Energy dependence and functional reconstitution of the gamma-aminobutyric acid carrier from synaptic vesicles.

The energy dependence of gamma-aminobutyric acid (GABA) uptake was characterized in rat brain synaptic vesicles and in proteoliposomes reconstituted with a new procedure from vesicular detergent extracts. The proteoliposomes displayed high ATP-dependent GABA uptake activity with properties virtually identical to those of intact vesicles. GABA uptake was similar at chloride concentrations of 0 and 150 mM, i.e. conditions under which either the membrane potential (delta psi) or the pH difference (delta pH) predominates. Delta psi was gradually dissipated by increasing the concentration of SCN-. GABA uptake was reduced by 10 mM SCN-, showing less sensitivity to delta psi reduction than glutamate uptake but more than dopamine uptake. Dissipation of delta pH with NH+4 abolished GABA uptake at pH 7.3, whereas no significant inhibition occurred at pH 6.5. In contrast, dopamine uptake was inhibited more strongly, even at pH 6.5, and glutamate uptake was not reduced in either condition. We conclude that GABA uptake is driven by both components of the proton electrochemical gradient, delta pH and delta psi, and that this is different from the uptake of both dopamine and glutamate, which is more strongly dependent on delta pH and delta psi, respectively. Thus, our data suggest that GABA uptake is electrogenic and occurs in exchange for protons.


Energy Dependence and Functional Reconstitution of the y-Aminobutyric
Acid Carrier from Synaptic Vesicles* (Received for publication, July 10,1989) Johannes W. Hells, Peter R. Maycox, and Reinhard Jahn From the Department of ~eurochem~t~, fax-P~nck Institute for Psych~ut~, The energy dependence of y-aminobutyric acid (GABA) uptake was characterized in rat brain synaptic vesicles and in proteoliposomes reconstituted with a new procedure from vesicular detergent extracts. The proteoliposomes displayed high ATP-dependent GABA uptake activity with properties virtually identical to those of intact vesicles.
GABA uptake was similar at chloride concentrations of 0 and 150 mM, i.e. conditions under which either the membrane potential (A*) or the pH difference (ApH) predominates. A9 was gradually dissipated by increasing the concentration of SCN-. GABA uptake was reduced by 10 mM SCN-, showing less sensitivity to A9 reduction than glutamate uptake but more than dopamine uptake. Dissipation of ApH with NH4+ abolished GABA uptake at pH 7.3, whereas no significant inhibition occurred at pH 6.5. In contrast, dopamine uptake was inhibited more strongly, even at pH 6.5, and glutamate uptake was not reduced in either condition. We conclude that GABA uptake is driven by both components of the proton electrochemical gradient, ApH and A@, and that this is different from the uptake of both dopamine and glutamate, which is more strongly dependent on ApH and Atk, respectively.
Thus, our data suggest that GABA uptake is electrogenic and occurs in exchange for protons.
Synaptic vesicles are specialized secretory organelles that store neurotransmitters in a concentrated form within the neuron. Upon stimulation, they release their transmitter content by exocytosis. After membrane retrieval, they are reloaded with their respective transmitter by specific carrier systems and enter another round of exo-endocytotic membrane cycling (for review, see Ceccarelli and Hurlbut, 1980;Reichardt and Kelly, 1983;De Camilli and Jahn, 1990).
In recent years, evidence has accumulated that synaptic vesicles possess specific carriers for monoamines (Maron et al., 1979; for review, see Njus et al., 1986;Johnson, 1988), acetylcholine (Anderson et al., 1982), glutamate Ueda, 1983, 1985;Maycox et al., 1988), GABA (Fykse and Fonnum, 1988;Hell et al., 1988), and glycine (Kish et al.,19&g), which are thought to be localized on different vesicle populations (Fischer-Bovenkerk et al., 1988, Kish et al., 1989 The kinetic properties, substrate specificities, inhibition profiles, and energy dependence of these carriers are clearly different from those of the plasma membrane transporters (Kanner and Schuldiner, 1987). Transmitter uptake by synaptic vesicles is ATP dependent and sensitive to uncouplers. This indicates that all vesicular carriers are driven by a proton electrochemical potential (Apg).' This gradient is generated by an H+-ATPase of the vacuolar type which is present in the vesicle membrane Cidon and Sihra, 1989).
A detailed analysis of the energy dependence has been performed only for the monoamine carrier (Njus et al., 1986;Johnson, 1988) and to some extent for the glutamate carrier . Using chromaffin granules as a model system, the monoamine carrier has been shown to be driven by both components of the electrochemical gradient ASH+, the membrane potential (A%) and the pH gradient (ApH), the latter being more effective (for review, see Njus et al., 1986;Kanner and Schuldiner, 1987;Johnson, 1988). In contrast, glutamate uptake by synaptic vesicles is dependent predominantly on A* with little uptake occurring in the presence of high ApH .
In the present study, the relative dependence of vesicular GABA uptake on A* and ApH was investigated. Two approaches were used to manipulate A'@ and ApH. First, the chloride concentration was varied. The proton pump forms a large A\k in the absence of permeant anions, e.g. Cl-, which can provide charge balance. An increase in the Cl-concentration results in a gradual shift from AhJ! to ApH while the total electrochemical potential ASH+ remains unchanged Van Dyke, 1988;Maycox et al., 1988). Second, the A* or ApH was selectively dissipated using SCN-and NH: ions, respectively . In all experiments, GABA uptake activity was correlated with the relative size of ApH and A\k, which was monitored using acridine orange and oxonol VI as indicator dyes, respectively. In addition, the uptake activities of glutamate and dopamine were measured to allow a comparison of all three carriers under identical experimental conditions. Synaptic vesicle preparations from the mammalian central nervous system are probably heterogeneous, with only a fraction being GABAergic. It cannot be excluded that changes in A* and ApH are specific for GABAergic vesicles and are different from those of the total vesicle population. Therefore, parallel experiments were performed using proteoliposomes that were prepared from detergent extracts of synaptic vesicles. Since, in this preparation, all protein components are '  for 10 min, and the supernatant was chromatographed on a controlled pore glass bead column as described previously . For most experiments, fraction P, was used since no significant differences were observed in comparison with controlled pore glass beadpurified vesicles. All vesicle fractions were stored at -70 "C without loss of activity.
When vesicles were prepared at pH 6.5, the HEPES buffer was replaced with 10 mM MES-KOH in all fractionation steps. Vesicle protein was determined according to Bradford (1976 mM). After 5, 6, or 10 min at 32 "C for glutamate, GABA, or dopamine uptake, respectively, 3 ml of ice-cold standard assay buffer was added followed by immediate filtration through nitrocellulose filters (0.45-pm pore size). An extension of these incubation times did not lead to an increase of uptake for any of the transmitters.
For measurements performed at pH 6.5, HEPES was replaced by 10 mM MES-KOH in the standard assay buffer. The filters were washed four times with 3 ml of ice-cold assay buffer, and bound radioactivity was determined by liquid scintillation counting.
Filter blanks as shown in Table I   . Vesicle proteins were solubilized with sodium cholate, cleared from insoluble material by ultracentrifugation, and reconstituted by dilution after addition of solubilized brain phospholipids. Fig. 1  the time course of ATP-dependent GABA uptake in proteoliposomes. The kinetics of uptake was similar to that of intact synaptic vesicles (not shown; see also Fykse and Fonnum, 1988). Uptake was almost complete after 10 min, and therefore this was the chosen incubation period in all further experiments involving proteoliposomes. The uptake activity recovered after reconstitution was higher than in the intact synaptic vesicles used as starting material (Table I). As in intact synaptic vesicles, uptake in proteoliposomes was inhibited by the uncoupler FCCP. Together, these data show that the carrier was reconstituted in an active form and that all components required for uptake are constituents of the synaptic vesicle membrane. Chloride Dependence of GABA Uptake-The Cl-concentration was varied in order to study the relative dependence of the vesicular GABA uptake on ApH and A9. Maximal uptake occurred in a concentration range of 4-50 mM Cl- (Fig. 2). When no Cl-was present, GABA uptake was reduced by about 40% of maximal activity. A similar reduction was observed at a Cl-concentration of 150 mM. To study the effect of Cl-in more detail, changes of A3 and ApH were monitored at three Cl-concentrations (0,4, and 150 mM) and correlated with GABA uptake. Parallel experiments were performed with proteoliposomes (4 and 150 mM Cl-; since the reconstitution procedure involves Cl-, reconstituted vesicles could not be tested under Cl-free conditions). In addition, GABA uptake was compared with that of glutamate (which has been shown to be solely dependent on A*; Maycox et al., 1988), and dopamine (which is preferentially driven by ApH; Njus et al., 1986;Johnson, 1988).
In the absence of Cl-, A* was maximal in intact vesicles (Fig. 3a). In addition, a small preexisting ApH was present (Fig. 4a). At 4 mM Cl-, a slight ATP-dependent acidification was observed (Fig. 4a), which was associated with a significant reduction of A\k (Fig. 3a). At 150 mM Cl-, ApH was maximal, whereas A\k was barely detectable (Figs. 3a and 4a). Similar traces were obtained when proteoliposomes were used (Figs. 3b and 4b). GABA uptake was observed at 0 and 150 mM Cl-, suggesting that both ApH and A\k can act as the driving force. A comparison with the uptake of glutamate and dopamine (Fig. 5) revealed differences in the Cl-dependence of the three carrier systems. Dopamine uptake was observed at 0 and 4 mM Cl-but displayed 2-fold higher activity at 150 mM, which is in agreement with its preference for ApH as the driving force. In contrast, glutamate uptake was maximal at 4 mM Cl-and was strongly reduced at 150 mM, which is in agreement with its dependence on A*. The reduced activity in the absence of chloride probably reflects a direct involvement of Cl-in the glutamate uptake process. These findings are in agreement with earlier reports Naito and Ueda, 1985;Maycox et al., 1988).
Influence of NH:-The experiments described above suggest that GABA uptake is active under conditions in which either ApH or A* predominates. In order to support this view further, we analyzed the effects of reagents that are commonly used to dissipate ApH or A\k selectively. These experiments were performed at a Cl-concentration of 4 mM, which is close to the intracellular concentration of this ion within mammalian neurons (Hansen, 1985).
NH: was used to dissipate ApH since it equilibrates the intravesicular pH with that of the incubation medium. As shown in Fig. 4, 20 mM NH: is sufficient to destroy ApH completely at 4 mM Cl-in intact and reconstituted vesicles. Further additions of NH: had no effect. At 150 mM Cl-, a slight residual pH difference was observed which was dissipated by a further addition of 20 mM NH:. At 4 mM Cl-, 20 mM NH: induced an increase in the membrane potential which was similar in both intact and reconstituted vesicles (Fig. 6, a and b, left). This corresponds with the complete dissipation of ApH indicating that A~LH~, the sum of ApH and A\k, remains constant. Virtually identical traces were obtained at pH 6.5 instead of pH 7.3 (data not shown).
The uptake of GABA was almost completely inhibited by 20 mM NH: at pH 7.3 in both intact and reconstituted vesicles (Fig. 7, left), which parallels the dissipation of ApH (cf. Fig.  4). In contrast, no inhibition by NH: was observed at pH 6.5 (Fig. 7, right). This indicates that a rise in the intravesicular proton concentration (at pH 6.5, it is g-fold higher than at pH 7.3) is sufficient to allow GABA uptake to proceed in the complete absence of ApH. Under these conditions, it is solely driven by A*.
A comparison of GABA uptake with that of glutamate and dopamine confirmed the different dependence of the three carriers on ApH and A\k. Addition of NH: increased glutamate uptake at pH 7.3 (Fig. 7, left). This parallels the slight increase in the membrane potential (Fig. 6, left) and confirms the strict A\k dependence of the glutamate carrier. In contrast, dopamine uptake was strongly reduced by NH: at both pH 7.3 and pH 6.5, which is in accordance with the preference of the monoamine carrier for ApH.
Influence of XX---In the following experiments, the lipophilic anion SCN-was used to dissipate A\k. SCN-penetrates membranes easily, thereby neutralizing the positive charge within the vesicle. SCN-can act as a chaotrope at higher concentrations. It is known to affect hydrophobic interactions and has adverse effects on membrane protein function. Therefore, the SCN-concentration was titrated to analyze its effect on A* in detail. As shown in Fig. 6, a and GABA uptake in this and the following experiments was normalized to that observed at 4 mM Cl-. Acidification was monitored by following the absorbance of acridine orange at pH 7.3. When indicated, a final concentration of 20 mM NH: was added from a stock solution of 3 M (NH&SOa. b, (right), the effects of increasing concentrations of SCN-on A\k in intact synaptic vesicles (Fig. 6a) were similar to those obtained in reconstituted vesicles (Fig. 6b). Two mM SCNreduced the oxonol signal by about 40%. Higher SCN-con- centrations led to a further reduction, reaching 80% at 10 mM. Fig. 8 shows the effects of SCN-on GABA uptake in intact and reconstituted vesicles. At 2 mM SCN-, GABA uptake was only slightly inhibited (less than 20% inhibition). This contrasts with glutamate uptake, which is already reduced by 50% at 2 mM SCN-, parallel to the reduction of A* (see Fig.  6, a and b, right). At 10 mM SCN-, GABA uptake is inhibited almost completely in intact vesicles and by more than 50% in reconstituted vesicles. In contrast, dopamine uptake is only slightly reduced by 10 mM SCN-(see also Apps et al., 1980). This shows that the inhibition of GABA uptake at this SCNconcentration is not due to a dissipation of the overall energy gradient A~cLH+. It further supports the view that the dopamine carrier is predominantly dependent on ApH. Together, the data indicate that GABA uptake is less sensitive to SCNthan the A@-dependent glutamate uptake but more sensitive than dopamine uptake.

DISCUSSION
In the present study, we have analyzed the energy dependence of GABA uptake by synaptic vesicles in intact and reconstituted vesicle preparations. Our reconstitution procedure involves the formation of proteoliposomes from soluble detergent extracts after addition of exogenous phospholipids. This ensured a random distribution of all protein components. Absorbance of oxonol VI was followed under standard assay conditions. FCCP and Gramicidin D (Gram I) was added at a final concentration of 10 or 20 pM, respectively.
It resulted in high activities of the endogenous proton pump as well as of the GABA carrier and allowed a precise comparison of the intact and reconstituted systems. Under all experimental conditions, the two systems exhibited similar properties with respect to AQ, ApH, and GABA uptake. These observations allow the conclusion that the subpopulation of GABAergic vesicles responds to perturbations of the energy gradient in a way similar to the whole vesicle population. Preliminary results indicate that this applies also to the glutamate carrier." Several approaches were used to modify the relative proportions of ApH and A\k. In addition to varying the chloride concentration of the assay buffer, we used SCN-and NH.$ to dissipate A* and ApH, respectively.
The concentration of SCN-or NH: was kept as low as possible to avoid adverse effects on the integrity of membrane proteins. Neither ion affected the ATPase activity of the H' pump at the maximal concentrations used (10 and 20 mM, respectively) (data not shown). However, higher concentrations of NH: and SCNled to nonspecific effects including reduction of the energy gradient and inhibition of the carrier activities. These findings emphasize the necessity for a careful control of alI parameters in experiments involving these ions. conditions, GABA uptake can be driven solely by AQ, without involvement of ApH. First, addition of NH: did not inhibit GABA uptake at pH 6.5, whereas ApH was completely abolished. Similarly, glutamate uptake, which is known to be A\k dependent, was not affected. Second, GABA uptake was only reduced by 40% in the absence of chloride, whereas ApH is barely measurable under these conditions. It could not be documented with the same stringency that ApH can act as the driving force for GABA uptake. However, it is supported by the following observations. First, GABA uptake was only reduced to 40-50% of its maximal activity at 150 mM Cl-. A\k was reduced to a far greater extent, which was reflected by a parallel inhibition of glutamate uptake (>80%). Second, GABA uptake was less sensitive to inhibition by low concentrations of SCN-than glutamate uptake (Fig.  8). Since there is still a residual amount of A?ir in both experimental conditions, it remains to be established whether ApH can drive GABA uptake in the complete absence of A*. This is currently being investigated in our laboratory. It is evident, however, that GABA uptake is strongly influenced by the intravesicular proton concentration. This is demonstrated by the response of GABA uptake to NH: at different pH values. NH: equilibrates the internal pH with that of the extravesicular medium, inhibiting GABA uptake at pH 7.3 but not at pH 6.5. ApH is completely dissipated GABA reconst. under both conditions. This indicates that the intravesicular proton concentration must be sufficiently high for GABA uptake to proceed at significant rates. These findings can be explained in two ways. First, it is possible that protonation of a side group of the GABA carrier protein(s) is required for its activation. Second, protons may be directly involved in GABA uptake by a coupled exchange mechanism. In this case, the increased activity at low pH (in the absence of ApH) reflects the affinity of the carrier to protons as co-substrate. We favor the latter interpretation, as an exchange of GABA for protons offers a plausible mechanistic explanation since both A\k and ApH can probably act as the driving force (see below). The model in Fig. 9 summarizes our results and depicts a possible mechanism of the GABA carrier. For comparison, models of the glutamate and the dopamine carriers are shown as well. As discussed above, our data can be explained most readily by a model involving a coupled exchange of GABA for protons. In this model, transport is associated with the translocation of a net positive charge out of the vesicle. This explains why an inside positive membrane potential can act as the driving force. Furthermore, it also explains why an outwardly directed proton gradient leads to GABA accumulation. In contrast, glutamate uptake does not seem to involve protons and is clearly independent of ApH. This is indicated by the observation that dissipation of ApH by NH: does not inhibit but rather activates glutamate uptake. Additionally, glutamate uptake is very similar at pH 7.3 and 6.5 in the presence of NH:, indicating that it is not influenced by the intravesicular proton concentration. It is interesting to note that glutamate uptake is strongly reduced in the absence of chloride ( Fig. 5; see also Naito and Ueda, 1985) although Ah\k is maximal under this condition (Fig. 3). We assume that a minimal concentration of Cl-is required for optimal activity of the carrier. It is possible that Cl-is directly involved in glutamate uptake, e.g. by a coupled substoichiometric exchange with glutamate. However, more evidence is needed to support this view.
In the case of the monoamine carrier, the view that one positively charged monoamine molecule is exchanged for two protons has received strong support ( Fig. 9; Knoth et al., 1981). Despite this, an alternative model involving the exchange of one uncharged monoamine molecule for one proton has been proposed (e.g. Scherman and Henry, 1981; for review see Njus et al., 1986 andJohnson, 1988). In our experiments, dopamine uptake appears to be more dependent on ApH than on A\k. It is 2-fold higher at 150 mM Cl-compared with 4 or 0 mM Cl-, hardly affected by 10 mM SCN-, and reduced by NH: even at pH 6.5. These findings are in agreement with earlier studies and support the idea of the participation of two protons in the transport cycle.
These models are still hypothetical and need further support by additional experiments. We hope that our analysis of the GABA carrier and its functional reconstitution in proteoliposomes will provide an experimental basis for its further characterization at the molecular level.