The binding sites for cocaine and dopamine in the dopamine transporter overlap

Abstract

Cocaine is a widely abused substance with psychostimulant effects that are attributed to inhibition of the dopamine transporter (DAT). We present molecular models for DAT binding of cocaine and cocaine analogs constructed from the high-resolution structure of the bacterial transporter homolog LeuT. Our models suggest that the binding site for cocaine and cocaine analogs is deeply buried between transmembrane segments 1, 3, 6 and 8, and overlaps with the binding sites for the substrates dopamine and amphetamine, as well as for benztropine-like DAT inhibitors. We validated our models by detailed mutagenesis and by trapping the radiolabeled cocaine analog [3H]CFT in the transporter, either by cross-linking engineered cysteines or with an engineered Zn2+-binding site that was situated extracellularly to the predicted common binding pocket. Our data demonstrate the molecular basis for the competitive inhibition of dopamine transport by cocaine.

Figure 1

Models of DAT/ligand complexes. (a) Two-dimensional schematic representation of the human dopamine transporter (hDAT). Colored circles indicate residues that interact with either dopamine or the cocaine analog CFT in the molecular models. Red circles indicate side-chain interaction and orange circles indicate backbone interaction. (b) Structure of dopamine, cocaine and CFT. (c,d) Docked dopamine (c) and CFT (d) in DAT. Transmembrane domains 1, 3, 6 and 8 are shown in various shades of blue; the other transmembrane domains and intra- and extracellular loops have been removed for clarity. The ligands are shown in green. Sodium and chloride ions are shown as purple and salmon spheres, respectively. Encircled numbers refer to the specific interactions motifs. Motif 1 (c), salt bridge interaction between Asp79 and the protonated amine of dopamine; motif 1 (d), polar interaction between Asp79 and the amine of CFT; motif 2, aromatic-aromatic stacking interaction between Tyr156 and the catechol ring of dopamine; motif 3, hydrogen bond between the OH group of Tyr156 and Asp79; motif 4, hydrophobic-aromatic interaction between Leu80 and Tyr156; motif 5, interaction between Tyr156 and the 2β-methylester substituent of CFT; motif 6, interaction between the nitrogen of Asn157 and the fluoride atom of CFT.

Figure 2

Evidence for involvement of Val152, Tyr156, Asn157, Val328 and Ser422 in binding of dopamine and/or CFT by DAT. (a) [³H]dopamine uptake (shown as saturation curve) in COS7 cells expressing wild-type DAT (■), V152M (○), V328I (▼) or S422A (□). (b) Inhibition of [³H]dopamine uptake by CFT in COS7 cells expressing wild-type DAT (■), V152M (○), V328I (▼) or S422A (□). (c) [³H]dopamine uptake (shown as saturation curve) in COS7 cells expressing wild-type DAT (■), Y156F(▲) or N157C (◇). (d) [³H]CFT binding (shown as saturation curve) in COS7 cells expressing wild-type DAT (■), Y156F(▲) or N157C (◇). (e) [³H]dopamine uptake (shown as saturation curve normalized to maximum uptake) in COS7 cells expressing wild-type DAT (■), Y156F(▲) or N157C (◇). (f) [³H]CFT binding (shown as saturation curve normalized to maximum binding) in COS7 cells expressing wild-type DAT (■), Y156F(▲) or N157C (◇). All data are means ± s.e.m. of 3–8 experiments carried out in triplicate. The saturation curves shown are transformations of the [³H]dopamine uptake and [³H]CFT binding competition assays that were used for calculation of the K_M and V_MAX values for [³H]dopamine and the K_D and B_MAX values for [³H]CFT (Supplementary Table 1).

Figure 3

Validation of the CFT docking model using intramolecular cross-linking and cysteine-reactive reagents. (a) Engineering of a Zn²⁺-binding site between TMDs 1 and 3 in an extracellular vestibule above the predicted binding site resulted in partial trapping of [³H]CFT in its binding site. Left, model of CFT docked into L80H-I159C with subsequent docking of a Zn²⁺ ion in the site created by the mutated residues. Zn²⁺ is shown as a purple sphere. The Zn²⁺ binding site is positioned in the same locus of the transporter as the noncompetitive binding site for tricyclic antidepressants in LeuT, (illustrated by a spherical model of clomipramine bound to LeuT). CFT is shown in green. Middle, dissociation of prebound [³H]CFT in COS7 cells expressing the double mutant L80H-I159C in the background construct E2C H193K with (■) or without (○) 100 μM Zn²⁺. Right, effect of Zn²⁺ (100 μM) on [³H]CFT off-rate in the DAT background construct E2C H193K (named ‘background’) and in the mutations L80H-I159C, L80H, I159C, L80H-I159A or L80K-I159C (all made in E2C H193K). (b) Cysteine-mediated cross-linking between TMDs 1 and 3 resulted in partial trapping of [³H]CFT in its binding site. Left, model of CFT docked into W84C-I159C with subsequent cross-linking of MTS-3-MTS. Middle, dissociation of prebound [³H]CFT in COS7 cells expressing the double mutant W84C-I159C in the E2C background with (■) or without (○) 0.5 mM MTS-3-MTS. Right, effect of 0.5 mM MTS-3-MTS on [³H]CFT off-rate in the DAT background construct E2C (named ‘background’) and in the mutations W84C-I159C, W84C, I159C, W84C-I159A or W84A-I159C (all made in E2C). (c) Slower dissociation of [³H]CFT by reacting I159C or W84C with the bulky MTS-reagent BZ-MTS. Left, model of CFT docked into I159C with subsequent reaction of I159C with BZ-MTS. Middle, dissociation of prebound [³H]CFT in COS7 cells expressing the mutant I159C in the background construct E2C with (■) or without (○) 0.5 mM BZ-MTS. Right, effect of BZ-MTS on [³H]CFT off-rate in the DAT background construct E2C (named ‘background’) and in the mutations I159C, I159A, W84C or W84A. All data in middle panels are percentages of bound [³H]CFT at t = 0 (means ± s.e.m., n =3–5). All data in right panels are the ratio between the t½ (min) for [³H]CFT dissociation with and without Zn²⁺ or MTS reagent (means ± s.e.m., n = 3). The [³H]CFT off-rates (t½in min) in the absence of Zn²⁺ or MTS reagent are given in the Supplementary Note. * P < 0.05), *** P < 0.001; one-way ANOVA with Newman-Keuls multiple comparison post hoc test.

Figure 4

Molecular docking models of cocaine, amphetamine (AMPH) and JHW007 with experimental validation. (a,c,d) Models of cocaine, amphetamine and JHW 007 docked in DAT, shown in the plane of the membrane. Transmembrane domains 1, 3, 6 and 8 are shown in various shades of blue; the other transmembrane domains and intra- and extracellular loops have been removed for clarity. The ligands are shown in green and interacting residues in yellow and brown. Sodium and chloride ions are shown in purple and salmon spheres, respectively. Note that the docking of cocaine does not allow the formation of a hydrogen bond between Asp79 and Tyr156, whereas both AMPH and JHW 007 allow this interaction to occur. (b,d,f) Inhibition of [³H]CFT binding by cocaine (b), amphetamine (d) and JHW 007 (f) in COS7 cells expressing wild-type DAT (■) or DAT Y156F (○). Data are means ± s.e.m. of 3–4 experiments carried out in triplicate.

Figure 5

Comparison of the CFT-binding mode with the binding mode of dopamine and other DAT ligands. (a) Chemical structures of all compounds docked into the DAT model. Compounds that hindered the Asp79-Tyr156 hydrogen bond when docked are shown in black; compounds that allowed the hydrogen bond are shown in gray. (b) Computationally estimated distance between the oxygen atoms of Tyr156 and Asp79 plotted against ligand Y156F/wild-type affinity ratio (values from Supplementary Table 2). The time-dependent behavior of the Tyr-Asp hydrogen bond was explored with molecular dynamics simulations. The hydrogen bond was continuously present (distance <3.5 Å) in all trajectories (gray) except for those of CFT and cocaine (black). Error bars are standard deviations of the estimated distances. Dotted lines indicate the minimal distance for hydrogen-bond formation (3.5 Å, horizontal) or an arbitrary fivefold affinity ratio (vertical). Inset, model of the Tyr-Asp interaction seen for the two classes of ligands. (c) Rendering of water permeation into the dopamine- (top) and CFT-binding (bottom) sites. Right, the distribution of water shown over a 25 Å distance along the membrane normal in the region indicated on the left by the rectangular box. The distribution was calculated from the last 5 ns of each of three separate simulations carried out for each ligand-DAT complex (red, blue and black traces, respectively). The locations of the geometric centers of Tyr156 and Asp79 and ligands at the end of the simulations are indicated by blue and green arrows, respectively, and water molecule numbers are indicated as a function of distance. Left, water distribution in context of the structure; silver-colored surfaces indicate the water penetration, with the structures of Tyr156, Asp79 and the ligands rendered.