Structure of the human dopamine transporter and mechanisms of inhibition

Abstract

The neurotransmitter dopamine has central roles in mood, appetite, arousal and movement¹. Despite its importance in brain physiology and function, and as a target for illicit and therapeutic drugs, the human dopamine transporter (hDAT) and mechanisms by which it is inhibited by small molecules and Zn²⁺ are without a high-resolution structural context. Here we determine the structure of hDAT in a tripartite complex with the competitive inhibitor and cocaine analogue, (-)-2-β-carbomethoxy-3-β-(4-fluorophenyl)tropane² (β-CFT), the non-competitive inhibitor MRS7292³ and Zn²⁺ (ref. ⁴). We show how β-CFT occupies the central site, approximately halfway across the membrane, stabilizing the transporter in an outward-open conformation. MRS7292 binds to a structurally uncharacterized allosteric site, adjacent to the extracellular vestibule, sequestered underneath the extracellular loop 4 (EL4) and adjacent to transmembrane helix 1b (TM1b), acting as a wedge, precluding movement of TM1b and closure of the extracellular gate. A Zn²⁺ ion further stabilizes the outward-facing conformation by coupling EL4 to EL2, TM7 and TM8, thus providing specific insights into how Zn²⁺ restrains the movement of EL4 relative to EL2 and inhibits transport activity.

Fig. 1. Function and architecture of inhibitor-bound ∆-hDAT.

a, Saturation uptake of [³H]dopamine in HEK293 GnTI⁻ cells expressing ∆-hDAT (black) and full-length hDAT (blue). Uptake in the presence of 10 µM MRS7292 is shown in orange and pink for ∆-hDAT and full-length hDAT, respectively. The Michaelis constant (K_m) values for [³H]dopamine uptake by ∆-hDAT and full-length hDAT were 0.55 ± 0.07 and 0.56 ± 0.13 μM, and reaction rate at infinite substrate concentration (V_max) values were 342.8 ± 11.7 and 190.9 ± 10.7 fmol min⁻¹ per well, respectively. Data were analysed using a Michaelis–Menten kinetics model. The uptake assay was performed in n = 3 biological replicates, with each in technical triplicate. Data are mean ± s.d. b, Scintillation proximity assay (SPA) using [³H]WIN35428 and purified His-tagged ∆-hDAT. The dissociation constant (K_d) for [³H]WIN35428 binding by ∆-hDAT was 6.5 ± 0.91 nM. Data are mean ± s.d. Assays were done in n = 3 independent replicates, each with technical triplicates. c, Structure of ∆-hDAT showing β-CFT in the central site and MRS7292 in the allosteric site. NAG represents an N-acetylglucosamine modification at N188. d, Slab view of ∆-hDAT in a surface representation showing how the transporter adopts an outward-open conformation. e, Chemical structure of β-CFT (prepared using ChemDraw 18.2). f, Density associated with β-CFT, contoured at 10σ, within 2 Å of the ligand atoms. g, Close-up representation of β-CFT bound to the central site. Hydrogen-bonding interactions are shown as black, dashed lines. Source Data

Fig. 2. Delineation of the MRS7292 allosteric site.

a, MRS7292 binds underneath EL4 and adjacent to TM1b, in a hydrophobic pocket additionally defined by TM7. Key residues in the MRS site are indicated in stick representation with the carbon atoms of MRS7292 in orange. Subsites I, II and III are shown. b, Chemical structure of MRS7292 (prepared using ChemDraw 18.2). c, Density associated with MRS7292, contoured at 12σ within 2 Å of the ligand atoms. d, Effect of MRS7292 on [³H]dopamine uptake in ∆-hDAT and ∆-hDAT mutants. Data were analysed using nonlinear regression (Methods). IC₅₀ measurements were performed in n = 3 biological replicates (each in technical triplicate). Data are mean ± s.d. e, Superposition of the MRS site in ∆-hDAT and the equivalent site in hSERT (PDB: 7LIA) using α-carbon atoms of TM3 and TM8. f, Close-up view of the MRS site showing the cryo-EM model and the final conformations of the simulation replicas. Source Data

Fig. 3. The zinc site bridges EL2 and EL4.

a, Location and close-up view of the Zn²⁺ site showing the multivalent coordination of the divalent ion. The Zn–O and Zn–N distances are expressed in Å. b, Superposition of the Zn²⁺ site in ∆-hDAT and a predicted model of ∆-hDAT in an inward-open state showing the displacement of H375 on EL4 by 4.2 Å between the outward-open to inward-open conformations. TM3 and TM8 were aligned using α-carbon atoms. c, Characterization of [³H]dopamine uptake in HEK293 GnTI⁻ cells expressing ∆-hDAT and the mutants T211E and T211H in the presence of Zn²⁺. Data were analysed using a nonlinear regression model (Methods). The experiments were performed in n = 3 biological replicates with each in technical triplicate. Data are mean ± s.d. Source Data

Fig. 4. Mechanisms of inhibition.

β-CFT binds at the central or S1 site, stabilizing the outward-open conformation of ∆-hDAT. Binding of MRS7292 at the allosteric binding pocket near TM1b and below EL4 leads to a conformational change in TM1b, TM6 and EL4 that precludes their movement, further stabilizing the outward-open conformation and enhancing β-CFT binding at the central site. Binding of Zn²⁺ restrains the EL4 loop, precluding its movement upon transition to the inward-open conformation, thus inhibiting transport.

Extended Data Fig. 1. Biochemical characterization of ∆-hDAT.

(a) Fluorescence-detection size-exclusion chromatography (FSEC) of purified ∆-hDAT protein. (b) SDS-PAGE analysis of the FSEC sample was done once and visualized by silver staining. See Supplementary Fig. 1 for gel. Source Data

Extended Data Fig. 2. Cryo-EM image processing and analysis.

(a) Cryo-EM workflow. A detailed description of the image processing steps and parameters is included in the Methods section. Scale bar = 50 nm. (b) Local resolution map along with (c) the gold standard Fourier shell correlation (GSFSC) curve and (d) angular sampling of the cryo-EM reconstruction. (e) FSC_work/FSC_free and map versus model curve. Source Data

Extended Data Fig. 3. Density associated with transmembrane helices, EL4, and the C-terminal latch.

Isomesh map features are contoured at 8 σ and within 2 Å of the atoms associated with each feature.

Extended Data Fig. 4. C-terminal latch of ∆-hDAT and structural comparison with other NSS transporters.

(a) Comparison of the C-terminal latch of ∆-hDAT with the human serotonin transporter (hSERT; PDB: 7LIA), human GABA transporter (GAT1; PDB: 7SK2), and human glycine transporter (GlyT1, PDB: 6ZBV). The structures were aligned using α-carbon atoms. (b) Close up view of the C-terminal latch of ∆-hDAT showing the proximity of TM12, TM3, and TM10.

Extended Data Fig. 5. Lipid and lipid-like features in the ∆-hDAT reconstruction.

(a) Overall distribution of lipid-like features in the cryo-EM reconstruction of ∆-hDAT. CHS and alkyl chains of possible lipid molecules are represented as orange and magenta sticks, respectively. The associated cryo-EM density features are shown in isomesh map representation contoured at 8 σ within 2 Å of the atoms of putative CHS and alkyl chain molecules. (b) Structural alignment of ∆-hDAT with dDAT (PDB: 4XPG) and (c) with hSERT (PDB: 5I73) showing the positions of CHS or cholesterol molecules in the proximity of lipid molecules in ∆-hDAT. Structural superposition was done using α-carbon atoms of the whole structures. (d) Additional putative CHS molecule in a groove formed by TMs 4, 5 and 8.

Extended Data Fig. 6. Central binding site of ∆-hDAT and comparison with related transporters.

(a) Overlay of the central binding pockets of S-citalopram-bound hSERT (5I73) and β-CFT-bound ∆-hDAT. The structural alignment was done using α-carbon atoms of TM3 and TM8 of ∆-hDAT. (b) Superposition of the central sites occupied by β-CFT in dDAT (PDB: 4XPG), cocaine in pSERT (PDB: 8DE3) and β-CFT in ∆-hDAT. (c) Key phenylalanines ‘above’ the central sites in pSERT (PDB: 8DE3) and dDAT (PDB: 4XPG) that participate in defining the occluded or outward open states. The gray sphere represents the center of mass of β-CFT in ∆-hDAT. (d) (d) Superposition of the central site showing the relative pose of β-CFT in the cryo-EM reconstruction of ∆-hDAT and five replicas of MD simulation derived poses. All structural superpositions in (b)-(d) were done using α-carbon atoms of whole structures. (e) Coordination of a Na⁺ ion at the Na2 site. The distances between the Na⁺ ion and the coordinating oxygen atoms of the likely interacting residues are shown in yellow dashed lines and represented in Å.

Extended Data Fig. 7. The MRS7292 site in ∆-hDAT and structural comparison with hSERT and dDAT.

(a) The locations of allosteric sites for the respective transporter complexes. The allosteric ligands are shown in sphere representation, and the central site molecules are shown as sticks. (b) Illustration of the MRS site showing the position of the methyl ester moiety (dashed circle). (c) Effect of MRS7292 on [³H]dopamine uptake in ∆-hDAT mutants. Data was plotted and fitted using a non-linear regression model as described in ‘Methods’. Data from n = 3 biological replicates, each performed in technical triplicate, are represented as mean values +/− standard deviation. (d) Total, specific [³H]dopamine uptake for ∆-hDAT and ∆-hDAT mutant controls from the IC₅₀ experiments. Data from n = 3 biological replicates, each performed in technical triplicate, are represented as mean values +/− standard deviation. (e) Accommodation of the thienyl moiety (dashed circle) into the MRS subsite. (f) Alignment of the MRS7292 (shown as teal sticks) binding pocket with the binding pocket of hSERT in complex with serotonin (7LIA) and (g) dDAT in complex with β-CFT (4XPG). Superposition of structures was performed using α-carbon atoms of TM3 and TM8 of ∆-hDAT. The superposition shows how TMs 1b and 6a are displaced and reoriented whereas TMs 1a and 6b superimpose relatively well. Source Data

Extended Data Fig. 8. Allosteric site in hSERT and comparison to ∆-hDAT.

(a) Structural superposition of the allosteric binding pocket for serotonin (5-HT) (PDB: 7LIA) with the equivalent region in ∆-hDAT. The α-carbon to α-carbon distance (yellow dashed line) is indicated in Å. (b) Superposition of the S-citalopram (CIT) (PDB: 5I73) binding pocket with the equivalent region in ∆-hDAT. Superpositions were done using α-carbon atoms of TM3 and TM8. Selected residues and the allosteric molecules are represented as sticks.

Extended Data Fig. 9. Zinc site analysis and alignments with other NSSs.

(a) Isomesh map representation of coulomb density associated with the Zn²⁺ site in the ∆-hDAT cryo-EM reconstruction, contoured at 8 σ within 2.0 Å of the atoms associated with the structural feature. (b) Alignment of the amino acid sequence encompassing the Zn²⁺ binding region of ∆-hDAT with the equivalent sequences of related transporters. Residues that directly coordinate Zn²⁺ in ∆-hDAT are outlined in red. (c)-(e) Alignment of the zinc binding site of ∆-hDAT with the equivalent regions in hSERT (PDB:7LIA), the human glycine transporter (GlyT1; PDB:6ZBV), and the human GABA transporter (GAT1; PDB:7SK2) demonstrates the unique position of EL2 in Zn²⁺-bound ∆-hDAT compared to related transporters. All structural alignments used α-carbon atoms of TM3 and TM8 of ∆-hDAT. (f,g) Simple models of T211E and T211H were created by substituting the mutated residue for T211 and selecting rotamers with favorable chi1 and chi2 angles to estimate distances to the zinc ion (Å). (h) Inhibition curve for T211E at pH 7.5. Data was analyzed using a nonlinear regression model as described in ‘Methods’. Data from n = 3 biological replicates, each performed in technical triplicate, are represented as mean values +/− standard deviation. Source Data