The glial and the neuronal glycine transporters differ in their reactivity to sulfhydryl reagents.

The neuronal (GlyT2) and glial (GlyT1) glycine transporters, two members of the Na(+)/Cl(-)-dependent neurotransmitter transporter superfamily, differ by many aspects, such as substrate specificity and Na(+) coupling. We have characterized under voltage clamp their reactivity toward the membrane impermeant sulfhydryl reagent [2-(trimethylammonium)-ethyl]-methanethiosulfonate (MTSET). In Xenopus oocytes expressing GlyT1b, application of MTSET reduced to the same extent the Na(+)-dependent charge movement, the glycine-evoked current, and the glycine uptake, indicating a complete inactivation of the transporters following cysteine modification. In contrast, this compound had no detectable effect on the glycine uptake and the glycine-evoked current of GlyT2a. The sensitivities to MTSET of the two transporters can be permutated by suppressing a cysteine (C62A) in the first extracellular loop (EL1) of GlyT1b and introducing one at the equivalent position in GlyT2a, either by point mutation (A223C) or by swapping the EL1 sequence (GlyT1b-EL1 and GlyT2a-EL1) resulting in AFQ <--> CYR modification. Inactivation by MTSET was five times faster in GlyT2a-A223C than in GlyT2a-EL1 or GlyT1b, suggesting that the arginine in position +2 reduced the cysteine reactivity. Protection assays indicate that EL1 cysteines are less accessible in the presence of all co-transported substrates: Na(+), Cl(-), and glycine. Application of dithioerythritol reverses the inactivation by MTSET of the sensitive transporters. Together, these results indicate that EL1 conformation differs between GlyT1b and GlyT2a and is modified by substrate binding and translocation.


SUMMARY
The neuronal (GlyT2) and glial (GlyT1) glycine transporters, two members of the Na + /Cldependent neurotransmitter transporter superfamily, differ by many aspects, such as substrate specificity and Na + coupling. We have characterized under voltage-clamp their reactivity toward the membrane impermeant sulfhydryl reagent [2-(trimethylammonium)-ethyl]-methanethiosulfonate (MTSET). In Xenopus oocytes expressing GlyT1b, application of MTSET reduced to the same extent the Na +dependent charge movement, the glycine-evoked current and the glycine uptake, indicating a complete inactivation of the transporters following cysteine modification.
In contrast, this compound had no detectable effect on the glycine uptake and the glycine-evoked current of GlyT2a. The sensitivities to MTSET of the two transporters can be permutated by suppressing a cysteine (C62A) in the first extracellular loop (EL1) of GlyT1b and introducing one at the equivalent position in GlyT2a, either by point mutation (A223C) or by swapping the EL1 sequence (GlyT1b-EL1 and GlyT2a-EL1) resulting in AFQ ßà CYR modification. Inactivation by MTSET was 5-times faster in GlyT2a-A223C than in GlyT2a-EL1 or GlyT1b, suggesting that the arginine in position +2 reduced the cysteine reactivity. Protection assays indicate that EL1cysteines are less accessible in the presence of all co-transported substrates: Na + , Cland glycine. Application of dithioerythritol (DTE) reverses the inactivation by MTSET of the sensitive transporters. Together, these results indicate that EL1 conformation differs between GlyT1b and GlyT2a and is modified by substrate binding and translocation.

INTRODUCTION
Na + ,Cl --coupled transporters control the extracellular and intracellular concentrations of neurotransmitters like bioamines, γ-aminobutyric acid (GABA) and glycine. Through the thermodynamically coupled uptake of Na + , Cland the transmitter, they help terminate synaptic transmission, keeping a low extracellular concentration of neurotransmitter and limiting the spill-over of neurotransmitter between neighboring synapses. At some pre-synaptic boutons, their concentrative power may also be required for an efficient supply of neurotransmitter to the low affinity vesicular transporters.
Within this superfamily, the two glycine transporters (GlyT1 and GlyT2) have distinctive properties. A well established difference concerns their substrate specificity, as sarcosine is a substrate of GlyT1b (1) but not of GlyT2a (2). Recently, other pharmacological differences have been identified: the tricyclic antidepressant amoxapine and ethanol both inhibit selectively GlyT2a (3,4). Electrophysiological characterization of the two transporters has revealed additional features differentiating GlyT2a from GlyT1b: (i) a stoichiometry of 3 Na + / 1 Cl -/glycine compared to only 2 Na + /1 Cl -/glycine for GlyT1b (resulting in a charge to glycine flux ratio of a different charge coupling of ~2 for GlyT2a vs. 1 for GlyT1b) (5); (ii) a reduced capacity for reverse uptake (5); (iii) an uncoupled, outwardly-rectifying Clconductance (Roux M.J. and S. Supplisson, submitted); (iv) a bi-exponential transient current recorded in the absence of glycine (Roux and Supplisson, unpublished observations). The amino-acids involved in these differences between GlyT1b and GlyT2a have not been identified yet.
Sequence comparison of EL1 in the Na + , Clcoupled transporter superfamily shows that GlyT2a together with PROT (6) and ATB 0,+ (7) are the only members that lack a conserved cysteine in EL1 that has been identified in GAT-1 (C74, (8)) and SERT (C109, (9)) as a primary target for the membrane-impermeant sulfhydryl reagent MTSET. Mutation in GlyT2a first extracellular loop (EL1) induced an increase in the outwardly rectifying leak current characteristic of this transporter (Roux M.J. and S. Supplisson, submitted), which further suggested that the conformation of this loop may differ between GlyT1b and GlyT2a. To assess this issue, we tested the reactivity to sulfhydryl reagents of GlyT1b and GlyT2a.
Transporter inhibition by MTSET required prolonged application and high concentrations of this reagent, which suggests a low accessibility of the reactive cysteine either because it is partially buried inside the protein or because it is infrequently exposed to the surface (8). In fact, the orientation of EL1 has been a matter of debate (8,(10)(11)(12). The original topological model proposed for the GABA transporter GAT-1 (13) predicted cytoplasmic N -and C -termini and 12 transmembrane segments (TM1-12) linked by 6 extracellular loops (EL1-6) and 5 intracellular loops (IL1-5). In accordance to this model, most positions in EL1 were found to be accessible to extracellularly applied MTSET when mutated to cysteine in . In addition, MTSET increased dramatically the oocyte leak current (8).
We characterized under voltage-clamp the effects of MTSET and (2sulfonatoethyl) methanethiosulfonate (MTSES) on the glycine uptake, the glycineevoked current and the charge movement of the two wild-type glycine transporters, as well as on those of two point-mutants of the first amino acid of EL1, GlyT1b-C62A and GlyT2a-A223C, that exchanged the MTSET reactivity. The difference in residue number for an equivalent position in EL1 sequence is explained by the much longer N-terminal of GlyT2a (15). In addition, the GlyT2a-EL1 chimera resulting from the mutation of the first three amino acids of GlyT2a-EL1 (AFQ) to the GlyT1b sequence (CYR) was studied in detail because it influenced MTSET reactivity and had a dramatic effect on the outward Clcurrent specific of GlyT2a (Roux M.J. and S. Supplisson, submitted).

EXPERIMENTAL PROCEDURES
Site-directed mutagenesis, RNA synthesis and expression -Rat GlyT1b (gift from K. (16)) and rat GlyT2a (15) cDNAs were subcloned as indicated in (1). Point mutations were performed using the method of (17). cRNAs were transcribed in vitro using the T7 mMessage-mMachine kit (Ambion, Austin, TX) and 10-100 ng were injected into oocytes using a nanoliter injector (World Precision Instruments, Sarasota, FL). Xenopus laevis oocytes were prepared and incubated as indicated in (5). After complete wash-out and the current back to the baseline level, oocytes were transferred in 5 ml ice-cold Ringer solution and washed twice, then lysed in vials containing 500 µl SDS (2%) and measured for radioactivity. The charge coupling was determined as the ratio between the glycine-evoked current integral and the glycine uptake as described in (5).

Transfer of EL1 confers MTSET sensitivity from GlyT1b to GlyT2a
The glycine transporters GlyT1b and GlyT2a, expressed in Xenopus oocytes, showed differential sensitivities to MTSET. The glycine-evoked current (I T ) (  Table 1 for sequence comparison). With GlyT2a-A223C and GlyT2a-EL1, application of MTSET reduced I T (Fig. 1 e,g), the transient currents and the outwardly-rectifying leak conductance (Fig. 1f,h). However, the inhibition of GlyT2a-EL1 was partial, as for GlyT1b, while GlyT2a-A223C was fully inhibited under the same conditions.

MTSET and leak current
MTSET did not change the leak current in oocytes expressing either GlyT1b (Fig.   1b) or GlyT2a ( Fig. 1d) but it affected the leak current of GlyT2a-A223C and GlyT2a-EL1. Like GlyT2a, both of these constructs exhibit an outwardly-rectifying "leak" conductance at depolarized potentials, which is particularly large in the case of GlyT2a-EL1 and that is described in detail in Roux and Supplisson (submitted).
While this outward "leak" conductance was not changed by MTSET in GlyT2a (Fig.   1d), it was suppressed in both GlyT2a-A223C and GlyT2a-EL1 ( Fig. 1 f,h). In contrast, we observed after MTSET treatment a limited increase in the inward steady-state leak current at hyperpolarized potential for these two constructs (Fig.   1f,h). GlyT2a-A223C is shown in Figure 3 with the unsubstracted OFF-currents recorded during the repolarization step to the holding potential in NaCl Ringer, CholineCl Ringer and after MTSET application.

MTSET blocks the uptake current and the charge movement in the same proportion
In the potential range from -140 to +50 mV, the I T -voltage relationships of GlyT1b, GlyT2a and GlyT2a-A223C (Fig. 2b,d,g) were quasi-linear while that of GlyT2a-EL1 was inwardly rectifying (Fig. 2 f). For comparison, I T values were normalized to the Q max determined before MTSET treatment. The transport currents were reduced in the same proportion at all potentials by MTSET for GlyT1b, GlyT2a-EL1 and GlyT2a-A223C transporters, as shown in Figure 2 (b, f, h). For each oocyte, the percentage of the remaining transport current was subsequently calculated as the mean of the experimental values between -140 and 0 mV (every 10 mV). Values obtained at voltages above 0 mV were ignored because of their small amplitudes (especially for GlyT2a-EL1). The reduction in I T after MTSET treatment was comparable to the reduction observed in Q max as indicated in Table 2.

EL1 sequence influences the cysteine accessibility
To compare the accessibility of the EL1 cysteine for the various sensitive constructs, we studied the time course of the MTSET inhibition. The progressive decrease in uptake current plotted as a function of the cumulative exposure to MTSET is shown in Figure 4. The decay in current had a time constant of 3.8 min and 3.9 min for GlyT1b and GlyT2a-EL1 respectively, but was 5-fold faster with GlyT2a-A223C, with a time constant of 0.67 min. The difference in the time constants indicates that in glycine transporters, the accessibility of the EL1-cysteine to the extracellular solution and its reactivity to MTSET are strongly influenced by the next two residues that are modified in GlyT2a (FQ vs. YR in GlyT1b).
To confirm that EL1 was fully responsible for GlyT1b sensitivity to MTSET, we constructed the reverse chimera GlyT1b-EL1. For this transporter, no change in the amplitude of glycine-evoked current was observed for period of incubation up to 20 minutes (Fig. 4).

MTSET have the same effect on glycine uptake and glycine-evoked current
To verify that the reduction of the glycine-evoked current by MTSET was reflecting a comparable change in the influx rate of glycine in oocytes expressing sensitive glycine transporters, we repeated the MTSET inactivation protocol using tracer flux of glycine. A summary of these uptake experiments is shown in Figure 5A, that confirms the large reduction of the Na + -dependent glycine uptake with oocytes expressing GlyT1b, GlyT2a-EL1 and GlyT2a-A223C. No significant inhibition by MTSET are observed with GlyT2a, GlyT1b-EL1 and the point mutant GlyT1b-C62A.
Finally, we checked the charge coupling of each transporter by applying radiolabelled glycine under voltage-clamp condition as described in (5). The results shown in Figure 5B indicate that the specific charge coupling of the wild-type transporters (1 and ~2 for GlyT1b and GlyT2a respectively) are not affected by EL1 sequence modification, and that glycine-evoked current is proportional to glycine uptake.
Na + , Cland glycine limit EL1-cysteine accessibility MTSET applications (1 mM during 5 minutes at -40 mV) were performed on GlyT1b, GlyT2a-EL1 and GlyT2-A223C in the presence or absence of the transporter substrates to see if these substrates protected against MTSET. As described in Figure 6, inhibition by MTSET was reduced by the joint presence of Na + and Cl -, and was further reduced by glycine. Sarcosine, a specific substrate of GlyT1b, protected this transporter to a similar extent as glycine, but did not protect GlyT2a-EL1. In all transporters tested, Na + , Clor glycine alone did not induce any protection (Fig. 6). This is in good agreement with the fact that neither the uptake current nor the charge movement can be observed when one of the co-transported ions is absent ( (5) and unpublished observations). The best protection was observed for GlyT2a-EL1, for which a quasi-complete protection by NaCl + glycine was found.

Voltage-dependence of MTSET and MTSES reactivity on EL1-cysteine
GlyT1b oocytes were perfused with 1 mM MTSET for 5 minutes in cholineCl Ringer at holding potentials between -120 mV and 0 mV. Inhibition increased with hyperpolarization, from 62.3 ± 5.2% (n = 4) at 0 mV to 84.4 ± 0.9% (n = 4) at -120 mV (Fig. 7A). To test if this voltage-dependence was due to a voltage-dependent transition of the transporter in the absence of Na + or to the positive charge of MTSET, we repeated the experiments using the negatively charged MTSES. Figure   7A shows that MTSES inhibition of GlyT1b had the same sign of voltage-dependence as MTSET, with the inhibition increasing from 51.5 ± 1.3% at 0 mV to 79.0 ± 0.9% at -120 mV.
The inhibition of GlyT2a-EL1 by MTSET or MTSES was not voltage-dependent (Fig. 7B). In addition, the potency of MTSES on GlyT2a-EL1 and GlyT2a-A223C ( Fig. 7C) was much lower than that of MTSET, as expected from their reported relative reactivity (18). This contrasts with the fact that the two reagents inhibited GlyT1b with comparable potencies (Fig. 7A).

Inactivation of Glycine Transporters by MTSET can be reversed with DTE.
Covalent modification of cysteine by MTSET produced an inactivation of sensitive glycine transporters which contain a cysteine in their EL1 sequence. This inactivation persisted after complete wash-out of MTSET (Fig. 8B). We tested if the application of a reducing agent like DTE may reverse the alkylation by MTS reagents, as reported for the glucose transporter SGLT1 (19). The data, shown in Figure 8 indicate that superfusion with DTE (5 min, 10 mM in CholineCl Ringer) restored partially or totally the glycine-evoked current, with a more dramatic effect with GlyT2a-A223C (Fig. 8). In addition, the small increase in leak current observed with GlyT2a-EL1 (Fig. 1h) and GlyT2a-A223C ( Fig. 1f and 8A-B) was reversed by DTE.

DISCUSSION
The topological organization of the Na + , Clneurotransmitter transporter family has now received good experimental support in favor of the original model of 12 putative transmembrane segments proposed for GAT-1 (13). Nevertheless the precise organization of the first third of the protein, which contains amino acids critical for function, is still debated (8)(9)(10)(11)(12)14,20,21). Recent insight into the localization of the loops and the accessibility of membrane segments has been obtained (8,9,12,14) through the identification of solvent-accessible residues using the substitutedcysteine accessibility method (SCAM, see review in (18)). This method has been extensively applied to determine the residues lining the pore of ionic channels and receptors (22)(23)(24)(25) and to map the ligand-binding crevice in G -protein coupled receptors (26).
Glycine transporters are good candidates for SCAM because one of them (GlyT2a) is totally insensitive to MTS reagents in uptake assays and voltage-clamp studies thus allowing further structure/function characterization without the need for a cysteine-less mutant (18). Sequence comparison reveals that GlyT2a lacks the wellconserved cysteine in EL1 that is present in GlyT1b and to which MTSET sensitivity was attributed in GAT-1 (8) and SERT (9). This cysteine substitution (A223) is the first of three consecutive differences between GlyT1b (CYR) and GlyT2a (AFQ) within a stretch of 25 otherwise conserved amino-acids in the TM1-TM2 region (table   1) which is known to be critical for function (27)(28)(29)(30).
We decided to probe EL1 accessibility and function in glycine transporters combining glycine uptake assay and transport current recordings. We have shown previously that charge and glycine uptake are precisely correlated in oocytes expressing GlyT1b (+1 elementary charge (e)/glycine) and GlyT2a (+~2 e/glycine) (5) and that the two transporters differ notably in their pre-steady state kinetics and leak current. The interest of a dual approach was reinforced by the fact that conflicting results have been reported on GAT-1 concerning MTSET accessibility (8,10,14).
We report here that the first amino-acid of EL1 (C62) in GlyT1b which is also found in GAT-1 (C74), SERT (C109) and DAT (C90), is accessible to extracellularly applied MTSET and MTSES. All transporter-associated currents ( pre-steady-state and uptake) were blocked similarly after cysteine modification by MTS reagents. In contrast, we found that none of the electrophysiological properties (glycine-evoked current, charge movement, leak current) of the wild-type GlyT2a were affected by MTSET. Introduction of a cysteine (A223C), either by point mutation or by replacing EL1 with the GlyT1b sequence, conferred MTS reagent sensitivity.
Previous results on the cysteines of Na + /Clcoupled transporters involved high concentrations of MTS reagents (≥ 1 mM) and prolonged applications (5 to 10 minutes in most cases) to achieve only partial inhibition (8)(9)(10)12,14). Similarly, the reduction of uptake current of the wild-type GlyT1b was only partial after 5-minute  (18) or for exposed cysteines near the entrance of the pore in acetylcholine receptor (31). Nevertheless, these rate constants are in the range reported for cysteines located near the cytoplasmic end of the acetylcholine receptor pore in the absence of ligand (31). This indicates that even in the more reactive GlyT2a-A223C the EL1-cysteine is exposed only transiently to the extracellular solution or is located in a diffusion-limiting crevice.
The rate of inactivation was influenced by the two other non-conserved residues of EL1. Differences in reactivity may be attributed to the presence (GlyT1b, GlyT2a- GlyT1b, which contrasts with the > 10-fold difference in reactivity in solution between these two compounds (18). Comparable changes in reactivity ratio have been observed in the case of the CFTR channel (32), for which residues located deep inside the pore (T351, Q353) show a higher reactivity for MTSES than MTSET. In this channel, the arginine at position 352 is considered as a key residue for anion selectivity (32). In the wild-type GlyT1b, R64 may facilitate the access to C62 for the negatively charged but less reactive MTSES.
The orientation of the arginine, in addition to its charge, plays a role in controlling the access to EL1-cysteine, as GlyT2a-EL1 was much less sensitive to MTSES than to MTSET (by a factor of ~5, as estimated from the inhibition after 5 minutes). In addition, position 225 of GlyT2a (an arginine in GlyT2a-EL1) is not accessible to extracellularly-applied MTSET (López-Corcuera B., R. Martínez-Maza, E. Núñez, A. Geerlings and C. Aragón, submitted). Together, these results suggest that EL1 has a different conformation in GlyT1b and GlyT2a.
Another difference between the two transporters was that membrane potential       GlyT2a-A223C