxin-insensitive mutants of G i␣ and G o␣ subunits. Each of those constructs, all characterized previously (31)(32)(33), was created by mutating the cysteine four residues from the carboxyl terminus of each G protein to remove the ADP-acceptor site for the ribosylation catalyzed by pertussis toxin. By transfecting those constructs and subsequently neutralizing the endogenous G i␣ and G o␣ subunits, we were able to study the effects of each G protein ␣-subunit individually. Thus, this model system allowed us to examine the specific contribution of G i/o/z proteins to the regulation of NHE-1 activity.
Microphysiometry-Na ϩ /H ϩ exchange activity was measured in real time as the rate of decrease in extracellular pH in intact cells placed in an eight chamber Cytosensor™ microphysiometer (Molecular Devices Corp., Sunnyvale, CA) (36,37). The microphysiometer uses a light addressable silicon sensor to detect extracellular protons, which can be derived primarily from Na ϩ /H ϩ exchange and glycolysis, and from other metabolic pathways (36,37). Rate data transformed by a personal computer running Cytosoft™ version 2.0 (Molecular Devices Corp.) were presented as the extracellular acidification rate (ECAR) in microvolts/s, which roughly correspond to millipH units/min (Nernst equation). To facilitate comparison of data between two channels, values were expressed as a percentage of the base line as determined by computerized analysis of the five data points prior to exposure of the cell monolayers to a test substance. CHO-K1 cells expressing the 5-HT 1A receptor were grown in Ham's F-12 medium supplemented with fetal bovine serum (10%), penicillin (100 units/ml), and streptomycin (100 g/ml). The night prior to experimentation, cells were replated onto polycarbonate membranes (3 pore size, 12 mm size) at a density of 300,000 cells per insert. The day of the study, cells were washed with serum-free, bicarbonate-free Ham's F-12 medium, placed into the microphysiometer chambers, and perfused at 37°C with the same medium or balanced salt solutions as described in figure legends or text. For most studies, the pump cycle was set to perfuse cells for 60 s, followed by a 30-s "pump-off" phase, during which proton efflux was measured from the 6th through the 28th second. Cells were exposed to the test agent for two or three cycles (180 -270 s). Valve switches (to add or remove test agents) were performed at the beginning of the pump cycle. In some cases where the response was expected to be both rapid and very transient (i.e. recovery from sodium-free conditions) the valve switch was performed 55 or 58 s into the perfusion phase, allowing 7-10 s for solution mixing prior to rate measurements. Data points were then acquired every 90 s. The peak effect during stimulation was expressed as percentage increase from average basal ECARs from five rate measurements prior to application of the test agent(s). Typical basal ECARs in glucose-containing solutions were in the Ϫ120 -200 V/s range, regardless of the presence of sodium. In glucose-depleted pyruvate-containing solutions, basal rates were in the range of Ϫ25-60 V/s.
For some experiments, intracellular pH (pH i ) was fixed by a method similar to that used by Azarani et al. (38) to measure NHE activity by 22 Na ϩ uptake. That method employs nigericin, a K ϩ /H ϩ exchange ionophore. In using this tool, two major assumptions were made. (i) At equilibrium the desired pH i could be calculated from the [K ϩ ] gradient and pH o using the following formula: (ii) The intracellular [K ϩ ] was assumed to be 140 mM. Briefly, cells were perfused with a low-sodium balanced salt solution containing (in mM) 118 choline chloride, 20 NaCl, 10 glucose, 5 KCl, 1.3 CaCl 2 , 0.8 K 2 HPO 4 , 0.5 MgCl 2 , and 0.1 KH 2 PO 4 , pH Ϸ 7.4. One or two ECAR readings were obtained after stimulation with 1 M 5-HT, then cells were perfused for 10 min with the same salt solution supplemented with 10 M nigericin except that the [K ϩ o ] was adjusted to provide a calculated pH i of 6.6. The choline content was adjusted such that the sum of choline chloride and KCl equaled 123 mM. To achieve a pH i of 7.0, [K ϩ o ] and pH o were increased accordingly. After perfusion with nigericin, cells were quenched with the same buffer supplemented with 0.5% bovine serum albumin (minus nigericin) to remove free nigericin, then exposed to 1 M 5-HT. The presence or absence of the bovine serum albumin did not affect the basal ECAR, nor did it affect 5-HT-stimulated ECAR.
Determination of pH i Using BCECF-AM-Dye Technique-Cells grown on coverslips were loaded with 7 M BCECF-AM in a Hepesbuffered, bicarbonate-free balanced salt solution at 37°C for 1 h. Cells were washed and loaded into cuvettes that were perfused at 37°C. Measurements were made with a Perkin-Elmer LS50 fluorescence spectrometer using an emission wavelength of 530 nM and excitation wavelengths of 500 and 440 nm. Periodic measurements were made to calculate the excitation ratio (500/440), which were compared with a calibration curve that was generated by clamping pH i with nigericin as described above.
Expression of ␤ARK-1 Minigene in CHO-K1 Cells-Cells were transfected with 1 g of pRK-␤ARK-1-(495-689) or empty pRK vector in the presence of LipofectAMINE™ liposomes or infected with 50 plaque-forming units/cell of ␤ARK-1-(495-689)-adeno or a control virus. Transfections for microphysiometry were performed with 1 g of DNA/300,000 cells and for immunoblot with 2 g of DNA/1,000,000 cells. After 40 -48 h, microphysiometry was performed as described above. For immunoblots, cells were scraped into Laemmli buffer, boiled for 3 min, and subjected to SDS-polyacrylamide gel electrophoresis under reducing conditions with 4 -20% precast gels (Novex, San Diego, CA). After semidry transfer to polyvinylidine difluoride membranes, proteins were subjected to immunoblot with anti-␤ARK-1 rabbit serum (1:8,000) (39), and immunoreactive bands were visualized using an enhanced chemiluminescence technique (ECL, Amersham Corp.). We used several marker constructs including a 12CA5 epitope-tagged 5-HT 1A receptor and ␤-galactosidase staining to estimate the transfection efficiency of both protocols. Lipopolyamine-mediated gene transfer resulted in expression of the marker in at least 50 -80% of cells at 48 h, and adenovirus-mediated transfer resulted in expression in essentially 100% of the cells.
Expression of G Protein Constructs in CHO-K1 Cells-All G protein constructs were subcloned into the pCDNA3 vector, which incorporates a G418 resistance motif. Stable expression of G z␣ was achieved by co-transfection with the cDNA encoding the human 5-HT 1A and a hygromycin resistance motif; selection in the presence of both G418 and hygromycin produced several clones, one of which contained Ϸ800 fmol of 5-HT 1A receptor/mg of protein and expressed G z␣ by immunoblot, was used for the current study. Transient expression of other constructs was achieved in those cells by transfection (2 g of each G protein cDNA/one well of a six-well dish) in the presence of LipofectAMINE™ (2 h in serum-and antibiotic-free medium). Further studies were performed as described above.
MAPK Assay-MAPK activity was measured in anti-MAPK immunoprecipitates using myelin basic protein as a phosphorylation substrate as described previously (27,39).
High Affinity Agonist Binding-High affinity binding was defined as the amount of [ 3 H]8-OH-DPAT (1 nM) displaced by 100 M GTP, and the assays were performed as described previously (40). Under these conditions, each tube contained about 3,000 cpm of specifically bound [ 3 H]8-OH-DPAT, and GTP-displaced binding was about 90% of the specific binding (displaced by 10 M 5-HT).

RESULTS
To establish the ability of the Cytosensor™ to detect changes in Na ϩ /H ϩ exchange activity under our experimental conditions, cells were perfused with a sodium-free balanced salt solution (138 mM choline chloride, 10 mM glucose, 5 mM KCl, 1.3 mM CaCl 2 , 0.8 mM K 2 HPO 4 , 0.5 mM MgCl 2 , and 0.1 mM KH 2 PO 4 , pH 7.35) for 20 -30 min, then switched to the same solution where NaCl was substituted for choline chloride. Fig.  1A shows that a rapid burst of proton efflux occurred when cells were exposed to sodium. Under identical conditions in which cells were exposed to amiloride analogs EIPA (5, 10, or 20 M) or MIA (10 M) during the choline sodium-free perfusion, the burst of proton efflux was completely blocked upon exposure to sodium (Fig. 1A). Those studies confirm the presence of an amiloride-inhibitable sodium-dependent proton efflux pathway in CHO-K1 cells. We next wanted to establish that the ECAR could be acutely increased by maneuvers that are known to activate fibroblast NHE as measured by other techniques. Such measures include hypertonic stress, phorbol 12-myristate, 13acetate acting through protein kinase C, maneuvers that increase intracellular Ca 2ϩ levels, and growth factors such as fibroblast growth factors (FGF) and thrombin (41)(42)(43)(44). Fig. 1B demonstrates that exposure to hypertonic solutions (270 mM NaCl or 170 mM sucrose ϩ 100 mM NaCl) resulted in brisk increases in ECAR, as would be expected if the proton efflux is mediated through NHE-1 (5, 6). Fig. 1C shows that phorbol 12-myristate, 13-acetate, and the calcium ionophore A23187, each rapidly activate ECAR in CHO cells, as do the growth factors basic FGF (100 pg/ml) and thrombin (1 unit/ml) (Fig.  1D). In aggregate, the studies presented in this Fig. 1 show that ECAR as measured by microphysiometry corresponds very closely with what is known about the activity and properties of NHE.
It was important to next demonstrate that the G proteincoupled 5-HT 1A receptor can activate ECAR in these cells. This evidence was provided by demonstrating that the agonists 5-HT and 8-OH-DPAT (at 1 M) stimulate ECAR in Ham's F-12 by 42 Ϯ 3% (n ϭ 22) and 38 Ϯ 3% (n ϭ 19), respectively. Those effects could be blocked effectively by co-incubation with 1 M of the 5-HT 1A receptor antagonist (S)-UH-301, which itself had no effect (Figs. 2A). Moreover, there was no stimulation of ECAR by 5-HT or 8-OH-DPAT in cells that were not transfected with the 5-HT 1A receptor, confirming that the increase in ECAR is indeed mediated through that receptor (n ϭ 12).
It was next necessary to provide evidence that the receptormediated increases in ECAR were due to activation of NHE. Fig. 2, C and D, support that hypothesis by demonstrating that 5-HT increased ECAR in a sodium-containing balanced salt solution and that the amiloride analog EIPA (10 M) attenuated the 5-HT (1 M)-mediated increase in acidification rate by Ϸ70 -80%. Because hormones and growth factors can have complex cellular effects, including activation of metabolic pathways (glycolysis and mitochondrial respiration) which can lead to intracellular accumulation of protons (37,38), the increase in ECAR could either be due to a primary activation of the NHE molecules through a regulatory pathway or secondarily to a decrease in pH i leading to activation of NHE through a protonsensing site located within the protein itself (44). To address The response to all of the stimulating agents (B-D) was rapid and relatively shortlived after removal of the agent, with the exception of basic FGF, which intiated a slower and less abrupt increase in ECAR, which nevertheless persisted long after the stimulus was removed (Ͼ30 min). Shaded areas represent times during which cells were exposed to test agents. these possibilities, we performed three sets of experiments. First, pH i was measured before and after treatment with 5-HT. The resting pH i of nonacid-loaded CHO-K1 cells in monolayer at 80% confluence was determined to be 7.23 Ϯ 0.03 by the BCECF-AM fluorescence method, and this was not decreased by treatment with 5-HT (Fig. 2B, n ϭ 5). Second, glucose was removed from the perfusion medium, and pyruvate was substituted as the carbon source for some microphysiometry studies. Removal of glucose rapidly and dramatically reduced ECAR by nearly 80% within one pump cycle (minutes); ECAR returned quickly to the previous base line (several minutes) after restoring the glucose to the perfusate solution (not shown). Thus, CHO-K1 cells appeared to rely quite heavily on glucose for maintaining metabolism and ECAR under our conditions. Therefore, if the increase in ECAR mediated by 5-HT was mediated solely through augmented glycolysis, removal of glucose would be expected to abrogate the stimulation of ECAR by 5-HT. In the pyruvate-containing, glucose-free solution, 5-HT increased acidification rate by 28 Ϯ 4% (n ϭ 6 in duplicate or triplicate), and that stimulation was almost completely blocked by 10 M EIPA (n ϭ 12) (Fig. 2D). In a third set of experiments, the pH i was clamped either to 6.6 or 7.0 by manipulating the [K ϩ o ] and [pH o ] in the presence of nigericin. Under those conditions, 5-HT was still able to increase ECAR by 25 Ϯ 3% and 29 Ϯ 4%, respectively, confirming that the increase in ECAR can occur under a condition (low pH i ) in which a further reduction in pH i would be unlikely to affect the proton-sensing site of the NHE molecule (44). The experiments in Fig. 2, C-E, provide consistent evidence against a passive mechanism of activation of ECAR that is secondary to decreased pH i resulting from increased metabolic activity.
The data in Fig. 2F  We next explored potential pathways by which the 5-HT 1A receptor increased NHE, with particular emphasis upon identifying the G proteins involved in the process. Fig. 3A shows that pertussis toxin treatment (200 ng/ml overnight) completely abolished the ability of 5-HT 1A receptors to increase ECAR without altering the ability of ATP (acting through an endogenous G q␣ -linked purinergic receptor) to do the same. Because pertussis toxin shows a selective uncoupling of the 5-HT 1A receptor from downstream effectors, its effects are not due to nonspecific cellular toxicity. Moreover, it provided a null background against which the effects of the various pertussis toxin-insensitive G proteins could be studied. Fig. 3B shows that a minigene construct encoding the carboxyl terminus of ␤ARK-1, which is a known sequesterer of G protein ␤␥-subunits (34,39), did not alter the ability of 5-HT to increase the rate of acidification. However, that treatment on the same batch of cells completely blocked the ability of 5-HT to stimulate the mitogen-activated protein kinases, ERK1 and -2 ( Fig.  3C), confirming both high transfection efficiency and functional expression of the sequestering minigene. The lack of effect upon NHE activity was apparent whether the minigene was delivered by liposome-mediated transfection or by infection with replication-deficient adenovirus. Regardless of the delivery system, ample expression of a 24-kDa ␤ARK-1 carboxyl terminus fragment (␤1-CT) was documented by immunoblot. The apparent densitometric immunoreactivity was at least 20fold greater for ␤1-CT than for the endogenous Ϸ80-kDa ␤ARK-1 holoprotein (not shown). Those results suggest a lack of critical involvement of ␤␥ G protein subunits and are consistent with the necessary involvement of pertussis toxin-sensitive G protein ␣-subunits in the activation of NHE by the 5-HT 1A receptor.
The final series of studies was performed to assess the role of the ␣-subunits of specific members of the G i/o/z family of G proteins in conveying the stimulatory signal from the 5-HT 1A receptor to the NHE molecule. In that respect, we limited our studies entirely to members of the family to which the 5-HT 1A receptor has been shown previously to couple, those being G i␣1 , G i␣2 , G i␣3 , G o␣ , and G z␣ (26, 28 -30). Transfection of each of the constructs encoding a specific pertussis toxin-resistant G protein ␣-subunit resulted in increased expression of that subunit as detected by immunoblot (Figs. 3D and 4). Although it is difficult to quantify the exact amount of a given G protein in membrane fractions due to issues of antibody specificity and  7). D, the same experiment as described in C was repeated in salt solution with pyruvate substituted for glucose to minimize the glycolytic component of the acidification response. E, 5-HT activates ECAR in the presence of intracellular acid load imposed by nigericin. F, ability of 5-HT to increase ECAR depends upon the extracellular concentration of Na ϩ . Typical basal ECARs were ϷϪ120 -200 V/s, with the exception of D, which was Ϸ Ϫ25-60 V/s. All tracings are representative examples, with the exception of that shown in D, which was derived from the pooled values of four chambers at one sitting. Shaded areas represent times during which cells were exposed to test agents. sensitivity, we probed membranes derived from cells transfected with the cDNAs of the various pertussis toxin-resistant G proteins with a polyspecific antiserum (P960, Fig. 3D) and more specific sera (Fig. 4). P960 is a rabbit serum raised against the common GTP binding region of G ␣ (45) and is capable of interacting with all of the known heterotrimeric G protein ␣-subunits. Transfection with all of the plasmids except the empty pcDNA3 vector increased the immunoreactivity detected by P960 from 130 -210% (Fig. 3D). Confirmation that G proteins were expressed after transfection with each plasmid was obtained (Fig. 4) with more specific antisera (23,26). Because different sera were used for each blot, one cannot directly compare the level of expression of each construct. Nevertheless, the results clearly document detectable expression of all of the G protein constructs which were tested in this work.
We next assessed whether the transfected G proteins were capable of interacting with the 5-HT 1A receptor. This was accomplished after transfection by measuring high affinity agonist binding ([ 3 H]8-OH-DPAT), which is a gauge of the ability of agonist-bound receptors to couple to G proteins. Transfected cells were treated with pertussis toxin or vehicle, and the ability of each construct to "rescue" high affinity binding after pertussis toxin treatment was measured. In cells transfected with empty vector, pertussis toxin eliminated all detectable high affinity binding as determined by the ability of GTP to suppress binding of [ 3 H]8-OH-DPAT. The pertussis toxin-resistant G proteins were able to restore high affinity binding in varying degrees as follows and as depicted in Fig. 3D: G o␣ , 37%; G i␣1 , 42%; G i␣2 , 55%; G i␣3 , 43%; and G z␣ , 13%. Thus, the transfected constructs were shown to be functional in their ability to couple to the 5-HT 1A receptors in CHO cells with a rank order of G i␣2 Ն G i␣3 ϭ G i␣1 ϭ G o␣ Ͼ G z␣ in our assay system. These results are similar to those previously published for 5-HT 1A receptors interacting with nonmutant G proteins in nonmammalian expression systems (28,30).
The expression of any of the G protein constructs did not significantly alter the basal ECAR (not shown) or the extent of 5-HT-mediated increases in ECAR (Fig. 3E, open bars). In contrast, pertussis toxin treatment reduced the basal ECAR by almost 50% in cells transfected with each of the constructs, as it did in nontransfected cells (not shown). There were clear differences among the individual constructs in the ability to rescue 5-HT-mediated increases of ECAR after treatment with pertussis toxin. Whereas 5-HT did not increase ECAR in pertussis toxin-treated cells that were transfected with empty vector (pcDNA3), G z␣ PT, G o␣ PT, or G i␣1 PT, it did substantially increase ECAR in cells transfected with G i␣2 PT (22.0 Ϯ 3.8%, n ϭ 5) or G i␣3 PT (29.8 Ϯ 7.8%, n ϭ 5) when compared with the same cells that had not been treated with pertussis toxin (36.4 Ϯ 6.7% and 40.5 Ϯ 10.4%, respectively; Fig. 3E) or when compared with cells transfected with empty vector and not treated with pertussis toxin (31.7 Ϯ 5.2%, n ϭ 3).
Thus, although we were able to document detectable expression of all five pertussis toxin-insensitive G protein constructs, only G i␣2 PT and G i␣3 PT were able to restore the activation of NHE by the 5-HT 1A receptor. Moreover, because sequestration of G protein ␤␥-subunits had no effect, the ␣-subunits themselves are likely to directly convey the stimulatory signal. DISCUSSION The current study was undertaken to define some of the G proteins involved in the short term activation of NHE activity by a prototypical cell surface receptor that couples exclusively to pertussis toxin-sensitive G proteins. Three major findings are presented within this report. First, a clear role for pertussis FIG. 3. Effects of G proteins on ECAR. A, pertussis toxin treatment (200 ng/ml ϫ 18 h) abrogates 5-HT-increased ECAR without any effect on ATP-activated ECAR (n ϭ 6 for each). Pertussis toxin reduced the basal ECAR by about 50% (not shown). B, ␤␥-sequestering reagent ␤1-CT does not affect 5-HT-activated ECAR, but markedly decreases the ability of 5-HT to activate MAPK (390 Ϯ 41% versus 20 Ϯ 20% increase; n ϭ at least 6 for each) (C). D, expression of various pertussis toxin-resistant G proteins proteins (G z␣ , G o␣ PT, G i␣1 PT, G i␣2 PT, and G i␣3 PT) and empty plasmid (pCDNA3) on P960 immunoreactivity (striped bars) and on high affinity agonist binding in pertussis toxintreated cells (speckled bars) (n ϭ 2 in duplicate for both assays). E, effects of expression of various pertussis-resistant G proteins and empty plasmid on ability of 5-HT to increase ECAR after pertussis toxin treatment). Pertussis toxin reduced the basal ECAR by about 50% in each condition (not shown). All of these experiments were performed at least five times in duplicate, with the exception of those with pCDNA3, which were performed three times. toxin-sensitive G proteins in the activation of NHE in CHO fibroblast cells was established. In the current study, a transfected receptor, which couples primarily or exclusively to G i/o/z proteins, was shown to activate ECAR. The response was due to the transfected 5-HT 1A receptor because (i) it was not present in nontransfected cells, (ii) it was initiated by a specific receptor agonist, and (iii) was blocked by a specific receptor antagonist. Furthermore, the increase in ECAR was mainly due to activation of NHE based on the following evidence: (i) dependence on [Na ϩ o ] with a half-maximal effect which approximates that described previously for NHE proteins in various systems; (ii) activation of proton efflux by hypertonic exposure, growth factors, calcium ionophore, and phorbol ester; and (iii) inhibition by amiloride analogs. The stimulation of NHE was not secondary to metabolic production of protons because (i) 5-HT did not decrease pH i , (ii) the stimulation occurred when glycolytic substrate was removed from the perfusate, and (iii) the stimulation occurred when pH i was fixed at Ϸ6.6 and 7.0 with the K ϩ /H ϩ ionophore, nigericin.
The second major finding was an apparent lack of involvement of G protein ␤␥-subunits in the signaling pathway connecting the G i -coupled receptor to activation of NHE. This conclusion was based on the inability of a G protein ␤␥-sequestering reagent (␤1-CT) to inhibit NHE activation despite effectively blocking activation of MAPK in the same cells. The third finding was that very specific members of the G i/o/z ␣-subunit family, namely G i␣2 and G i␣3 , convey the stimulatory signal to NHE. Each of these findings is highly significant in light of previous work.
The involvement of G i/o/z proteins in acute agonist-regulated NHE activity has been controversial. On the one hand, although many receptors that classically couple to G i/o/z proteins are capable of regulating NHE activity, this effect has most frequently been shown to be insensitive to pertussis toxin (5)(6)(7)(8)(9). Additionally, studies using constitutively activated G proteins have not supported a role for pertussis toxin-sensitive G proteins in NHE regulation (10,11). On the other hand, some more recent studies have indicated a role for pertussis toxin-sensitive G proteins in the activation of NHE by various agonists (13)(14)(15)(16). What, then, might account for the wide variety of findings? The differences likely involve factors related to the specific cell types and the different protocols used to measure NHE activity in those studies. Clearly, the complement of G proteins expressed within a cell might directly influence whether a given receptor modulates NHE activity and whether that modulation occurs in a positive or negative direction. Alternatively, different types of NHE with distinct regulatory characteristics might be expressed in the various cell types. As pointed out by Azarani et al. (46), many aspects of the experimental protocols could influence the observed results. Cell attachment has been identified as a major factor in the regulation of pH i and responsiveness to regulation of NHE activity (44). Most of the previous studies were performed in nonadherent cells shortly after being detached by calcium chelation or enzymatic digestion. In the current study, and in several others that used a microphysiometer (8,13,14), cells were studied in relatively undisturbed monolayers. An additional difference is that most of the prior studies used fluorescent dyes and measured the rate of recovery of cells after an imposed acid load, whereas neither dye nor acid loading is required for microphysiometric measurement. In a real sense, because microphysiometry allowed the preservation of cell to cell contact and the examination of NHE activity absent any acid loading protocols, it may more accurately preserve features that mimic agonist exposure of cells within functioning organs.
What of the differences between the current studies and those that utilized activated G protein mutants? Prior studies, which used constitutively active versions of the G proteins, would be expected to exert their effects over long periods of time (hours), whereas in the current studies, short term (minutes) effects were examined. Moreover, those studies examined the rate of recovery from an acid load absent any agonist stimulation, whereas the current studies examined the effects of brief exposure to agonists in nonacid-loaded cells.
The finding that ␤␥-subunits of G i proteins are not critically involved in conveying the stimulatory signal from the 5-HT 1A receptor dissociates the regulation of the growth-associated NHE activity (␣-subunit-mediated) from regulation of MAPK activity (␤␥-subunits) at a very proximal location in the signaling pathways. The lack of involvement of ␤␥ in activating NHE was also somewhat surprising in light of previously published data. Busch et al. (47) demonstrated that transducin ␤␥-subunits activate an endogenous NHE activity when microinjected into oocytes derived from Xenopus laevis, whereas microinjection of transducin ␣-subunits, which are closely related to G i␣ subunits, had no effect (47). That finding led us to study the role of ␣-subunits of specific members of the G i/o/z to which the 5-HT 1A receptor is known to couple. The members of the G i/o/z family showed a high degree of specificity for conveying the signal from the receptor to activation of NHE, despite the fact that the receptor has been shown to activate all of the subtypes studied here. Only G i␣2 and G i␣3 , and not G z␣ , G o␣ , or G i␣1 , were shown to be involved. Those differences cannot be accounted for by varying degrees of efficiency of expression or transfection of the various G proteins, because transfection efficiency was very high, and because increased expression of all of the target G proteins was significantly increased as determined by immunoblot. Moreover, all of the G proteins were capable of coupling to the 5-HT 1A receptor as assessed by high affinity binding, albeit G z␣ much less so than the other constructs.
The current studies are remarkable in that they demonstrate a high degree of specificity for activation of NHE among the members of the G i/o/z family of G proteins (when activated by a single receptor type), whereas the degree of specificity of those same G proteins to couple to the 5-HT 1A receptor appears to be much less constrained. G i␣2 and G i␣3 are the two most abundant G i/o/z family proteins in CHO cells, being expressed at 4.8 and 0.6 pmol/mg of membrane protein, respectively (48), and it is interesting to note that their pertussis toxin-resistant mutants were best able to reconstitute the activation of ECAR. Taken together, those results support a primary role for endogenous G i␣2 and G i␣3 in CHO cells in activating ECAR through the 5-HT 1A receptor.
Should we interpret these findings as evidence that only G i␣2 and G i␣3 , but not G z␣ , G o␣ , or G i␣1 , are capable of short term activation of NHE? We think not. It is important to understand that membrane preparations were used in the previous studies in which the 5-HT 1A receptor was shown to activate all five of the G proteins (26, 28 -30), as well as in the current study (Fig.  3D). By creating broken cells, those protocols may have removed important physical or functional constraints (such as cytoskeleton or membrane fences) that would have normally prevented the receptors from freely interacting with the entire array of G proteins contained within the cell (49). For example, one study showed that tubulin binds specifically to G s␣ and G i␣1 , but not to G o␣ , G i␣2 , or G i␣3 (50), thus providing one possible selective means of sequestering G i␣1 away from the 5-HT 1A receptor. Rather than proving that G z␣ , G o␣ , and G i␣1 are not capable of regulating NHE, our findings in intact cells might suggest that similar subtle constraints prevented the receptors from effectively interacting in situ with each or all of G z␣ , G o␣ , and G i␣1 . Clearly, further studies will be needed to resolve this important issue. It is critical to stress that the current studies do not definitively rule out a role for G z␣ , G o␣ , and G i␣1 in the regulation of NHE activity, and that these results should not be generalized to other cell types or receptors.
In summary, the current work provides evidence that a fibroblast Na ϩ /H ϩ exchange activity, putatively the ubiquitously expressed NHE-1, can be rapidly stimulated through the transfected human 5-HT 1A receptor. The activation pathway involves pertussis toxin-sensitive G protein ␣-subunits G i␣2 and G i␣3 , but not ␤␥-subunits or G z␣ , G o␣ , or G i␣1 , this despite the fact the receptor can couple efficiently to each of those subunits.