Ion coupling stoichiometry for the norepinephrine transporter in membrane vesicles from stably transfected cells.

We prepared membrane vesicles from stable LLC-PK1 cells expressing serotonin (5-HT) gamma-aminobutyric acid (GABA) and norepinephrine (NE) transporters (SERT, GAT-1, and NET). These vesicles accumulate transport substrates when the appropriate transmembrane ion gradients are imposed. For NET, accumulation of [3H]dopamine (DA) was stimulated by imposition of Na+ and Cl- gradients (out > in) and of a K+ gradient (in > out). The presence of Na+ or Cl-, even in the absence of a gradient, stimulated DA accumulation by NET, but K+ had little or no effect in the absence of a K+ gradient. Stimulation by a K+ gradient was markedly enhanced by increasing the K+ permeability with valinomycin, suggesting that net positive charge is transported together with DA. Cationic DA is likely to be the major substrate for NET, since varying pH did not affect Km. We estimated the Na+:DA stoichiometry by measuring the effect of the transmembrane Na+ gradient on peak DA accumulation. The results suggest a 1:1 cotransport of Na+ with DA. Taken together, the results suggest that NET catalyzes cotransport of one cationic substrate molecule with one Na+ ion, and one Cl- ion, and that K+ does not participate directly in the transport process.

The synaptic action of neurotransmitters released by nerve cells is terminated by a reuptake process in which the transmitters are transported back inside the nerve endings from which they were released. Recently, cDNAs encoding transporters for neurotransmitters, amino acids, and other substrates have been cloned. These include ␥-aminobutyric acid (GABA) 1 transporters and transporters for serotonin  and the catecholamines norepinephrine (NE) and dopamine (DA) (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16). Many of these transporters share extensive sequence homology and constitute a multigene family (17)(18)(19)(20)(21)(22). In addition to the similarity in primary sequence, these transporters share a functional dependence on Na ϩ and Cl Ϫ (23)(24)(25)(26)(27). In some cases, this dependence has been shown to reflect the cotransport of Na ϩ and Cl Ϫ with the neurotransmitter substrate (28 -30). This cotransport leads to a coupling of the downhill influx of Na ϩ and Cl Ϫ with the uphill influx of substrate and leads to the internal accumulation of substrate to concentrations hundreds of times higher than in the medium.
In addition to the difference in substrate specificities be-tween members of this family, there are also differences in ion coupling mechanisms. The 5-HT and GABA transporters (SERT and GAT-1) have been studied extensively in membrane vesicle preparations from platelets and synaptosomes, respectively (31,32). These studies reveal that SERT couples transport of its cationic substrate with one Na ϩ ion while GAT-1 mediates zwitterionic GABA transport together with two Na ϩ ions. Both transporters apparently use a single Cl Ϫ ion, but SERT couples 5-HT influx to efflux of one K ϩ , while GAT-1 mediated transport does not directly involve K ϩ . The net result is that 5-HT transport by SERT is an electrically neutral process, and that GABA influx generates a net inward current. It has become apparent recently that in addition to the coupled flux of Na ϩ , K ϩ , and Cl Ϫ the 5-HT, GABA, and NE transporters also catalyze uncoupled ion flux (33)(34)(35). This uncoupled flux is likely to represent rare events in which the transporter transiently behaves like an ion channel (36). In studies that attempt to measure solute transport by recording the electrical current associated with transport, the uncoupled ion flux can confound determination of transport stoichiometry and electrogenicity. For example, even though 5-HT transport is electroneutral, addition of 5-HT to oocytes expressing SERT leads to an inward current (33). Substrate accumulation, in contrast, depends on imposed ion gradients and electrical potentials but is not likely to be influenced by uncoupled currents carried by the transporter. Thus, flux studies performed with membrane vesicle preparations remain the experimental system of choice for determination of ion coupling stoichiometry. In this paper we use the term "stoichiometry" to refer only to the substrates and ions transported in the carrier cycle and not the variable number of ions translocated during uncoupled ion flux.
Other members of the NaCl-coupled transporter family have not been studied as thoroughly as SERT and GAT-1, partly due to the lack of suitable membrane vesicle preparations. One exception is the NE transporter (NET), which has been studied in membrane vesicles prepared from PC12 cells and placental syncytiotrophoblast (37,38). Harder and Bonisch (37) concluded that NE transport into PC12 vesicles was coupled to Na ϩ and Cl Ϫ , and was electrogenic, but they failed to arrive at a definitive coupling stoichiometry because of uncertainties about the role of K ϩ . According to their analysis, stimulation of NE influx by internal K ϩ resulted either from direct K ϩ countertransport as occurs with SERT (39), or from a K ϩ diffusion potential which drives electrogenic NE influx, as with GAT-1 (26). Ganapathy and co-workers (38) studied NET mediated transport of both NE and DA into placental membrane vesicles (both catecholamines are substrates for the cloned transporter (27)). They reached a similar conclusion regarding ion coupling, but also were left with some ambiguity regarding K ϩ . In fact, the effects of ions on NET-mediated DA accumulation were similar to those observed with SERT-mediated 5-HT transport in the same membranes and the two activities were distin-guished only by their inhibitor sensitivities (40). Part of the difficulty in interpreting and comparing these data stems from the fact that they were obtained in different cells, with unknown, and potentially very different conductances to K ϩ .
Two further problems make it difficult to interpret existing data on NET ion coupling. Both previous studies assumed that the cationic form of the catecholamine substrate was transported (37,38). However, both cationic and neutral forms of catecholamine substrates are present at physiological pH, and there is no previous evidence indicating that one or the other is the true substrate. In the case of the vesicular monoamine transporter, the ionic form of the substrate is a matter of some controversy (41)(42)(43). Furthermore, the number of Na ϩ ions cotransported with substrate was estimated from the dependence of transport rate on Na ϩ concentration (37,38). This method is capable of detecting the involvement of multiple Na ϩ ions only if they have similar binding affinity and kinetics. If two Na ϩ ions (for example) with widely different affinities or binding kinetics are cotransported, the dependence of transport rate on Na ϩ may reflect only binding of the lowest affinity or most slowly associating Na ϩ ion.
We have recently established LLC-PK 1 cell lines stably expressing the biogenic amine transporters SERT, NET, and DAT as well as the GABA transporter GAT-1. Using these cell lines, we have characterized and compared the transporters under the same conditions and in the same cellular environment (27). One attractive advantage of LLC-PK 1 cells is that it has been possible to prepare plasma membrane vesicles that are suitable for transport studies (44). We took advantage of this property to prepare membrane vesicles containing transporters for GABA, 5-HT, and NE, all in the same LLC-PK 1 background. These vesicles should have identical composition except for the heterologously expressed transporter. Moreover, these vesicles are suitable for estimating equilibrium substrate accumulation in response to imposed ion gradients. In this paper, we describe experiments that define the ion coupling stoichiometry for NET using the known stoichiometries for GAT-1 and SERT mediated transport as internal controls.

EXPERIMENTAL PROCEDURES
Materials-The parental LLC-PK 1 cells and cells expressing GABA transporter (LLC-GAT) were gifts from Dr. Michael Caplan, Department of Cell and Molecular Physiology, Yale University. The cDNA encoding the human NE transporter was supplied graciously by Dr. Susan Amara, Vollum Institute, Portland, OR. SERT cDNA, coding for the rat 5-HT transporter, was contributed kindly by Dr. Beth Hoffman, NIMH, Bethesda, MD. GAT-1 cDNA, encoding a rat brain GABA transporter, was donated generously by Dr. Baruch Kanner, Hebrew University, Israel. The generation and characterization of LLC-PK 1 cells expressing human NE transporter (LLC-NET) and the rat 5-HT transporter (LLC-SERT) were described previously (27). Cell Culture-The parental LLC-PK 1 cells were maintained in a modified Eagle's medium supplemented with 10% fetal bovine serum and 2 mM L-glutamine at 37°C, 5% CO 2 . The transfected cell lines were maintained in the same medium except that G418 (Life Technologies, Inc., Gaithersberg, MD) was added at a concentration of 1.8 g/liter.
Membrane Vesicle Preparation-Cells were grown to confluence as monolayers on 15-cm diameter tissue culture dishes. Following one wash with 10 ml of vesicle preparation buffer (VPB: 200 mM mannitol, 80 mM HEPES), adjusted to pH 7.4 with LiOH, 1 g/ml pepstatin, 17 g/ml phenylmethylsulfonyl fluoride, and 5 mM oxidized form of glutathione), cells were harvested in the same buffer (10 ml/dish) by scraping and collected at 2300 ϫ g for 10 min. The cells then were suspended in VPB (5 ml/dish) and homogenized with a tight-fitting Teflon homogenizer for 40 strokes. The nuclei were removed by a 10-min low-speed centrifugation (2300 ϫ g), and the membrane vesicles were collected by centrifugation at 48,000 ϫ g for 15 min. The vesicle pellets then were suspended in VPB, frozen quickly in 0.1-ml portions in liquid nitrogen and stored in a Ϫ80°C freezer. The whole process was performed at 0 -4°C.
Substrates and Buffers-Previous results indicated that DA influx into LLC-NET cells has a lower K m and higher V max than NE influx (27). For transport into NET vesicles, 50 M L-ascorbic acid was added to [ 3 H]DA solutions to stabilize the amine substrate. This concentration of ascorbate did not interfere with substrate uptake by LLC-NET cells.  Table I were used as described in each figure legend. Li ϩ was used to replace Na ϩ or K ϩ , and isethionate or SO 4 2Ϫ were used to replace Cl Ϫ in experiments where the concentration of Na ϩ , K ϩ , or Cl Ϫ was varied. We performed control experiments to make sure these ions do not interfere with the transport processes. We observed that addition of up to 50 mM of these ions (in addition to the normal components of the transport assay mixture) neither inhibited nor stimulated the activities of the transporters (data not shown).
Substrate Transport into Membrane Vesicles-Frozen membrane vesicles were thawed at room temperature, diluted at least 20-fold into internal buffer, and incubated at 37°C for 15 min. Following centrifugation at 48,000 ϫ g for 20 min at 4°C, the vesicle pellets were resuspended in the same internal buffer and incubated on ice for 10 -30 min. Ten l of the equilibrated vesicles (about 1.0 mg of protein) then were diluted into 1 ml of external buffer containing approximately 40 nM 3 H-labeled substrate and incubated for a given time at 22°C. Independent experiments (not shown) demonstrated no significant difference in transport properties if the incubations were performed at 37°C. Previous experience suggested that ion gradients would dissipate less rapidly at the lower temperature. The internal and external buffers are described in each figure legend. Transport was stopped by addition of 3 ml of ice-cold 0.2 M NaCl, and the suspensions were filtered immediately through a GN6 filter (Gelman Sciences, Ann Arbor, MI) followed by two washes of 3 ml of ice-cold 0.2 M NaCl. The filters were then counted in 3 ml of Optifluor scintillation fluid (Packard Instrument Co.) in a Beckman LS-3801 liquid scintillation counter.
Substrate Transport into Whole Cells-Cells were grown in 48-well plates at 37°C until they were confluent (or at least 48 h after plating). Following one wash with 1 ml of phosphate-buffered saline supplemented with 1 mM MgCl 2 and 0.1 mM CaCl 2 (phosphate-buffered saline/ magnesium/calcium), cells in each well were incubated in 0.15 ml of phosphate-buffered saline/magnesium/calcium containing about 40 nM 3 H-labeled substrate and other reagents (as described in each figure legend) for the indicated time at 22°C. At the end of the incubation, cells were washed three times with 0.5 ml of ice-cold phosphate-buffered saline and then dissolved in 0.25 ml of 1% SDS solution. The amount of substrate accumulated in the cells was determined by counting the solubilized cells in 3 ml of Optifluor scintillation fluid.
Data Analysis-Nonlinear regression fits of experimental and calculated data were performed with Origin (MicroCal Software, Northampton, MA), which uses the Marquardt-Levenberg nonlinear least squares curve fitting algorithm. The statistical analysis was done with data from single experiments. All the experiments were repeated a total of at least three times and in all cases gave essentially the same results. Within each experiment, data points are the averages of triplicates, and standard deviations are represented by error bars.

Transport by Membrane Vesicles Prepared from Stably
Transfected Cells-The generation and characterization of stable cell lines expressing NET, GAT, and SERT were described previously (27,45). When membrane vesicles prepared from these cell lines were equilibrated in a buffer composed predominantly of K 2 SO 4 (buffer B), and diluted into NaCl (buffer A) containing an appropriate 3 H-labeled substrate, the vesicles rapidly accumulated the labeled substrate to high internal concentrations. Fig. 1 shows time courses of GABA, 5-HT, and DA accumulation into vesicles prepared from LLC-GAT, LLC-SERT, and LLC-NET cells, respectively. Inwardly-directed Na ϩ and Cl Ϫ gradients and an outward K ϩ gradient were imposed across the vesicle membrane by dilution. These gradients slowly decay with time. Therefore, each time course consists of an accumulation phase, during which substrate enters the vesicle in response to the imposed ion gradients, and a decay phase, when the dissipating ion gradients are no longer sufficient to support the accumulated gradient of intravesicular neurotransmitter. At times close to the peak of accumulation (2-5 min after dilution), the substrate gradient is expected to be close to equilibrium with the driving forces of the imposed ion gradients. In subsequent experiments, the extent of substrate accumulation at 5 min after dilution was used to estimate the substrate gradient in equilibrium with imposed ion gradients. Transport Stimulation by Na ϩ , Cl Ϫ , and K ϩ Gradients-To learn more about the way that ion gradients drive substrate accumulation into NET vesicles, we measured the effect of Na ϩ , Cl Ϫ , and K ϩ gradients on transport. In the time courses shown in Fig. 1, Na ϩ , Cl Ϫ , and K ϩ gradients were imposed across the vesicle membrane. To examine which of these ion gradients contributed to substrate transport, we imposed these gradients separately and in combination, and measured their influence on peak substrate accumulation by each transporter. Since influx rates, which are sensitive to external ion composition, can affect the time at which peak accumulation occurs, we kept the external NaCl and internal K ϩ constant wherever possible and varied internal Na ϩ and Cl Ϫ and external K ϩ to alter the imposed ion gradients. In this way, each ion was present at a concentration sufficient to support maximal transport rates. Accumulation was maximal for GAT, SERT, and NET when all three gradients (the inward Na ϩ and Cl Ϫ gradients and the outward K ϩ gradient) were imposed (Fig. 2, column 1). At the other extreme, uptake was minimal when the same medium was present internally and externally (column 8).
For Na ϩ , Cl Ϫ , and K ϩ , we tested whether each ion was required for transport and if a gradient of that ion stimulated accumulation. The results for NET, GAT-1, and SERT are presented in columns 2-7 of Fig. 2. Adding K ϩ to the external buffer to eliminate the K ϩ gradient decreased accumulation by NET vesicles but had a smaller effect on SERT vesicles and essentially no effect on GAT-1 vesicles (compare column 2 to column 1). SERT is known to countertransport K ϩ with 5-HT (39), but such a direct interaction between K ϩ and NET had not been demonstrated in previous studies (37,40). Removing K ϩ entirely from both internal and external media had no further effect on NET vesicles, and little effect on GAT-1 vesicles, but significantly decreased uptake by SERT vesicles. The decreased uptake by SERT vesicles is consistent with the known participation of K ϩ in 5-HT transport, and the lack of GAT-1 inhibition by K ϩ removal also was consistent with evidence that K ϩ does not participate directly in that reaction (46) (see Ref. 47). The observation that K ϩ removal does not affect DA accumulation strongly suggests that K ϩ is not coupled directly to NET-mediated catecholamine transport.
Column 4 of Fig. 2 shows the results obtained when the Cl Ϫ gradient was eliminated by preloading LLC-GAT, LLC-SERT, and LLC-NET vesicles with buffer E, containing Cl Ϫ . For each transporter, there is a dramatic decrease in the amount of  Table I To test whether K ϩ is required by the transporters, substrate accumulation in 5 min was measured when inward Na ϩ and Cl Ϫ gradients were imposed across the vesicle membranes with K ϩ absent (column 3) or present on both sides of the membrane at equal concentration (column 2). Similarly the requirements of Cl Ϫ and Na ϩ were tested with an inward Na ϩ gradient and an outward K ϩ gradient imposed with Cl Ϫ absent (column 5) or present on both sides of the membrane at equal concentration (column 4), and with inward Cl Ϫ gradient and outward K ϩ gradient imposed with Na ϩ absent (column 7) or present on both sides of the membrane at equal concentration (column 6). Controls with all and none of the three gradients imposed are shown in columns 1 and 8, respectively. GAT, SERT, and NET represent vesicles prepared from LLC-GAT, LLC-SERT, and LLC-NET cells. The buffer conditions from left to right are (internal medium/external medium): column 1, buffer B/buffer A; column 2, buffer B/buffer D; column 3, buffer C/buffer A; column 4, buffer E/buffer A; column 5, buffer B/buffer F; column 6, buffer H/buffer A; column 7, buffer B/buffer G; column 8, buffer C/buffer C. substrate accumulated, consistent with the proposal that Cl Ϫ serves as a driving force for accumulation. Removal of Cl Ϫ entirely from both internal and external media (column 5) caused a further reduction in transport by each transporter, as expected if Cl Ϫ is directly involved in the transport reaction as a cotransported ion (28). Na ϩ cotransport is also a common theme for neurotransmitter transporters, including these three transporters (23,26,48). As shown in column 6 of Fig. 2, ablation of the Na ϩ gradient by the use of internal buffer H, which contained Na ϩ , dramatically decreased accumulation in all three cases. Removal of Na ϩ from both internal and external buffer reduced transport even further (column 7) indicating that both the presence of Na ϩ and a Na ϩ gradient stimulate accumulation, consistent with the cotransport of Na ϩ by GAT-1, SERT, and NET.
Electrogenicity of Transport-To assess the influence of a transmembrane electrical potential on transport, we added the K ϩ -specific ionophore valinomycin to vesicles in which a transmembrane K ϩ gradient (in Ͼout) had been imposed. Valinomycin-mediated K ϩ efflux generates a diffusion potential (internal negative) under these conditions. Given the 100-fold K ϩ gradient imposed across the vesicle membrane, the diffusion potential should approach Ϫ117 mV initially and then decay with the K ϩ gradient. This potential is expected to increase accumulation by any process that is coupled to net positive charge influx, to inhibit those processes coupled to net positive charge efflux, and to have no direct effect on processes that are not coupled to net charge movement. In the experiment shown in Fig. 3, we imposed inwardly directed Na ϩ and Cl Ϫ gradients and an outwardly directed K ϩ gradient to drive transport in the presence and absence of valinomycin. The figure shows that potentials generated by valinomycin addition did not stimulate SERT-mediated [ 3 H]5-HT accumulation, but significantly increased [ 3 H]GABA accumulation by GAT-1, and more importantly, [ 3 H]DA accumulation by NET. These results confirm previous conclusions that GABA but not 5-HT is transported with a positive charge (26,49). In addition, they demonstrate that [ 3 H]DA transport into membrane vesicles by NET is electrogenic, and that one or more positive charges cross the membrane together with DA in each catalytic cycle.
Na ϩ Coupling Stoichiometry of Transport-GAT-1 has been shown to cotransport two Na ϩ ions with each GABA molecule (30). In contrast, SERT cotransports a single Na ϩ ion together with 5-HT (29). This difference in Na ϩ stoichiometry is reflected both in steady-state (29,46) and kinetic (26,50) transport measurements. Kinetic measurements of NET Na ϩ dependence demonstrate a hyperbolic response consistent with involvement of a single Na ϩ ion in the process (37,38). However, multiple Na ϩ ions are involved in desipramine binding to NET (51), and kinetic measurements are less reliable than steady-state determinations for ion coupling stoichiometry (see "Discussion"). We therefore measured peak accumulation by GAT-1, SERT, and NET in LLC-PK 1 vesicles in response to varying the transmembrane Na ϩ gradient. The GABA and 5-HT transporters served as two known references in this vesicle system. We kept the external Na ϩ concentration constant and varied internal Na ϩ to reduce the transmembrane Na ϩ gradient. As shown in Fig. 4, an increase in internal Na ϩ decreases peak accumulation by each transporter, but the accumulation of [ 3 H]GABA by GAT-1 vesicles was much more sensitive to internal Na ϩ than the accumulation of [ 3 H]5-HT by SERT vesicles. This result is entirely consistent with the known Na ϩ stoichiometry of the two transporters. Fig. 4 also shows that the sensitivity of NET vesicles to increasing internal Na ϩ was quite similar to that of SERT and lower than the sensitivity of GAT-1. Together with the previous kinetic data these results strongly suggest that NET, like SERT, also cotransports one Na ϩ ion with each molecule of substrate.
The Ionic Form of DA Used as Substrate by NET-To deduce an ion coupling stoichiometry using the results presented above, it is important also to consider the ionic form of the substrate (DA) transported. The pK a values for dopamine are 8.9 for the catechol -OH group and 10.6 for the -NH 2 group. At the pH of the transport assays described above, almost all of the substrate is in the cationic form. However, as the pH increases from 6.5 to 7.5, the zwitterionic and neutral forms of DA increase from 0.4 to 4%, while the cationic form remains essentially unchanged at 99.6 and 96%, respectively. If the neutral or zwitterionic form of DA was the only substrate for NET, the apparent transport K m for DA (all species) should decrease about 10-fold as the pH rises. Fig. 5 shows the saturation of DA transport by NET at pH 6.5 and 7.5. The K m for DA was 0.32 Ϯ 0.03 M at pH 6.5 and 0.39 Ϯ 0.05 M at pH 7.5. The lack of a significant shift in K m suggests that the positively charged form of DA, rather than the neutral or zwitterionic form, is the predominant transport substrate for NET. FIG. 4. Transport dependence on Na ؉ gradient. Substrate accumulation in a 5-min incubation was measured as described under "Experimental Procedures." Inward Na ϩ and Cl Ϫ gradients and outward K ϩ gradient were imposed, and the Na ϩ gradient was varied by replacing varying amounts of Li ϩ in the internal medium with Na ϩ . The external buffer contained 210 mM NaCl, 5 mM K 2 SO 4 , and 45 mM Li 2 SO 4 . The internal buffers contained 100, 75, 50, 25, and 0%, respectively, of 100 mM Na 2 SO 4 , 50 mM K 2 SO 4 , and 10 mM NaCl with the remainder consisting of 100 mM Li 2 SO 4 , 50 mM K 2 SO 4 , and 10 mM NaCl. All above buffers also contained 10 mM lithium phosphate, pH 6.7, and 1 mM MgCl 2 . GAT, SERT, and NET represent vesicles prepared from LLC-GAT, LLC-SERT, and LLC-NET cells. DISCUSSION We have taken a novel approach to understand the ion coupling stoichiometry of the NET. We are able to express and compare, in the same parental cell line, transporters whose stoichiometry is known and also the transporter of interest. This system allows us to manipulate transmembrane ion gradients and electrical potentials directly, and to test the effects on transport. For comparison with NET we have chosen the 5-HT and GABA transporters (SERT and GAT-1) since the ion coupling stoichiometries for these two transporters are well understood and is different from one another (31,32). SERT cotransports cationic 5-HT with Na ϩ and Cl Ϫ and countertransports K ϩ . The 1:1:1:1 stoichiometry of this process results in electrically neutral 5-HT transport. In contrast, GAT-1 cotransports GABA with two Na ϩ ions and one Cl Ϫ and does not transport K ϩ . The resulting 1:2:1 stoichiometry leads to positive charge movement in the direction of GABA transport. Our results suggest that the stoichiometry of NET-mediated DA transport is different from those for both SERT and GAT-1.
The recent demonstration that neurotransmitter transporters also mediate uncoupled ion flux (33)(34)(35)52) has cast a measure of confusion on the terms used to describe these proteins and their properties. If, as it seems likely, these transporters transiently form conductive channels through the membrane, a distinction must be made between the types of electrical currents due to channel activity and substrate transport. The term "electrogenic" has traditionally been used to describe a coupled transport process in which net charge crosses the membrane. In the absence of ion gradients, an electrogenic transporter should generate an electrical potential in response to an imposed transmembrane substrate gradient. In contrast, the channel activity of such a transporter can only mediate energetically downhill ion flux. Thus, while SERT and GAT-1 both conduct ions by virtue of their intermittant channel activity, GAT-1 is an electrogenic transporter because it transports net charge with GABA and SERT is electroneutral because the 5-HT transport cycle itself does not move net charge across the membrane.
We previously described the generation of stable cell lines based on LLC-PK 1 cells that express biogenic amine transporters (27). We characterized the transporters with respect to their requirements for external Na ϩ and Cl Ϫ by varying the cells' external ion compositions. We observed that NET, as well as SERT and the DA transporter (DAT) require Na ϩ and Cl Ϫ from the outside of the cells for full activity. These experiments could not determine whether Na ϩ and Cl Ϫ were cotransported or merely required to activate the transporter. However, the response of transport rate to Na ϩ was sigmoidal for DAT and hyperbolic for NET, suggesting that a different number of Na ϩ ions participated in the two reactions. These differences had been observed previously (38,53), however, they had not been observed using transporters expressed in the same cell.
While kinetic measurements of Na ϩ dependence can suggest a cotransport stoichoimetry, their interpretation is not straightforward. For example, if two Na ϩ ions are cotransported with DA by NET, the Na ϩ dependence is expected to be sigmoidal. However, if one of those two Na ϩ ions binds much slower or more weakly than the other, the Na ϩ dependence might appear hyperbolic, giving the false impression that only one Na ϩ ion is involved in the process. Furthermore, a sigmoidal dependence on Na ϩ is expected if two Na ϩ ions are required for transport, even if only one of them actually is cotransported with substrate. To avoid these potential problems, we estimated the Na ϩ cotransport stoichiometry by varying the Na ϩ gradient and measuring the accumulation of substrate at a time when it is close to being in equilibrium with the imposed gradients.
Since the peak of substrate accumulation occurs at a time when transmembrane ion gradients are decaying, the absolute value for internal Na ϩ is higher than that initially present in the internal buffer. Rather than attempt to estimate internal Na ϩ , we have taken advantage of identical membrane vesicles from cells expressing SERT and GAT-1. At the time of peak uptake, we expect the Na ϩ gradient to have decayed to the same extent in each vesicle preparation. Transport into vesicles from cells expressing GAT-1 is much more sensitive to dissipation of the Na ϩ gradient than is transport into SERT vesicles (Fig. 4). This is expected since the Na ϩ gradient contributes more to the overall driving force for GABA accumulation than it does for 5-HT accumulation (32). Dissipating the Na ϩ gradient has a similar effect on NET and SERT vesicles. This similarity strongly argues that NET, like SERT transports with a Na ϩ :substrate coupling stoichiometry of 1:1.
Previous studies examining the stoichiometry of NE transport in membrane vesicles from PC12 cells (37) and placental syncytiotrophoblast (38) used kinetic measurements of NE or DA influx. Although those studies came to similar conclusions concerning the overall stoichiometry, there were significant uncertainties concerning the role of K ϩ . Internal K ϩ stimulated the rate of transport and the extent of accumulation. We also observed a stimulatory effect of the K ϩ gradient (in Ͼ out) on DA accumulation (Fig. 2, lower panel, columns 1 and 2). However, the results in Fig. 2, lower panel columns 2 and 3, indicate that this effect is not likely to result from an obligatory interaction of K ϩ with the transporter. In the absence of a K ϩ gradient, the presence of K ϩ had no effect on peak [ 3 H]DA accumulation by NET. In contrast, removal of K ϩ significantly impaired [ 3 H]5-HT accumulation by SERT (Fig. 2, middle  panel, columns 1 and 2), which is known to directly couple 5-HT and K ϩ countertransport (39). These results argue that the effects of K ϩ were indirect, and that K ϩ was not coupled to DA transport by NET.
Although K ϩ was not directly coupled to DA transport, an outwardly directed K ϩ gradient stimulated accumulation (Fig.  2). The most likely reason for this effect is that an endogenous K ϩ conductance in the membrane allowed K ϩ efflux to generate a membrane potential (negative inside) or to offset any potential (positive inside) generated by Na ϩ influx. Increasing the K ϩ permeability with valinomycin stimulated DA accumu- lation, as would be expected if DA influx was coupled to influx of net positive charge. This stimulation of NET-mediated transport by K ϩ or H ϩ diffusion potentials had been observed also in previous studies and had been attributed to coupled movement of substrate and positive charge (37,38).
Although it might be expected that GABA accumulation also would be stimulated by a K ϩ gradient in the absence of valinomycin, this was not observed. However, GABA accumulation was stimulated less than DA also when valinomycin was used to increase K ϩ permeability (Fig. 3). Since GAT-1 couples GABA transport with 2 Na ϩ ions, the Na ϩ gradient represents a much larger portion of the GAT-1 driving force (relative to the membrane potential) than it does for NET, which uses only one Na ϩ ion for cotransport with DA. While this would reduce the influence of a K ϩ diffusion potential on GAT-1, we cannot explain why the effect on GAT-1 of dissipating the K ϩ gradient was completely absent. Another potential explanation is that the K ϩ gradient does not, in the absence of valinomycin, induce a diffusion potential, and that the inhibition of NET results from an inhibitory effect of increased external K ϩ . At present we have no way to distinguish between these or other possibilities. Nonetheless, the lack of effect of K ϩ on NET in the absence of a K ϩ gradient argues against a direct role for K ϩ in NET-mediated DA transport.
The range of possible ion coupling stoichiometries for NET is constrained by the observations that the cationic form of DA is cotransported with one Na ϩ ion and that net positive charge accompanies these ions. Additional ion gradients that could be coupled to transport are the K ϩ gradient (in Ͼ out) and the Cl Ϫ gradient (out Ͼ in). A maximum of one cotransported Cl Ϫ ion or one countertransported K ϩ ion could be coupled to transport. One each of Cl Ϫ or K ϩ or two of either ion would force the overall process to be electroneutral. The results presented here and elsewhere (37,38) strongly suggest that transport is electrogenic, and that the presence of Cl Ϫ , as well as the Cl Ϫ gradient, are required for efficient transport by NET. In contrast, the presence of K ϩ does not stimulate transport in the absence of a K ϩ gradient. The most straightforward conclusion is that NET couples the cotransport of NE (or DA), Na ϩ , and Cl Ϫ with a 1:1:1 stoichiometry.
The three neurotransmitter transporters (GAT-1, SERT, and NET) in this gene family whose stoichiometry is now known have three different ion coupling stoichiometries. Moreover, kinetic measurements suggest that the DAT cotransports two Na ϩ ions with DA (53). Thus, even among the closely related biogenic amine transporters (SERT, NET, and DAT), ion coupling stoichiometries probably differ. Current efforts in this laboratory are directed toward understanding the stoichiometry of DAT using a similar approach to that described here. The ability to compare different transporters in identical membrane vesicle preparations has provided a new tool for understanding the types and mechanisms of ion coupling in transporters. By comparing the similarities in coupling between members of this gene family with regions of sequence similarity, it may be possible to identify structural regions of the transporters responsible for ion coupling.