Which Form of Dopamine Is the Substrate for the Human Dopamine Transporter: the Cationic or the Uncharged Species?*

The question of which is the active form of dopamine for the neuronal dopamine transporter is addressed in HEK-293 cells expressing the human dopamine transporter. TheK m value for [3H]dopamine uptake fell sharply when the pH was increased from 6.0 to 7.4 and then changed less between pH 7.4 and 8.2. The K I for dopamine in inhibiting the cocaine analog [3H]2β-carbomethoxy-3β-(4-fluorophenyl)tropane binding displayed an identical pH dependence, suggesting that changes in uptake result from changes in dopamine recognition. Dopamine can exist in the anionic, neutral, cationic, or zwitterionic form, and the contribution of each form was calculated. The contribution of the anion is extremely low (≤0.1%), and its pH dependence differs radically from that of dopamine binding. The increase in the neutral form upon raising the pH can model the results only when the pK a1 (equilibrium neutral-charged) is set to a much lower value (6.8) than reported for dopamine in solution (8.86). The sum of cationic and zwitterionic dopamine concentrations remained constant over the entire pH range studied. These forms are the likely transporter substrates with pH-dependent changes occurring in their interaction with the transporter. The binding of dopamine, a hydroxylated phenylethylamine derivative, displays the same pH dependence as guanethidine, a heptamethyleniminoethyl- guanidine derivative fully protonated under our conditions. An ionizable residue in the transporter could be involved that does not interact with or impact the binding of bretylium, a quaternary ammonium phenylmethylamine derivative that is always positively charged and shows only a minor reduction in K I upon increasing pH.

Na ϩ ions and one Cl Ϫ ion per DA molecule with the assumption that the cationic form of DA is the substrate for uptake (7)(8)(9). DA has an amino group that can accept a proton and a phenolic hydroxyl group that can donate a proton; the second phenolic group has a pK a value greater than 12. Therefore, except at extremely basic pH values where both hydroxyl groups can be dissociated, DA can exist as a cation ( ϩ H 3 NDOH, with D for DA skeleton), a zwitterion ( ϩ H 3 NDO Ϫ ), a neutral form (H 2 NDOH), or an anion (H 2 NDO Ϫ ). With pK a1 and pK a2 values of 8.86 and 10.5, respectively (10 -12) (see "Experimental Procedures"), it can be calculated that at physiological pH, DA exists mostly as a cation, which has prompted the assumption that this is the active form for transport (7)(8)(9). Indeed, in the analogous cases of neuronal uptake of serotonin and norepinephrine, strong evidence for translocation of the cationic form has been advanced (13,14). Indirect evidence in favor of the cation also being the active form for DA uptake comes from site-directed mutagenesis studies (15) and molecular modeling (16) implicating an interaction between the protonated amine group of DA and Asp-79 of the DAT. The question of the active monoamine form for uptake by the vesicular monoamine transporter was given a great deal of attention more than a decade ago, with mixed conclusions. A case has been made for the cationic (17,18) as well as the uncharged form (for review see Ref. 19), and to our knowledge no new information on this issue is available.
In the present study, the question of the active form of DA for the neuronal DAT is addressed by monitoring for the human DAT (hDAT) cloned by Janowsky and colleagues (20,21) and the pH dependence of (i) uptake and binding of DA, which can be in an ionized or neutral form, and (ii) binding of guanethidine, which is positively charged in the pH range studied (see below), and bretylium, which is always positively charged as a quaternary ammonium salt. Uptake was measured by monitoring the accumulation of [ 3 H]DA, and binding was assessed indirectly through inhibition of high affinity binding of the cocaine analog [ 3 H]2␤-carbomethoxy-3␤-(4-fluorophenyl)tropane (WIN 35428), which interacts with a domain on the DAT that overlaps with the DA domain (5,15,16,(22)(23)(24). By comparing DA uptake and binding, we can determine whether changes in uptake as a function of pH are solely determined by changes at the level of DA recognition, the first step in the DA translocation cycle. In contrast to DA, which is a dihydroxylated phenylethylamine, bretylium is a nonhydroxylated bromophenylmethylamine in which the amine is quaternary (25), and therefore the formation of an overall neutral species is not possible for this compound. Bretylium is usually employed for its inhibitory effect on norepinephrine release (see Ref. 26) and monoamine oxidase (27). In addition, it can be taken up into dopaminergic cells (28) and interferes with DA uptake into striatal synaptosomes (29). Guanethidine is a compound derived from guanidine. The latter is a strong base with a pK a of 12.5 (30) and is therefore predominantly in the positive guanidinium ion form over a wide pH range up to 11. Guanethidine has a pK a1 of 11.4 and a pK a2 of 8.3 (31). The former pK a value describes the ability of guanethidine, at the pH values between 6.0 and 8.2 studied here, to carry one proton at the guanidine residue, whereas the latter pK a value represents the ability of the guanido group to accept an additional proton as a function of pH (at pH 8.2 in ϳ50% of the molecules). In a similar but not identical manner as bretylium, guanethidine is known to interfere with catecholaminergic transmission in mammalian (32) and invertebrate (33) systems. Guanethidine can act as a substrate for the norepinephrine transporter (34) and has been reported to inhibit neuronal uptake of norepinephrine (25), but its affinity for the DAT, to our knowledge, has not been reported. . GF/C fiber filter mats and CytoScint ES mixture for the uptake assays were from Brandel (Gaithersburg, MD) and ICN (Costa Mesa, CA), respectively. The source of the DAT was the HEK-293-hDAT cell line developed by Janowsky and co-workers (21).

Materials-[
[ 3 H]WIN 35428 Binding Assay-The general conditions for growing the cells, working up the membrane preparations, and conducting the binding assays were as described previously by us (35). Briefly, after cell lysis, membranes were prepared and homogenized with the Brinkmann Polytron (setting 6, 15 s) in ice-cold saline. Binding assays were performed in ice-cold 30 mM sodium phosphate buffer (the result of mixing primary and half-strength secondary sodium phosphate buffer to a selected pH value between 6.0 and 8.2, as indicated, at room temperature) also containing 122 mM sodium chloride, 5 mM potassium chloride, 1.2 mM MgSO 4 , 10 mM glucose, 1 mM CaCl 2 , and 0.1 mM tropolone for catechol-O-methyltransferase inhibition (21) ("assay buffer"). The listed pH values are those measured at room temperature after all of the above components had been added. In this buffer, binding assays were conducted for 10 min at 21°C in 96-well plates with 4 nM [ 3 H]WIN 35428. The assays contained additionally, where appropriate, nonradioactive WIN 35428 (1-300 nM), DA (0.3-100 M), guanethidine (10 -3,000 M), or bretylium (10 -3,000 M) (for K D or K I estimation, respectively), HEK-293-hDAT membranes (ϳ120 g of protein as determined by the method of Lowry; see Ref. 35) with or without 100 M cocaine for assessment of nonspecific binding. Assays were terminated on glass fiber filter mats (presoaked in 0.05% (v/v) polyethyleneimine) with the MACH 3-96 Tomtec harvester (Wallac Inc.) using phosphate (10 mM)-buffered saline (pH 7.4). Filters were counted in a Microbeta Plus liquid scintillation counter (Wallac Inc.) at an average counting efficiency of 35%.
[ 3 H]DA Uptake Assay-The general conditions for conducting the uptake assays were as described by us previously for synaptosomal suspensions (22). Briefly, HEK-293-hDAT cells were lifted from the culture flasks with phosphate (10 mM)-buffered saline (pH 7.4) and centrifuged at 100 ϫ g for 6 min. The pellet of cells from two confluent 150-cm 2 flasks was resuspended in 6 ml of saline in a glass homogenizer with a motor-driven Teflon pestle (clearance of approximately 0.15 mm compared with a cell diameter of 0.01 mm). This step helped to produce even suspensions while not causing more cell damage than resuspension through vortexing (combined with repeated movements through a 5-ml pipette) as judged under the microscope by the Trypan Blue dye exclusion test. Uptake assays were conducted in a final volume of 0.4 ml of the above described assay buffer at selected pH values and contained 50 l of cell suspension, 20 l of [ 3 H]DA stock (contributing a final 6 nM concentration of tritiated DA (NEN Life Science Products) and 46 nM unradioactive DA), and for saturation studies 20 l of varying amounts of unradioactive DA (for final concentrations ranging from 0 to 1000 nM). Nonspecific uptake was defined by 100 M cocaine. After a preincubation of the cell suspension and assay tubes with assay buffer for 15 min in a 21°C water bath, 50 l of cell suspension was added to the tubes, which were then incubated while gently shaking in a 21°C water bath for 8 min. The reaction was terminated with the addition of 4 ml of ice-cold wash buffer (assay buffer minus tropolone). Subsequent filtration (within 4 min after stopping for all tubes) was performed with a 24-pin Brandel harvester through a Whatman GF/C glass fiber filter presoaked in 0.05% (w/v) poly-L-lysine with three additional rinses of 4 ml of ice-cold wash buffer. In control experiments for assay mixtures with stop buffer, after 6 min on ice no efflux of accumulated [ 3 H]DA occurred, as was the case for synaptosomal assays (22). Radioactivity on the filters was measured in 5 ml of CytoScint mixture with a Beckman model LS 6000IC scintillation counter with quench correction. The average protein content of the cell suspensions was estimated to be 250 g determined by the method of Lowry described above.
Distribution of Neutral and Ionic Forms of DA at Varying pH-The pK a1 value was set to 8.86 and the pK a2 value to 10.5, which is the average of the values of 8.8 -8.9 and 10.4 -10.6, respectively, reported by Armstrong and Barlow (10), Lewis (11), and Mack and Bönisch (12). The first pK a governs the transition from the cationic ϩ H 3 NDOH to the mixture of ϩ H 3 NDO Ϫ and H 2 NDOH (both overall neutral species) rather than representing the dissociation of the amino group only (10,12). Similarly, the second pK a describes the transition of the two neutral species into the anion H 2 NDO Ϫ and not just the dissociation of the hydroxyl group (10,12). This analysis does not consider extremely basic pH conditions in which ionization could occur in the second hydroxyl group, which has a pK a Ͼ 12.
The percent of total DA existing in the cationic form was calculated as follows: % positive ϭ 100/(1 ϩ antilog(pH Ϫ pK a1 ) ϩ antilog(2pH Ϫ pK a1 Ϫ pK a2 )); this was arrived at by rewriting the equations for the two equilibria described by Armstrong and Barlow (10) as shown for the single case of protonation of a base by Courtney and Strichartz (36). This derivation also takes into account that log( Ϫ pK a1 by solving both equilibrium equations for pH.
The percent of total DA existing in the anionic form was calculated as follows: % negative ϭ 100 Ϫ % neutral Ϫ % positive.
Prediction of DA Uptake Velocity Based on the Neutral Form of DA Being the Substrate-DA velocity for 56 nM [ 3 H]DA in the assay at varying pH was predicted based on fixed pK a1 and pK a2 values for the case that the neutral form is the active form for translocation. The pK a2 was always 10.5 (see above), and the pK a1 was 8.86 (theoretical value, see above) or set at a value between 6.8 and 9.5 (see "Results" and "Discussion"). At pH 8.2, the highest pH studied, more DA will be in the neutral form than at a lower pH, and for each pair of pK a1 and pK a2 , the K m of the neutral form of DA was computed from the "composite K m " according to the following: K m(neutral) ϭ K m(composite,pH 8.2) ⅐0.01⅐% neutral at pH 8.2. For example, if 10% of DA is neutral at pH 8.2, then the K m for the neutral form will be 10-fold smaller than the measured composite K m at pH 8.2 if the neutral form is the active form. Subsequently, K m(composite) was computed over the entire pH range from the following: K m(composite) ϭ K m(neutral} /(0.01⅐% neutral (a different set of calculations for each pair of pK a1 and pK a2 ). Finally, velocity, V, at each pH for a given set of pK a1 and pK a2 was calculated as V ϭ V max ⅐(56/(56 ϩ K m(composite) )) in which V max is the value measured across pH (1.10 pmol/mg protein/8 min) and 56 is the total composite [DA] in nM with K m expressed also in nM.
Data Analysis-IC 50 values for the inhibition of [ 3 H]WIN 35428 binding by compounds were estimated with the ALLFIT equation (37), and K D , K m , B max , and V max estimates were obtained with the nonlinear computer fitting program LIGAND (38) as described by us (39). IC 50 values were converted to K I values by the Cheng-Prusoff equation (40) taking into account the level of radioligand and the K D of [ 3 H]WIN 35428 binding at each pH. All results are expressed as means Ϯ S.E. Statistics included one-way ANOVA followed by the Student-Newman-Keuls multiple comparisons test or the least significant difference multiple range test and two-way ANOVA with factors A and B and interactions A ϫ B. Where needed, data were log-transformed for homogeneity of variance. The accepted level of significance was 0.05.

RESULTS
DA Uptake as a Function of pH-The K m for [ 3 H]DA uptake decreased approximately 3-fold upon increasing the pH from 6.0 to 7.4 and did not change upon further increasing the pH to 8.2 (Fig. 1A). There were no statistically significant differences in the V max values obtained in the 6.0 -8.2 pH range.
The velocity of [ 3 H]DA uptake at a fixed concentration of DA of 56 nM increased sharply upon increasing the pH from 6.0 to 7.0 and then remained at a plateau up to the highest pH applied, which was 8.2 (Fig. 1B).
A different pH dependence was calculated for the concentration of either the cationic form of DA ( ϩ H 3 NDOH), which at a pK a1 of 8.86 and pK a2 of 10.5 (10 -12), was in the majority at all pH values studied, or the neutral forms (zwitterion ϩ H 3 NDO Ϫ plus neutral form H 2 NDOH), which increased approximately 10-fold for each pH unit increase (Fig. 1C). Indeed, velocity predictions based on the neutral form being the substrate deviated substantially from the observed pH dependence of DA uptake (Fig. 1B, compare solid line with data points with broken line for pK a1 ϭ 8.86).
DA Binding as a Function of pH-The K I value for DA in inhibiting [ 3 H]WIN 35428 binding decreased approximately 3.5-fold upon increasing the pH from 6.0 to 7.4, paralleling the change in the K m for DA uptake, and showed a smaller decrease of approximately 1.5-fold upon further increasing the pH to 8.2 (Fig. 1A).
Guanethidine and Bretylium Binding as a Function of pH-The affinity of guanethidine for the DAT was generally lower than that of DA (for K I values see legend to Fig. 2). However, there was a resemblance between guanethidine and DA in the pattern of potency changes as a function of pH (Fig. 2). Thus, the K I value for guanethidine in inhibiting [ 3 H]WIN 35428 binding decreased approximately 3-fold upon increasing the pH from 6.0 to 7.4 and showed a smaller decrease of approximately 2-fold upon further increasing the pH to 8.2 (Fig. 2). WIN 35428 itself had a higher potency than DA or guanethidine (see legend to Fig. 2), but again the pattern of pH dependence was similar (Fig. 2) (note the lack of a statistically significant interaction factor in analysis of variance of the data for DA, guanethidine, and WIN 35428). In contrast, the total decrease in the K I of bretylium upon increasing the pH from 6.0 to 8.2 was only 2-fold as compared with the 5-7-fold reductions seen with DA, guanethidine, or WIN 35428 (Fig. 2) (note the statistically significant interaction factor in analysis of variance of the data for all compounds including bretylium).
No evidence was found for a deviation from a competitive mechanism of inhibition of [ 3 H]WIN 35428 binding by bretylium or guanethidine. First, the Hill numbers for the inhibition curves were close to unity over the entire pH range. Second, a fixed concentration of 700 M bretylium or guanethidine caused an increase in the K D value without changing the B max studied at both pH 6.0 and 8.2 (Table I). As expected, binding was increased at pH 8.2 as compared with 6.0; this happened in all cases because of a decrease in K D (Table I). DISCUSSION Predicted DA Uptake with Overall Neutral DA as the Substrate-If it is assumed that the increase in DA uptake with pH is accounted for solely by the shift in the distribution of DA forms toward the overall neutral forms ( ϩ H 3 NDO Ϫ plus H 2 NDOH), the predicted pH dependence does not coincide with the observed pattern at the pK a1 and pK a2 values reported for DA of 8.86 and 10.5, respectively (10 -12). It could be argued that the pK a1 value may be reduced in the vicinity of the binding site when its environment is more hydrophobic, making it easier for the compound to loose a proton. This pattern has been suggested for the binding of cocaine to the local anesthetic binding site in the voltage-dependent sodium channel, which is predicted more accurately by a pK a value of 7.1 than the value of 8.4 -8.7 measured in water (41). The present DA uptake data could be interpreted in such a context with a pK a1 value of 6.8 resulting in a reasonable fit for uptake at a fixed concentration of 56 nM [ 3 H]DA in the studied pH range of 6.0 -8.2. Another way the possibility of a pK a shift can be shown is by calculating the K m or K I of the neutral forms of DA from the observed "composite" K m,I taking into account the proportion of the neutral forms at each pH. If the calculated values remain constant across pH, this result would be in agreement with the concept that pH-dependent changes in proportion solely underlie the uptake or binding changes. At pK a1 8.86 the K m,I of the neutral forms was not constant across pH, but lowering the pK a1 to 6.8 gave reasonably constant values (Table II).
It can only be hypothesized as to how hydrophobic the environment is around the DA binding site to substantially reduce the pK a1 value below that found in a water milieu. DA is thought to bind to a site in a central aqueous cavity surrounded by amphipathic membrane-spanning helices (5). Transporter chimera and truncation studies support the idea that both substrates and blockers interact with residues in transmembrane domains lining the permeation pathway (reviewed in Ref. 4), which may include Asp-79 in transmembrane domain 1 (15). Prior to translocation, DA accesses its centrally located binding site, which is facing externally; translocation occurs through a conformational change exposing the binding site on the internal face (5,42). The presently observed similarity in the pH dependence of DA uptake and binding suggests that the pH effects occur at the level of DA recognition. It is possible that the DA binding site in the central crevice of the DAT is accessed by DA from the aqueous side, whereas the local anesthetic site near the inactivation gate in the Na ϩ channel (43) can be reached by the relatively more lipophilic cocaine (log P of 2.4 (44) compared with Ϫ0.99 for DA (12)) through the hydrophobic pathway in the membrane lipidic environment (45). Recent evidence for the serotonin transporter, which is closely related to the DAT, suggests binding of serotonin to Ile-72 and Tyr-176 (46). These residues lie in the ␣-helical structure of transmembrane domain 3 on a vertical patch that faces toward the binding pocket accessible to the external medium as judged from their sensitivity to membrane-impermeant, charged [2-(trimethylammonium)ethyl]methanethiosulfonate in the cysteine-scanning approach (46).
Predicted DA Uptake with Neutral, Uncharged DA as the Substrate-The composition of the mixture of the zwitterion ϩ H 3 NDO Ϫ and the neutral form H 2 NDOH is defined by the

FIG. 2. pH dependence of the potency of compounds in inhibiting [ 3 H]WIN 35428 binding. K I values were determined from inhibition curves for each compound as described under "Experimental
Procedures." Each point represents the mean Ϯ S.E. (vertical bar, shown where greater than point itself) of three independent experiments carried out in triplicate expressed as the percent of the value at pH 7.4. At this pH level, the K I value for WIN 35428 was 15 nM; for DA, 6.0 M; for guanethidine, 327 M; and for bretylium, 664 M. ‫ء‬ indicates p Ͼ 0.05 for interaction factor (F(9, 28) ϭ 2.08) (two-way ANOVA with pH as factor A and inhibitors guanethidine, DA, and WIN 35428 as factor B; both factors were highly significant with p Ͻ 0.001). $ indicates p Ͻ 0.001 for interaction factor (F(9, 28) ϭ 6.72) (two-way ANOVA with pH as factor A and inhibitors bretylium, guanethidine, DA, and WIN 35428 as factor B; both factors were highly significant with p Ͻ 0.001). b p Ͻ 0.05 (compared with control at same pH; least significant difference multiple range test following one-way ANOVA at same pH indicating significant differences (F(2, 9) Ն 7.95; p Յ 0.01)). c p Ͼ 0.9 (F(2, 18) ϭ 0.018 for factor B in two-way ANOVA with pH as factor A and added drug as factor B). d p Ͻ 0.05 (compared with guanethidine at same pH; least significant difference multiple range test following one-way ANOVA at same pH indicating significant differences (F(2, 9) ϭ 10.36; p ϭ 0.005)). equilibrium dissociation constant K Z ϭ ϩ H 3 NDO Ϫ /H 2 NDOH, which has been determined to be ϳ7.83 (the average of the range 5.8 -10.0) by various approaches (10). Thus, the neutral, uncharged form is less predominant than the zwitterion, but its increase as a function of rising pH is the same as for the total mix of ϩ H 3 NDO Ϫ plus H 2 NDOH. Indeed, when the DA uptake velocity at 56 nM [ 3 H]DA was calculated taking into account K Z and assuming the neutral, uncharged DA as the substrate rather than the mix, the curves were identical as shown for the mix in Fig. 1B. Thus, the hypothesis that this form of DA is the substrate also requires the assumption of a lower pK a1 value than is measured in a water environment.
Predicted DA Uptake with Zwitterionic DA as the Substrate-The increase of the zwitterion as a function of rising pH is the same as for the neutral forms of DA and again does not fit the experimental data at pK a1 8.86. Therefore, the zwitterion is not likely to be the sole form of DA acting as the substrate for the DAT. If a protonated amine group of DA is involved in the interaction of DA with Asp-79 in the DAT as suggested by Kitayama et al. (15) and Edvardsen and Dahl (16), then the contribution of the zwitterion and the cation may have to be considered together (see below). Because the contribution of the zwitterion at the pH values studied here is only a small fraction of that of the cation (see below), the difference between predicted DA uptake (binding) values with and without zwitterion is too small to draw conclusions regarding the comparison of the fit of these values to the experimental data. Therefore, the involvement of the zwitterion in addition to the cation is a possibility that cannot be excluded based on the present data.
Predicted DA Uptake with Anionic DA as the Substrate-If anionic DA is the substrate, it can be calculated that at pH 8.2 with a 0.1% contribution of the anion (at pK a1 8.86 and pK a2 10.5, see "Experimental Procedures"; at pH 7.4 the contribution of the anion is even less than 0.1%) and a K D(composite) ϭ 6.0 M (from observed inhibition by DA of [ 3 H]WIN 35428 binding at pH 7.4), the K D(anion) ϭ 6 nM. Such an affinity is typically high enough to allow measurement by radioligand binding approaches. The k ϩ1 for the binding of anionic DA can be estimated from the results of Meiergerd and Schenk (47) as k cat(anion) /0.001 K m(composite) in which k cat(anion) is the catalytic rate constant for overall movement of anionic DA across the membrane assuming a 0.1% contribution of the anionic species to total DA. Because under those conditions k cat(anion) equals 0.001 V max( composite) /T d , in which T d is the transporter density (47), we can write k ϩ1(anion) ϭ (0.001 V max(composite} /T d )/(0.001 K m(composite) ) ϭ (V max( composite) /T d )/(K m(composite) ), which is exactly the same as the k ϩ1 for composite DA calculated by Meiergerd and Schenk (47), i.e. of 1.4 ϫ 10 6 M Ϫ1 s Ϫ1 . With this value for k ϩ1 , the k Ϫ1(anion) (ϭK D ⅐k ϩ1 ) ϭ 6 ϫ 10 Ϫ9 M (affinity at pH 7.4). 1.4 ϫ 10 6 M Ϫ1 s Ϫ1 ϭ 0.0084 s Ϫ1 , indicating a t1 ⁄2 of 1.4 min. This off-rate is slow enough for detection of binding by rapid filtration, and yet [ 3 H]DA binding to the DAT has never been observed. Clearly, the hypothesis that anionic DA is the substrate for the DAT is untenable.
Predicted DA Uptake with Cationic DA as the Substrate-Most of the DA (at pK a1 8.86 and pK a2 10.5) was in the cationic form over the pH range studied, slightly decreasing from 99.9% at pH 6.0 to 82.0% at pH 8.2. If the protonated amine group is the important determinant, one should consider the sum of the cation and zwitterion, which stays virtually constant from 99.9% at pH 6.0 to 97.9% at pH 8.2. Therefore, if the cationic form (with or without zwitterion) is the substrate for the DAT, the increases in uptake with pH must result from mechanisms beyond increases in concentration of the active substrate. The parallel changes in K m for uptake and K I for binding suggest that changes in the affinity of DA for the DAT occur as a function of pH.
Consistent with the idea that changes occur in the interaction of cationic DA with the DAT as a function of pH, guanethidine, which is positively charged at the pH range of 6.0 -8.2 studied here, displays changes in affinity for the DAT that parallel the changes for DA itself. Although bretylium is always positively charged, it shows less dependence on pH in interacting with the DAT than guanethidine or DA, suggesting that elements are involved outside of the overlapping binding domains for bretylium and DA (see "pH-dependent Changes in Interaction between DA and DAT: Ligand or Transporter?" below). The lesser pH dependency is not explained by invoking a distinct mechanism of inhibition for bretylium, because both bretylium and guanethidine behave as competitive inhibitors of [ 3 H]WIN 35428 binding in saturation analysis, which is in agreement with the idea that a shared domain of the DAT is involved. Structurally, bretylium shares the phenyl ring with DA, but the carbon chain between the phenyl ring and the nitrogen (N ϩ ) is shorter than in DA. In contrast, this carbon chain has the same length in guanethidine as in DA, but the ring is made up of a heptamethylenimino rather than a phenyl structure in guanethidine. Most likely, these differences compared with DA contribute to the relatively lower potencies of bretylium and guanethidine at the DAT.
The observed K I of bretylium at pH 7.4 of 664 M is appreciably higher than the IC 50 value of 149 M reported by Kammerer et al. (29) for inhibition of striatal synaptosomal DA uptake. The most likely reason for this difference is that bretylium, like DA, is a substrate for the DAT, requiring higher concentrations for saturation of binding than uptake. The same TABLE II Estimates of the K m or K I of the neutral form of DA as substrate computed from the observed "composite" K m or K I at varying pK a1 (see text) If the neutral forms of DA ( ϩ H 3 NDO Ϫ plus H 2 NDOH) are the substrate for the DAT and if changes in uptake or binding with pH are only the result of changes in the concentration of the neutral forms as a function of pH, the K m or K I of the neutral forms can be computed from the observed "composite" K m,I taking into account the proportion of the neutral form. The pK a1 was set at the values shown, and the pK a2 was taken as 10.5 (see text). The "composite" K m,I is from three to four independent experiments assayed in triplicate. The corrected sum of squares is the sum of all squared differences between the mean and the individual K m,I values divided by the mean. Comparison with Other Transporters for Monoamines-The present data are consonant with the conclusion that the cationic form of DA, perhaps including the zwitterion, is the more likely substrate for the DAT. Accordingly, the results are consistent with the assumption that translocation of one DA molecule brings in one positive charge (9). pH dependence studies have led to a similar conclusion for the transport of serotonin by the plasma membrane serotonin transporter (13) and of norepinephrine by the plasma membrane norepinephrine transporter (14). For transport of monoamines by the vesicular monoamine transporter, the issue is still being debated (5). Although the majority of the earlier studies favored the idea of the neutral forms of monoamines acting as substrates (for review see Ref. 19), the pH dependence of the uptake of the positively charged 1-methyl-4-phenylpyridinium cast doubt on this conclusion (18). Thus, the uptake of 1-methyl-4-phenylpyridinium was found to have the same pH dependence as that of monoamines, even though as a quaternary ammonium base it is always positively charged. pH-dependent Changes in Interaction between DA and DAT: Ligand or Transporter?-pH-induced changes in the interaction between DA and the DAT could result from changes at the level of the ligand. For example, introduction of positive charges on the ligand by protonation may result in conformational changes in the ligand itself. Alternatively, changes occur at the level of the transporter. For example, the protonated amine group of DA (and of guanethidine) could interact with a negatively charged group in the DAT protein as suggested for the vesicular monoamine transporter by Darchen et al. (18). If the latter group is ionizable with a pK ϳ 6.8, it would explain the present pattern of uptake and binding of DA (and binding of guanethidine), which shows an increase with pH that is less than logarithmic. In this scenario, bretylium would not interact with the same ionizable group, perhaps because the distance between the carbon ring and the positively charged amine is one carbon shorter than in DA and guanethidine. Indeed, on the basis of extensive structure-activity studies on the inhibition of neuronal norepinephrine uptake by norepinephrine congeners and tricyclic inhibitors, Maxwell et al. (25) have postulated the phenyl ring and the N ϩ as two key elements in the binding of these compounds to the norepinephrine transporter.
Another possibility, again at the level of the DAT, is that the pH dependence of the interaction of many compounds with the DAT resides in an overlapping binding domain that is susceptible to pH. It may be recalled here that there is substantial evidence implicating a shared binding domain on the DAT for DA, cocaine-like inhibitors, and GBR 12935-like compounds (1, 5, 24, 42, 49, 50). The pH dependence observed here for the interaction of DA, guanethidine, and WIN 35428 with hDAT is very similar to that reported previously by us for the interaction of WIN 35428, cocaine methiodide, and benzocaine (39) and by Andersen for GBR 12935 (51) with rDAT. Among these compounds, cocaine methiodide is always protonated, benzocaine is always neutral, and the others are bases with pK a values Ն 8.4 (see Ref. 39). It could be concluded that a common binding domain (not shared by bretylium) confers the pH dependence, for example by ionizable groups (with a pK ϳ 6.8) in the DAT, impacting the conformation and thereby the binding potential. In this scenario it could be considered that either cationic or neutral DA binds with similar affinity and that the proportion at each pH determines how much of each form is translocated. Thus, at pH 7.4 most of the DA is in the cationic form, overwhelmingly resulting in translocation of the cationic form. The conclusion would be that the charge translocated with each DA molecule is a function of pH, with less charge translocated at extremely basic pH values, adding another variable to be taken into account in addition to membrane potential (9) in comparing observed charge transfers with expected values based on the stoichiometry Na ϩ :Cl Ϫ :DA ϩ ϭ 2:1:1. Such a phenomenon may be hard to establish because there is an appreciable leak conductance associated with the DAT, which may readily carry protons or otherwise be sensitive to pH changes, possibly by a pH-dependent ion selectivity (9).