The role of conserved tryptophan and acidic residues in the human dopamine transporter as characterized by site-directed mutagenesis

Authors

  • Nianhang Chen,

    1. Department of Biomedical and Therapeutic Sciences, University of Illinois College of Medicine, Peoria, Illinois, USA
    2. Department of Pharmacology, Nanjing Medical University, Nanjing, China
    Search for more papers by this author
  • Roxanne A. Vaughan,

    1. Department of Biochemistry and Molecular Biology, University of North Dakota School of Medicine and Health Sciences, Grand Forks, North Dakota, USA
    Search for more papers by this author
  • Maarten E. A. Reith

    1. Department of Biomedical and Therapeutic Sciences, University of Illinois College of Medicine, Peoria, Illinois, USA
    Search for more papers by this author

Address correspondence and reprint requests to Nianhang Chen, Department of Biomedical and Therapeutic Sciences, University of Illinois College of Medicine, Box 1649, Peoria, IL 61656, USA. E-mail: nhc@uic.edu

Abstract

The human dopamine (DA) transporter (hDAT) contains multiple tryptophans and acidic residues that are completely or highly conserved among Na+/Cl-dependent transporters. We have explored the roles of these residues using non-conservative substitution. Four of 17 mutants (E117Q, W132L, W177L and W184L) lacked plasma membrane immunostaining and were not functional. Both DA uptake and cocaine analog (i.e. 2β-carbomethoxy-3β-(4-fluorophenyl)tropane, CFT) binding were abolished in W63L and severely damaged in W311L. Four of five aspartate mutations (D68N, D313N, D345N and D436N) shifted the relative selectivity of the hDAT for cocaine analogs and DA by 10–24-fold. In particular, mutation of D345 in the third intracellular loop still allowed considerable [3H]DA uptake, but caused undetectable [3H]CFT binding. Upon anti-C-terminal-hDAT immunoblotting, D345N appeared as broad bands of 66–97 kDa, but this band could not be photoaffinity labeled with cocaine analog [125I]-3β-(p-chlorophenyl)tropane-2β-carboxylic acid ([125I]RTI-82). Unexpectedly, in this mutant, cocaine-like drugs remained potent inhibitors of [3H]DA uptake. CFT solely raised the Km of [3H]DA uptake in wild-type hDAT, but increased Km and decreased Vmax in D345N, suggesting different mechanisms of inhibition. The data taken together indicate that mutation of conserved tryptophans or acidic residues in the hDAT greatly impacts ligand recognition and substrate transport. Additionally, binding of cocaine to the transporter may not be the only way by which cocaine analogs inhibit DA uptake.

Abbreviations used
CFT

2β-carbomethoxy-3β-(4-fluorophenyl)tropane

DA

dopamine

DAT

dopamine transporter

dBET

drosophila betaine transporter

dSERT

drosophila serotonin transporter

GABA

γ-aminobutyric acid

hBGT

human betaine/GABA transporter

hDAT

human dopamine transporter

hGAT

human GABA transporter

hGLYT

human glycine transporter

hPROT

human proline transporter

[125I]RTI-82

[125I]-β-(p-chlorophenyl)tropane-2β-carboxylic acid

mGAT

mouse GABA transporter

MTSEA

(2-aminoethyl)methanethiosulfonate

PBS

phosphate-buffered saline

rGAT

rat GABA transporter

rPROT

rat proline transporter

TMD

transmembrane domains

SDS–PAGE

sodium dodecyl sulfate–polyacrylamide gel electrophoreses.

The Na+/Cl-dependent neurotransmitter transporters are carrier proteins located on the plasma membranes of nerve terminals and responsible for rapid uptake of neurotransmitters by neurons. This transporter family includes carriers for monoamines [dopamine (DA), norepinephrine and serotonin], γ-aminobutyric acid (GABA), amino acids (glycine and proline), taurine, betaine and creatine (Masson et al. 1999). All these members require Na+ and Cl for transport. They share a relatively high degree of homology in primary sequence and presumably have a common topology with 12 transmembrane domains (TMD) connected by alternating extracellular and intracellular loops with the N- and C-termini in the cytosol. Among these members, human dopamine transporter (hDAT) is an important molecular target for commonly abused psychostimulatory drugs including cocaine. Blockade of hDATs has been thought as one of major mechanisms underlying the reinforcing effect of cocaine (Giros et al. 1996; Amara and Sonders 1998). Recent structure-function studies for DAT has shown that domains for DA and various blocking drugs including cocaine are formed by interactions with multiple residues, some of which are separated in the primary structure but may lie together in the still unknown tertiary structures (for review, see Chen and Reith 2000).

The hDAT contains multiple tryptophan and acidic residues that are absolutely or highly conserved in the superfamily of Na+/Cl-dependent transporters (Fig. 1). These residues are presumably located either in hydrophilic loops or at the boundary between a hydrophobic transmembrane domain and a hydrophilic loop (Fig. 1). Sequence conservation of polar and charged amino acids from different transporters suggests functional importance. Generally, these conserved residues could reflect their involvement in features common for the Na+/Cl-dependent transporters, such as protein assembly and membrane insertion, conformational transitions required for translocation, and ion regulation of transporter function. Indirectly, these residues may also contribute to interactions of hDAT with its own substrates and blockers, because binding and transport at the DAT are highly dependent on conformations and ions (Rudnick 1997; Ferrer and Javitch 1998; Chen et al. 2000). Further, π-electron-carrying residues or negatively charged residues have the potential to directly interact with DA or cocaine analogs that are positively charged at physiological pH. Thus, in order to identify the tertiary binding pocket for cocaine and substrates, it is necessary to understand what function the conserved residues endow and whether this function contributes to the topological organization of cocaine and substrate binding sites.

Figure 1.

Distribution of conserved tryptophan and acidic residues in putative topology of the hDAT. Dark circles represent residues absolutely conserved for 26 Na+/Cl-dependent transporters, and white circles represent residues conservatively replaced in some members of the Na+/Cl-dependent transporter family. D68 is E in dSERT, hGAT3, rGAT2, mGAT2, hBGT1, hGLYT1c and dBET. W267 is Y in dSERT and all members of GABA transporter and amino acid transporter subfamilies. D313 is E in hPROT and rPROT. D476 is E in hSERT, rSERT and bSERT, and is N in dSERT.

In the present study, by using non-conservative substitution for the conserved tryptophan and acidic residues, we have explored the consequence of removing the indole ring or negative charges from these positions on surface expression, DA uptake and cocaine analog binding. To facilitate comparisons between uptake and binding results, binding assays were conducted in the same way as uptake assays (with intact cells in uptake buffer at 25°C). The results suggest that the mutations have profound influences on membrane expression, ligand recognition, and substrate transport. These findings may have implications for other Na+/Cl-dependent transporters.

Experimental procedures

Materials

The pCIN4 vector and the wild-type pCIN4-hDAT construct were generous gifts from Dr Jonathan A. Javitch of Columbia University. 2β-Carbomethoxy-3β-(4-fluorophenyl)tropane ([3H]CFT; 84.5 Ci/mmol), [3H]mazindol (24.0 Ci/mmol), and [3H]DA (60 Ci/mmol) were purchased from NEN Life Sciences Products (Boston, MA, USA). [125I]-3β-(p-Chlorophenyl)tropane-2β-carboxylic acid ([125I]RTI-82; 2175 Ci/mmol) was synthesized by Dr Ivy Carroll, and was radio-iodinated by Dr John Lever, Johns Hopkins University. Unlabeled CFT was from the Research Triangle Institute (Research Triangle Park, NC, USA). Other chemicals were provided by commercial sources.

Generation of cell lines stably expressing wild-type and mutant hDATs

Site-directed mutagenesis of the hDAT was performed using the QuickChange Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The following mutations were introduced: leucine (L) for tryptophan (W), glutamine (Q) for glutamate (E), and asparagine (N) for aspartate (D). The mutants were screened by restriction enzyme mapping, and the mutation was confirmed in both directions by dye terminator cycle sequencing (DNA Core Facility, University of Illinois, IL, USA). Human embryonic kidney cells (HEK 293, ATCC CRL 1573) were grown in Dulbecco's modified Eagle's medium/Ham's F12 medium supplemented with 10% bovine calf serum and 2 mml-glutamine at 37°C and 5% CO2. For stable transfection, cells were seeded into six-well plates and allowed to grow to 60% confluence. Then, each well of cells were transfected with 1 µg of wild-type or mutant pCIN4-hDAT, or pCIN4 vector alone using 6 µL Lipofectamine (Life Technologies, Inc., Gaithersburg, MD, USA) in 1 mL serum-free growth medium. Three days after transfection, the cells were selected in the growth medium containing 800 µg/mL geneticin. The pCIN4 vector (commercially described as pIRESneo) permits the translation of the recombinant protein and the antibiotic resistance marker from a single mRNA, thereby allowing all antibiotic-resistant clones to express the gene of interest (Rees et al. 1996). Thus, the resistant colonies of each transfection were pooled two weeks later and maintained in the growth medium containing 200 µg geneticin for experiments.

Transport and binding assays

Cells were dissociated with trypsin/EDTA (0.25%/0.1%, 1 mL for one 150 mm dish) and resuspended in growth medium containing 20% bovine calf serum. After a 1-h incubation at 21°C, the dissociated cells were harvested by centrifugation at 400 g for 2 min and washed once with phosphate-buffered saline (PBS). The resulted cell pellets were resuspended in uptake buffer (10 mm HEPES, 150 mm NaCl, 1.2 mm KH2PO4, 1.2 mm MgSO4, 1 mm CaCl2, 3 mm KCl, 10 mm glucose, 0.1 mm tropolone, and 1 mm ascorbic acid, pH 7.4) for either uptake or binding assays. Trypsinized cells showed better uptake activity than non-enzymatically (scraping or using commercial media) dissociated cells. No significant difference in the binding assays was observed with trypsinized and non-enzymatically dissociated cells. [3H]DA uptake assays were conducted in 96-well Multiscreen-FB plates (1.0 µm FB glass fiber, Millipore, Bedford, MA, USA) precoated with 0.05% poly-l-lysine. Preliminary experiments showed that, at 25°C, uptake was linear with the time up to 10 min and linear with the cell protein up to 250 µg. At 37°C, uptake reached plateau within 1 min, which made it impossible to measure the initial rate at the linear phase. Thus, all uptake assays were conducted at 25°C for 5 min. Uptake was initiated by adding 50 µL cell suspensions (50–200 µg) in a final volume of 200 µL uptake buffer containing 16 or 10 nm[3H]DA. The uptake reaction was terminated by rapid filtration on MultiScreen vacuum manifold (Millipore) followed by three washes with 300 µL per well of ice-cold PBS. Binding to intact cells was measured in regular 96-well plates. Binding were initiated by adding 50 µL cell suspensions (40–100 µg) in a final volume of 200 µL uptake buffer containing 2–8 nm[3H]CFT or 4 nm[3H]mazindol. Assay mixtures were incubated at 25°C for 15 min, or, in some experiments where indicated, at 4°C for 2 h. The binding reactions were terminated by filtration through 0.1% polythyleneimine presoaked glass fiber filtermat A (Wallac, Gaithersburg, MD, USA), and then the filter was washed with ice-cold PBS using a 96-well Tomtec cell harvester (Wallac). The radioactivity of uptake or binding samples was determined in SuperMix (Wallac) or Betaplate Scint (Wallac) scintillation cocktail using a 1405 Microbeta liquid scintillation counter (Wallac). In some experiments, cells expressing wild-type hDAT or D345N were incubated with 4 nm[3H]CFT in a final volume of 400 µL uptake buffer for 2 h at 4°C. The binding reaction was terminated by rapid filtration and washes (4 mL × 3) over 0.1% polyethylenimine presoaked GF/F glass fiber filters on a single-manifold Millipore filtration apparatus, and samples were counted for radioactivity in separate scintillation vials by a Beckman LS 6000 liquid scintillation counter (Palo Alto, CA, USA). In both uptake and binding assays, cells transfected with pCIN4 vectors alone were tested in parallel and used to estimate non-specific uptake and binding. For saturation or competition analysis, appropriate concentrations of non-radioactive DA, CFT, or other tested compounds were included in the assay. Kinetic parameters such as Km, Vmax, Kd, Bmax, IC50 and Hill were calculated using the computer-fitting programs LIGAND (Biosoft, Cambridge, UK) and ORIGIN (Microcal Software, Inc., Northampton, MA, USA). Apparent Ki values were calculated by the Cheng–Prusoff equation (Cheng and Prusoff 1973).

Immunostaining of transfected cells

Transfected cells were grown on 15-mm coverslips in 12-well plates for 24 h to achieve 80% confluence. Thereafter, the cells were fixed with 4% paraformaldehyde in PBS for 25 min at 21°C. Fixed cells were incubated with 0.6% H2O2 and 10% methanol in PBS for 10 min at 21°C to suppress endogenous peroxidases. After the cells were permeabilized and blocked with 2% normal goat serum and 0.2% Triton X-100 in PBS for 1 h at 21°C, they were incubated with 1 µg/mL of the rabbit anti-hDAT polyclonal antibody in blocking buffer overnight at 4°C. This primary antibody was raised against the intracellular C-terminal peptide from amino acids 598–619 (Chemicon International, Inc., Temecula, CA, USA). After incubation with the primary antibody, cells were incubated with a biotinylated goat anti-rabbit IgG (1 : 200 dilution) in blocking buffer for 1 h at 21°C. Then, cells were incubated with avidin–biotinylated horseradish peroxidase complex solution, and subsequently with diaminobenzidine–nickel–H2O2 substrate solution as described by the manufacturer (Vectastain Elite ABC Kit, Vector Laboratories, Inc., Burlingame, CA, USA). Stained cells were viewed with a Nikon Diaphot microscope (Nikon, Inc., Garden City, NY, USA) and MicroComputer Image Device (Imaging Research Inc., St. Catharines, Ontaria, Canada).

Western blot and photoaffinity labeling of hDAT membranes

After being lysed in 2 mm HEPES and 1 mm EDTA for 10 min at 4°C, transfected cells were scraped from plates and centrifuged at 20 000 × g for 20 min at 4°C. The pellet was homogenized with a Brinkmann Polytron (setting 6 for 15 s) in ice-cold PBS and the homogenate was centrifuged at 20 000 g for 12 min at 4°C. The resulting membranes were used either for western blot or photoaffinity labeling. For western blot, the membranes were solubilized in sodium dodecyl sulfate–polyacrylamide gel electrophoreses (SDS–PAGE) sample buffer (62.5 mm Tris-HCl, 2% SDS, 2% 2-β-mercaptoethanol, 10% glycerol, pH 6.8). Sixty micrograms of samples were subjected to 7.5% SDS–PAGE gel electrophoresis. The separated proteins were transferred to nitrocellulose membranes overnight at 4°C. After being blocked with SuperBlock blocking buffer (Pierce, Rockford, IL, USA) for 1 h, the nitrocellulose blots were incubated first with rabbit anti-hDAT polyclonal antibody (0.3 µg/mL) for 5 h at 4°C, and then with horseradish peroxidase-conjugated goat anti-rabbit IgG (0.04 µg/mL) for 1 h at 21°C. The horseradish peroxidase signal was visualized by using the SuperSignal West Pico Chemiluminescence Substrate Kit as suggested (Pierce, Rockford, IL, USA). Immunoblots were quantitated by using a Personal Densitometer SI system (Molecular Dynamics Inc., Sunnyvale, CA, USA). The membrane hDATs were photoaffinity-labeled with [125I]RTI-82 as previously described (Vaughan 1995; Vaughan and Kuhar 1996) with minor modifications. Briefly, the membranes were resuspended in ice-cold uptake buffer to 30 mg protein/mL and incubated for 1 h at 4°C with 5 nm[125I]RTI-82. Membranes were then irradiated with UV light for 45 s, washed twice by centrifugation, and solubilized in SDS–PAGE sample buffer. Equal amounts (60, 120, or 150 µg of initial membrane protein) of samples were electrophoresed on 7.5% SDS–PAGE. Gels were air-dried and then exposed to Kodak BioMax MS film for 3–5 days at −80°C.

Kinetic analysis of transport

The process of the DA uptake can be described in a simplified carrier model consisting of three different steps (Stein 1986; Schömig et al. 1988; Zimányi et al. 1989): (i) free extracellular substrate (So) binds to the free outward-facing carrier (Co) (k1 and k−1 for association and dissociation); (ii) the carrier–substrate complex (CoSo) translocates and releases substrate, leaving the carrier inward-facing (Ci) and substrate inside (Si) (k2 and k−2 for inward and outward movement; (iii) the inward-facing unloaded carrier reorientates (k3 and k−3 for outward and inward movement).

According to this model, we have:

image(1)
image(2)
image(3)

Ks represents the binding affinity of DA to outward-facing hDAT. kt represents the turnover number, which is the function of k2 and k3. Km is the half-saturation concentration for DA uptake. The k−1 for DA binding to the transporter is expected to be extremely rapid. If DA binds to the hDAT with an affinity of about 3 µm as indicated by the potency of DA to compete for CFT binding sites in the present study, and if an association rate of 106 M/s is assumed for DA, as observed for ligands binding to neurotransmitter receptors (Bennett 1978), the k−1 for DA binding to the transporter is 180 min−1 and therefore k−1 >> k2. Also, the chance for empty outward-facing transporter to translocate is assumed to be negligible in the presence of extracellular substrates, and therefore k−3 < < k3. Thus, Equation 3 can be simplified as:

image(4)

From Equations 2 and 4, we can compute:

inline image

From Equation (2), we can compute:

inline image

In the analysis, Ks was the DA Ki for inhibiting [3H]CFT binding, kt was estimated from the ratio of Vmax for [3H]DA uptake to Bmax for [3H]CFT binding.

Results

Mutants lacking plasma membrane immunostaining are not functional

Essentially no immunostaining was observed in cells transfected with pCIN4 vector alone. Cells expressing wild-type hDAT displayed hDAT immunostaining in both plasma membrane and cytosol (Fig. 2). Mutations of E117, W132, W177 and W184 altered the expression pattern. These mutants showed little hDAT immunostaining on the plasma membranes but some immunoreactivity in cytosol, especially the perinuclear area (Fig. 2). Noticeably, all the mutants with no expression on plasma membrane were located in a limited region from the intracellular side of TMD2 (E117Q), the first intracellular loop (W132L), to the second extracellular loop (W177L and W184L) (Fig. 1). At a single concentration of [3H]DA (16 nm) or [3H]CFT (8 nm), These mutants displayed little uptake and binding (Fig. 3). Thus, they were not included in further studies.

Figure 2.

Anti-hDAT antibody immunostaining of HEK-293 cells transfected with either the pCIN4 vector containing no hDAT cDNA, or the pCIN4 vector containing cDNAs encoding the wild-type or mutant hDAT. No staining was observed in mock-transfected cells. Only perinuclear staining was observed in cells expressing E117Q, W132L, W177L, or W184L. Perinuclear staining and cell surface staining was observed in cells expressing wild-type or other mutant hDATs.

Figure 3.

Specific [3H]DA uptake and [3H]CFT binding at a single concentration in wild type and mutant hDATs. Cells were incubated with 16 nm[3H]DA for 5 min or 8 nm[3H]CFT for 15 min at 25°C. Each value represents means ± SEM of 8–16 experiments performed in triplicate. N.D., not detectable. *p < 0.05 (Kruskal–Wallis one way anova followed by Dunn's test for multiple comparisons vs. wild-type group).

All the other mutants showed hDAT immunostaining patterns similar to that of wild-type, although the immunostaining of some mutants did not appear as dense as that of wild-type (Fig. 2). However, a tremendous loss in [3H]DA uptake was observed in W63L and D313N (less than 1% of wild-type), while [3H]CFT binding was only marginally detected in W63L, W311L and D345N (less than 6% of wild-type).

Mutations alter the recognition of DA and cocaine analogs at hDATs

To determine if the conserved tryptophan and acidic residues contribute to recognition of DA and cocaine analogs, we tested [3H]DA uptake and [3H]CFT binding in the presence of various concentrations of unlabeled ligands in 13 mutants that displayed normal or near normal patterns of cell surface immunoreactivity. Although at a single concentration, marginal [3H]DA uptake by W63L, as well as marginal [3H]CFT binding to W63L and D345N, could be observed in the initial experiments, the follow-up experiments showed that these uptake and binding values were too low to detect reliably and hardly inhibited by raising the concentration of unlabeled ligands. Thus, the related parameters for those mutants could not be measured with this approach.

Most mutations changing the Km for [3H]DA uptake also altered DA Ki for inhibiting [3H]CFT binding in the same direction (Table 1). In these mutants, there was a positive correlation between the Km and the Ki values (r = 0.93, p < 0.0001). In D68N and D436N, both DA Km and DA Ki were significantly reduced, suggesting an enhanced binding affinity for DA. Conversely, in D313N and D476N, both DA Km and DA Ki were dramatically increased, suggesting a reduced binding affinity for DA. One exception was W311L. Although DA had a Km value less than 300 nm for this mutant, DA up to 300 µm was unable to inhibit [3H]CFT binding.

Table 1.  Apparent affinity of wild-type and mutant hDATs for dopamine and cocaine analogs
 [3H]Dopamine uptake[3H]CFT binding
hDATDA KmM)Cocaine KI (nM)CFT Kd (nM)DA KiM)
  1. ND, not detectable. Each value represents means ± SEM of 3–12 experiments performed in triplicate. *p < 0.05 vs. wild-type (Dunnett's test following log transformation of data).

Wild-type0.450 ± 0.03837.0 ± 5.207.68 ± 0.942.84 ± 0.42
D68N0.223 ± 0.009*170 ± 13*24.8 ± 8.45*0.95 ± 0.22*
W84L0.595 ± 0.0546.63 ± 1.23*1.82 ± 0.25*1.25 ± 0.15*
E215Q1.411 ± 0.040*46.5 ± 9.6620.1 ± 2.92*3.75 ± 0.70
W267L0.208 ± 0.015*125 ± 13.3*15.4 ± 1.76*1.67 ± 0.41
W311L0.289 ± 0.023*110 ± 13.6*82.2 ± 16.3*> 300 µm
D313N2.788 ± 0.559*11.2 ± 2.15.69 ± 0.7515.2 ± 2.15*
D345N0.093 ± 0.007*127 ± 20.8*NDND
E428Q0.410 ± 0.004100 ± 11.7*28.4 ± 2.58*1.32 ± 0.26
D436N0.094 ± 0.007*128 ± 10.1*24.4 ± 5.49*0.39 ± 0.13*
D476N3.324 ± 0.336*94.7 ± 10.9*30.6 ± 3.94*11.9 ± 1.61*
W520L0.437 ± 0.06270.3 ± 7.7*17.3 ± 3.89*4.14 ± 0.73
W556L0.432 ± 0.040113 ± 13.1*23.4 ± 4.40*4.14 ± 0.66

CFT affinity increased (Kd decreased) by more than four-fold at W84L. In contrast, CFT affinity decreased by 2–4-fold at D68, E215, W267, W428, D436, D476, W520 and W556, and by 10-fold at W311L. To examine whether the high affinity binding of CFT reflects the site at which cocaine inhibit DA uptake, the potency for cocaine to inhibit [3H]DA uptake was measured. Mutations modifying CFT Kd also changed cocaine IC50 for DA uptake in the same direction. However, there was no correlation between the two parameters (r = 0.47, p = 0.12). For instance, W311L showed a nine-fold higher CFT Kd, but only two-fold higher cocaine IC50 for inhibiting DA uptake. Additionally, cocaine inhibited DA uptake by D345N at a potency not weaker than that for many other mutants, although little [3H]CFT binding was observed in D345N (Table 1).

At 4°C, the CFT binding to intact cell expressing wild-type had a Kd of 19.1 ± 1.1 nm and a Bmax of 3.49 ± 0.52 pmol/mg, which was similar to that determined at 25°C. DA displaced [3H]CFT binding completely at both temperatures with similar Ki values (3.39 ± 0.17 µm at 4°C and 2.84 ± 0.42 µm at 25°C). This indicates that [3H]CFT binding to intact cells mainly reflects surface binding, and even at 25°C, DA transport did not play a significant role in the DA potency to inhibit CFT binding. [3H]CFT binding to membrane preparations was also examined in wild-type and some mutants. The binding properties of wild-type and the mutants examined were similar between intact cell and membrane preparations (data not shown). No membrane binding was detected in W63L and D345N mutants.

Mutations shift the relative selectivity of the hDAT for DA and cocaine analogs

Cocaine/DA IC50 ratio for [3H]DA uptake and CFT/DA Ki ratio for [3H]CFT binding were calculated to evaluate if the conserved acidic residues play a role in the relative selectivity of the transporter for DA and cocaine analogs. The ratio for wild-type was set as 1. As shown in Fig. 4, the mutation-induced change in Ki ratio for [3H]CFT binding was roughly in agreement with mutation-induced change in IC50 ratio for [3H]DA uptake. The shift in selectivity appeared opposite between internal and external residues, and remarkable shifts were observed in most aspartate mutants (Fig. 4). D68N and D436N displayed nine- and 16-fold higher ratio in [3H]DA uptake, and nine- and 23-fold higher ratio in [3H]CFT binding, due to both increased DA affinity and decreased cocaine analog affinity. Modest increase in the ratios were also observed in E428Q. Thus, these mutations enhanced the selectivity for DA. Additionally, D345N displayed 16-fold higher IC50 ratio in DA uptake, due to a lower IC50 for DA, but a higher IC50 for cocaine. In contrast, D313N showed almost 20-fold lower ratio in [3H]DA uptake, and 10-fold lower ratio in [3H]CFT binding, mostly due to decreased DA affinity in D313N. Thus, this mutation enhanced the selectivity for cocaine analogs. For E215Q and D476N, a trend similar to that of D313N was also observed in the ratio for DA uptake (Fig. 4).

Figure 4.

Mutations of conserved acidic residues shift the relative selectivity of hDAT for DA and cocaine analogs in uptake and binding assays. Data derived from experiments in Table 1. The ratio was calculated as follows: for uptake, Cocaine/DA IC50 ratio = (IC50,COC /IC50,DA)mutant /(IC50,COC /IC50,DA)wild-type; For binding, CFT/DA Ki ratio = (Kd,CFT /Ki,DA)mutant /(Kd,CFT/Ki,DA)wild-type. The average values of Kd, Ki, or IC50 were used in the equations.

Mutations differentially affect translocation steps

All of the tested mutants showed reduced Vmax values (Table 2), some of which were likely due to their modest expression levels as indicated by immunostaining (Fig. 2). Indeed, the transporter density associated with [3H]CFT binding to intact cells was significantly lower in W84L, E215Q, W267l, D313N and W520L (Table 2). Because the transporter density for D345N could not be obtained from [3H]CFT binding assays, it was calculated by considering the band density (BD) of D345N relative to wild-type in western blots as follows: BDD345N/BDwild-typeBmax,wild-type. The estimated transporter density for D345N was 1.43 pmol/mg.

Table 2.  Transport characteristics of wild-type and mutant hDATs
hDATVmax (pmol/mg/min)Bmax (pmol/mg)kt (/min)kt/Km (s/M/ 105)k2 (/min)k3 (/min)
  1. Vmax, maximal velocity of [3H]DA uptake; Bmax, maximal capacity of [3H]CFT binding; kt, turnover number; kt/Km, apparent second-order rate constant when substrate concentration is far below the Km; k2, first-order rate constant for loaded transporters to translocate inwardly and release substrates inside; k3, first-order rate constant for unloaded transporters to reorientate. ND, not detectable. Each value represents means ± SEM of 3–12 experiments. aBased on western blot results, assume the density of D345N hDAT is 1.43 pmol/mg (see Results). *p < 0.05 vs. wild-type (Dunnett's test following log transformation of data).

Wild-type17.7 ± 1.423.19 ± 0.335.56 ± 0.442.16 ± 0.2539.6 ± 4.206.62 ± 0.52
D68N10.1 ± 0.38*2.17 ± 0.494.65 ± 0.183.48 ± 0.0819.8 ± 0.436.08 ± 0.30
W84L4.23 ± 0.38*1.35 ± 0.11*3.14 ± 0.28*0.88 ± 0.026.59 ± 0.11*6.15 ± 1.02
E215Q6.87 ± 0.18*0.74 ± 0.12*9.28 ± 0.251.10 ± 0.0124.7 ± 0.0514.9 ± 0.66*
W267L3.52 ± 0.27*1.64 ± 0.15*2.14 ± 0.16*1.77 ± 0.2017.7 ± 1.952.45 ± 0.19*
W311L0.82 ± 0.11*2.30 ± 0.530.36 ± 0.05*0.21 ± 0.02*NDND
D313N0.64 ± 0.13*0.72 ± 0.07*0.89 ± 0.18*0.05 ± 0.008*4.83 ± 0.25*1.13 ± 0.26*
D345N1.56 ± 0.07*ND1.09 ± 0.05a*1.67 ± 0.12NDND
E428Q13.0 ± 0.18*2.33 ± 0.265.57 ± 0.082.27 ± 0.0417.9 ± 0.328.07 ± 0.11
D436N3.59 ± 0.24*2.12 ± 0.471.69 ± 0.12*2.93 ± 0.036.86 ± 0.06*2.26 ± 0.21*
D476N28.3 ± 2.29*2.18 ± 0.3312.9 ± 1.05*0.66 ± 0.0446.8 ± 3.0618.1 ± 1.96*
W520L1.88 ± 0.23*1.88 ± 0.24*1.00 ± 0.12*0.39 ± 0.035*9.71 ± 0.87*1.13 ± 0.15*
W556L4.05 ± 0.21*2.79 ± 0.241.45 ± 0.08*0.59 ± 0.07314.7 ± 1.831.62 ± 0.08*

The value of kt, estimated based on the ratio of Vmax/Bmax, was reduced by more than 10-fold at W311L, and by 5–6-fold at D313N, D345N, W520L and W556L (Table 2). The kt/Km can be considered as an apparent second-order rate constant when substrate concentration is far below Km (Fersht 1985). In all tested hDATs, the kt/Km values were below the range of 106−108 /sM, indicating steps after substrate binding, probably the interconversion between outward-facing and inward-facing forms of the transporter, is the rate-limiting step during transport process (Stein 1986). Further dissecting the kt value revealed that the reorientation of inward-facing transporter (k3) was the slowest step and dominated the estimation of turnover rate kt(Table 2), at least, under the present conditions. The mutations differentially affected the k2 and k3: reducing both k2 and k3 for D313N, D436N and W520L, only k2 for W84L, and mainly k3 for W267L and W556L. Although k2 and k3 values could not be analyzed for W311L and D345N, their low kt values indicated damaged reorientation of the unloaded transporter, i.e. reduced k3, because kt values largely reflected k3 values (Table 2).

Calculation of k2 and k3 was based on the assumptions of k−1 >> k2 and k−3 << k3. The assumptions may still held true for D68N, W84L, E215Q, W267L, E428Q, W520L and W556L, because their affinities for DA were not far different from that of wild-type. However, for mutants in which the DA affinity was substantially altered, k2 and k3 values should be interpreted with caution. Higher DA affinity might be due to a reduced k−1 and therefore k−1 >> k2 may not apply; lower DA affinity might allow more free transporters to turn to the inside, which would not support the assumption of k−3 >> k3.

Cocaine analogs effectively inhibit DA uptake at D345N with a modified mechanism

Cocaine, CFT, RTI-121 and mazindol remained relatively potent in preventing D345N from taking up DA (Table 3). The kinetic mechanism by which CFT inhibited DA uptake was explored by saturation analysis of DA uptake in the absence and presence of CFT. CFT primarily affected the Km for wild-type, but affected both Km and Vmax for D345N (Table 4), suggesting different mechanisms of inhibition. Although CFT raised Km for D345N, the effect was significantly smaller than that on wild-type (six-fold in D345N vs. 30-fold in wild-type at 200 nm).

Table 3.  Inhibition of dopamine uptake by various hDAT blockers: comparison of wild-type and D345N hDATs
 Wild-typeD345N
BlockersIC50 (nM)Hill numberIC50 (nM)Hill number
  1. Each value represents means ± SEM of 2–4 experiments performed in triplicate. ND, not determined. *p < 0.05 vs. wild-type (t-test).

Cocaine37 ± 4.90.85 ± 0.06131 ± 17*0.92 ± 0.08
CFT8.9 ± 1.10.98 ± 0.0223.2 ± 1.2*0.92 ± 0.03
Mazindol7.6 ± 0.21.25 ± 0.0514.3 ± 0.3*1.28 ± 0.08
RTI-121NDND11.8 ± 0.70.95 ± 0.05
Table 4.  Effect of CFT on kinetic parameters of DA uptake: comparison of wild-type and D345N hDATs
 Wild-typeD345N
CFT (nM)Vmax (pmol/mg/min)Km (nM)Vmax (pmol/mg/min)Km (nM)
  1. Each value represents mean ± SEM of 4–10 experiments performed in triplicate. *p < 0.05 vs. 0 nm CFT (Dunnett's test after log transformation of data).

015.94 ± 1.94460 ± 351.51 ± 0.0784 ± 6
2019.13 ± 4.251288 ± 97*1.17 ± 0.10*115 ± 6*
20014.45 ± 0.7813734 ± 394*0.94 ± 0.12*519 ± 31*

[3H]CFT and [3H]mazindol do not bind to D345N

For D345N, an effort was made to measure the binding by using increasing concentrations of [3H]CFT. In these assays, the final CFT concentration varied from 4 to 300 nm, constructed by mixing [3H]CFT and unlabeled CFT at either 1 : 1 or 1 : 9 ratio. Both methods gave similar results. With cocaine (100 µm) to define the non-specific binding, ‘specific binding’ was observed at each [3H]CFT concentration tested. Because cocaine also inhibited [3H]CFT binding to intact cells expressing pCIN4 alone (− 50% at 30 µm), we turned to using cells expressing pCIN4 vector alone to define the non-specific binding. With the latter approach, specific binding was only observed at [3H]CFT concentration higher than 200 nm. Either approach produced an unsaturable concentration-response curve for the ‘specific binding’. The Kd and Bmax values could not be reliably fitted from these curves. The data also suggests that cocaine-definition of non-specific binding caused overestimation of the specific binding to intact cells, especially when the binding level of the mutants was low.

Since cocaine analogs inhibited DA uptake by D345N effectively, one possibility is that D345N displayed conformational states during the uptake cycle that had a high affinity for cocaine analogs, whereas such conformational states were not prevalent under the conditions of the binding assay. To explore this possibility, we examined the [3H]CFT binding in the presence of DA (10 nm) at a condition identical to that for DA uptake. This approach did not improve [3H]CFT binding (Fig. 5). Changes in hDAT structure may alter the association and dissociation rates such that the binding cannot be measured at 25°C. In an attempt to slow down the dissociation rate, we incubated the cells with [3H]CFT at 4°C for 2 h and terminated the reaction by either plate harvester or more rapid single filtration. These procedures failed to improve [3H]CFT binding (Fig. 5).

Figure 5.

[3H]CFT or [3H]mazindol binding to intact cells expressing wild-type or D345N under different conditions. Cells were incubated with [3H]CFT (2 nm) for 15 min at 25°C, or with [3H]CFT (4 nm) or [3H]mazindol (4 nm) for 2 h at 4°C. Each value represents means ± SEM of 4–8 experiments performed in triplicate.

Evidence from binding studies has suggested the involvement of common binding domains for CFT and mazindol (Xu and Reith 1997). Thus, we also determined whether mutation of D345 affected [3H]mazindol binding. [3H]mazindol binding to intact cells was performed at 4°C for wild-type and D345N. For wild-type hDAT, the Kd was 5.88 ± 2.47 nm and the Bmax was 3.04 ± 0.67 pmol/mg. DA displaced [3H]mazindol with a Ki value of 2.23 ± 1.04 µm. However, D345N did not bind [3H]mazindol, either (Fig. 5). Therefore, the effect of D345N mutation on mazindol and CFT binding is similar.

A photoaffinity cocaine derivative, [125I]RTI-82, does not label D345N proteins

Irreversible binding of a cocaine analog, [125I]RTI-82, was examined using total membrane preparations. In western blot for wild-type hDAT (Fig. 6), the antibody recognized two bands; the intense band between 66 and 97 kDa is the totally processed transporter present at plasma membrane, and the lower molecular weight band at approximately 50 kDa may correspond to the immature protein (Daniels and Amara 1999). D345N also exhibited immunoreactive bands at both positions with the higher molecular weight band as the major form, confirming that full-length and totally processed transporters were produced in D345N. With equal amounts of membrane sample, the D345N protein was quantified by densitometry at 45% relative to the wild-type protein. In photoaffinity labeling studies on cells expressing wild-type hDAT, [125I]RTI-82 labeled the proteins migrating between 66 and 97 kDa (Fig. 6). The [125I]RTI-82 labeling of wild-type can be blocked by coincubation with 1 mm DA (data not shown). However, no specific [125I]RTI-82 labeled proteins were observed in D345N (Fig. 6). Increasing the D345N membrane protein for SDS–PAGE up to 150 µg enhanced the background but failed to show any specific band labeled by [125I]RTI-82 (data not shown). No transporter was detected in cells transfected with vector alone in the immunoblotting and photoaffinity labeling studies (Fig. 6).

Figure 6.

Immunoblotting and photoaffinity labeling of wild-type and D345N. (a) Immunoblots with hDAT specific antibody. (b) Photoaffinity labeling with [125I]RTI-82. The total membrane protein for each lane was 60 µg. The positions of the molecular weight standards are indicated at the left. Data presented are representative of three separate experiments.

Discussion

Mutation of four conserved residues, E117Q, W132L, W177 and W184, disrupted plasma membrane expression, causing little uptake and binding activity. Given the absolute conservation of these four residues in 26 Na+/Cl-dependent transporters, our findings might generalize to other Na+/Cl-dependent transporters. Indeed, charge-neutralizing mutations of the E117 cognate in GABA transporters also causes no function (Keshet et al. 1995), although the membrane targeting was not reported. W177 and W184 each precedes one canonical N-glycosylation site (N181 and N188) by three residues. The two canonical N-glycosylation sites may be required for cell surface expression as reported for norepinephrine transporters (Melikian et al. 1996; Nguyen and Amara 1996). It is possible that the mutations altered the local environment of amino acids or the protein folding patterns, thereby affecting the glycosylation and membrane targeting (Opdenakker et al. 1993).

Obviously, a role common for the other six conserved tryptophans is their contribution to both DA translocation and cocaine analog binding, although in different manners. Tryptophans have been suggested to be especially prominent at cation–π interactions (Dougherty 1996; Gallivan and Dougherty 1999). The conserved tryptophans, preferentially bordering hydrophobic regions and neighboring with cationic residues (W63 with R60, K65 and K66, W84 with R85, W267 with K264 and W520 with R517 and K521), might interact with intramolecular residues to mediate the hDAT oscillation between inward- and outward-facing states. Thus, their loss could lock the transporter in one state or another. Evidence has recently been advanced that the same transporter can drive substrates in and out through separate pathways participated by different residues (Lanyi 1997; Chen and Justice 2000). In this regard, W84 seems to mainly contribute to inward orientation (markedly reduced k2 after mutation), whereas W267, W311, W520 and W556 appear more involved in outward orientation of the transporter (substantially low kt after mutation). Additionally, deletion or mutation to non-aromatic residues of the tryptophan aligned with W63 in GABA transporter impairs reorientation of the unloaded transporter (Bennett et al. 2000), indicating a similar consequence might also occur at W63L hDAT. Interestingly, cocaine analogs showed higher affinity for W84L in which the hDAT was mainly arrested at an outward-facing state, but lower affinity for the W-L mutants in which the hDAT seems more readily trapped in an inward-facing state. This is in line with the notion that the transporter conformation for cocaine binding is outward facing (Chen and Justice 1998).

In rat DAT (rDAT), five cognates (W84, W266, W310, W523 and W561) of the tryptophans targeted in our study (W84, W267, W311, W520 and W556, respectively) have recently been examined by alanine substitution (Lin et al. 2000). Although most of results from the two studies are comparable, there are interesting differences between W311L hDAT (present results) and W310A rDAT (Lin et al. 2000). Thus, in CFT binding assays, W311L hDAT displayed low binding affinity for both cocaine analogs and DA while W310A rDAT displayed high binding affinity for both. This discrepancy could be due to differences in amino acids (leucine vs. alanine) used for substitution and temperatures (25°C for both vs. 4°C for binding and 37°C for uptake) used for assays. The size and hydrophobicity of alanine are substantially lower than tryptophan while those of leucine are relatively compatible. Thus, the mutation-induced changes in the structural or hydrophobic features of the hDAT might differ. Indeed, the expression level of W311L hDAT is better than that of W310A rDAT. It is also possible that the mutation could induce temperature-sensitive changes in conformation. Both possibilities could impact ligand binding. Nevertheless, this conserved residue seems important for recognition of both cocaine and DA in that substitution with either alanine or leucine results in dramatical changes in ligand affinity. Despite the difference, W311L hDAT and W310A rDAT share a common character: dissociation between Km for DA uptake and Ki for DA to inhibit CFT binding, though in different direction. This dissociation could be interpreted as the mutation-induced separation between binding sites for DA and CFT, i.e. DA binding affinity at the transporter no longer equals the Ki for DA in competition for [3H]CFT. Alternatively, there are extreme changes in translocation rates as deduced from the description Km =Ksk3/(k2 + k3), which offsets the value of Ks.

The effect of mutating conserved aspartates seems to depend on their locations. Substitutions of D313 and D476 at the external surface reduced DA affinity, whereas substitutions of D68 and D436 at the internal surface enhanced DA affinity. After replacing another intracellular aspartate, D345, we observed a lower Km and considerable uptake especially at lower [3H]DA concentrations, which might also be due to an enhanced DA affinity. These results may point out a significant difference in the role of conserved aspartates between external and internal residues. Thus, the external aspartates could be preferentially involved in recognition of external DA, while the internal aspartates might be important for stabilizing a hDAT state binding DA weakly. A weak DA binding could be an advantage for subsequent release of DA inside. Four out of five mutations in conserved aspartates (D68N, D313N, D345N and D436N) displayed striking shift in the relative selectivity of the hDAT for DA and cocaine analogs, strongly suggesting that the conserved aspartates are critical in guiding the hDAT recognition of DA and CFT and that the binding domains for DA and CFT are not identical.

The most intriguing observation is that mutant D345N does not appear to bind cocaine analogs but cocaine analogs still inhibit DA uptake by this mutant. The loss of CFT binding was not caused by improper expression or post-translational processing of the protein, not temperature-dependent, and not alleviated by substrate-induced conformational changes. In spite of the small Vmax value, at low levels of [3H]DA (10 nm), DA uptake by D345N was approximately 30% of wild-type under control condition and 140% of wild-type in the presence of 200 nm CFT. Considering its lower expression level (45% of wild-type), this uptake ability is even more impressive. Thus, it also seems less likely that the structure of D345N was changed drastically. It is possible that [3H]CFT could bind to D345N but could not be detected because: (i) the bound ligands dissociate too fast to endure any separation steps used in equilibrium binding assays; (ii) the weak β emission from [3H] only allows detection of the bound [3H]signals in a particular conformation that has been affected by the mutation; and (iii) there are isotopic artifacts confounding the results. To exclude these possibilities, we performed covalent labeling with an irreversible cocaine-like inhibitor, [125I]RTI-82. The observation that [125I]RTI-82 failed to label any of the D345N proteins strongly argues against the above possibilities and reinforces the speculation that cocaine analogs do not bind to D345N. Noteworthily, two residues below D345 is a cytoplasmic cysteine (C342), which displays transport-dependent accessibility to the membrane permeable sulfhydryl reagent (2-aminoethyl)methanethiosulfonate (MTSEA) (Chen et al. 2000). Thus, the region in which D345 is located may participate in conformational changes. Since the binding of cocaine analogs is conformation-dependent, it is tempting to speculate that the mutation of D345 converts hDAT to a partially active state, which loses its high affinity for cocaine.

Although not binding to D345N, cocaine analogs still potently inhibited DA uptake through a mechanism deviated from that for wild-type (mixed inhibition vs. competitive inhibition). This questions whether the binding of cocaine to hDAT is solely responsible for its inhibition on DA uptake. It has been noted that cocaine analogs can potently inhibit DA uptake even though the high affinity of [3H]CFT binding is severely compromised (Lee et al. 1996) or absent (Sugamori et al. 1999), although the lack of binding in the latter study with naive COS-7 cells could be due to an exceedingly low density of the transporter. A study on norepinephrine transporters suggests that the transporters may occasionally form channel-like pores that substrates can readily move through (Galli et al. 1998). If the D345N behaves like a channel, cocaine analogs could compete with DA to flow through the same pores or physically obstruct the entry of DA into the pores. It should also be recalled that DA transport might be regulated by multiple intracellular signal transduction pathways (Reith 2000; Vaughan 2000). Interactions of cocaine analogs with intracellular signal transduction might diminish DA uptake, as well. [3H]CFT binding to intact cells has showed significant ‘background’ level to sites other than hDAT but blocked by cocaine analogs (Reith et al. 1996) (present data). These sites could be unknown cocaine targets. Evidently, further clarification of the dissociation between binding to the hDAT and inhibin of DA uptake is of great significance to ongoing effort in developing anticocaine medications.

For transporters, aromatic and acidic residues have been implicated in direct interactions with cationic neurotransmitters (Kitayama et al. 1992; Barker et al. 1999; Lin et al. 1999; Lin et al. 2000), alkali ion binding (Poolman et al. 1996; Martin et al. 1999), and proton binding/transfer (Steiner-Mordoch et al. 1996; Lanyi 1997). The fact that the residues tested here are conserved in the monoamine transporters as well as in the other members of the Na+/Cl-dependent transporters makes it more plausible that these residues interact with cations, protons, or intramolecular cationic residues, to trigger conformational changes. Of particular interest are the results with W311L, D313N and D345N, as well as the activity pattern of aspartate mutants. These mutants displayed dissociation between dopamine binding/uptake and cocaine analog binding, which might provide insight into regulating factors or structural determinants specific for either process. Thus, investigation of the potential regulatory mechanisms underlying the observed dissociation is in progress. It will also be important in future studies to systematically investigate the effect of conservative and non-conservative mutations at and nearby these locations to elucidate the molecular basis for this dissociation.

Acknowledgements

We thank Dr Jonathan A. Javitch for the gifts of the pCIN4 vector and the human dopamine transporter cDNA. This work was supported by NIDA grants DA 08379 and DA 13261 (to MEAR).

Ancillary