SLC6 Neurotransmitter Transporters: Structure, Function, and Regulation

1. Molecular Organization.

The molecular cloning of the cDNA encoding the NTTs in the early 1990s revealed a highly similar primary structure of these transporters with an average amino acid sequence identity of 40%, suggesting that NTTs share a similar general structure (Guastella et al., 1990; Blakely et al., 1991; Hoffman et al., 1991; Pacholczyk et al., 1991; Liu et al., 1992a, 1993). Most of the NTTs contain approximately 600 amino acids, ranging from 599 (GAT1) to 632 (SERT), except that the two glycine transporters each contain approximately 700 amino acids. Sequence analysis combined with biochemical approaches suggested a membrane topology with 12 transmembrane (TM) helices with intracellular N- and C-termini and a large glycosylated loop between transmembrane helices III and IV (see section V) (Table 2). Except for NET, GLYT1, and GLYT2, only single isoforms have been identified for each NTT (Fig. 3). This membrane topology has been experimentally verified across the monoamine (Torres et al., 2003a), GABA (Kanner, 2006) and glycine transporter subfamilies by a large number of studies, mainly using specific antibodies (Melikian et al., 1994; Brüss et al., 1995; Qian et al., 1995; Nirenberg et al., 1996; Chen et al., 1997a) or site-selective biochemical labeling of Lys and/or Cys residues (Guastella et al., 1990; Pacholczyk et al., 1991; Giros and Caron, 1993; Nirenberg et al., 1996; Povlock and Amara, 1996; Chen et al., 1998; Ferrer and Javitch 1998). Another structural feature of all NTTs is a pair of Cys residues in extracellular loop 2 that form an intraloop disulfide bridge. The loop also contains sites for N-linked glycosylation (Patel et al., 1994; Olivares et al., 1995; Qian et al., 1995; Nguyen and Amara, 1996; Chen et al., 1997a, 1998; Ramamoorthy et al., 1998b; Cai et al., 2005) (section V) (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Membrane topology of SLC6 NTTs and splice variants. The NTTs contain 12 TM-spanning regions connected by intra- and extracellular loops with N and C termini located intracellularly. The extracellular loop between TM3 and TM4 contains sites for N-linked glycosylation. The genes for NET, GLYT1, and GLYT2 give rise to multiple splice variants. Five GLYT1 splice variants with variations in the N-terminal (GLYT1a, -1b, and -1c) and C-terminal (GLYT1d and -1e) and two splice variants of GLYT2 (GLYT2a and -2b) have been reported (Supplisson and Roux, 2002). For NET, two splice variants (NET1 and NET2) have been described with variations in the C-terminal region (Kitayama and Dohi, 2003).

2. Analyses of Functional Domains.

In the decade after the cloning of the NTTs, substantial efforts were made to identify and characterize the regions and specific residues that contribute to core transporter functions such as substrate binding and translocation, PTMs (section IV), or interactions with intracellular proteins. These efforts also focused on probing for secondary structure elements as well as to establish proximity relationships between TM domains and specific residues. Experimental approaches mainly relied on site-directed mutagenesis in combination with recombinant expression of NTTs in heterologous expression systems. The substituted cysteine accessibility method (SCAM) (Akabas et al., 1992) in particular has been widely used, especially for DAT, SERT, and the GABA transporter GAT1, to probe for residue accessibility and secondary structure elements in TM and intra- and extracellular loop regions, as well as to define the functional role of individual residues for binding of substrate and inhibitors (section III). In SCAM studies, native residues are systematically replaced by cysteines and selectively derivatized with thiol-modifying agents (Akabas et al., 1992; Javitch, 1998; Karlin and Akabas, 1998). If the cysteine side chain is modified by agents only reactive in a hydrophilic environment, it implies that the residue is exposed at the water-accessible surface of the protein. Likewise, hydrophobic agents can be used to probe membrane-embedded or intracellular residues when the transporter resides in intact cells. SCAM analysis of entire TM regions and loop regions of SERT provided early evidence for α-helical structure of putative TM spanning domains and identified residues important for binding of substrates and inhibitors (Chen et al., 1997b; Henry et al., 2003; Keller et al., 2004b; Mitchell et al., 2004; Sato et al., 2004; Zhang and Rudnick, 2005, 2006).

DAT has also been subject to SCAM analysis, leading to insights into substrate- and cocaine-induced conformational changes in the transporter, oligomerization domains, and residues involved in substrate binding and/or binding of inhibitors such as cocaine (Ferrer and Javitch, 1998; Chen et al., 2000, 2004a; Loland et al., 2002, 2003, 2004; Hastrup et al., 2003; Uhl and Lin, 2003; Sen et al., 2005).

SCAM analysis of the GABA and glycine transporters has been carried out to a lesser extent, although several key residues for transporter function of GAT1 have been identified by mutagenesis studies (Kanner et al., 1994; Keshet et al., 1995; Mager et al., 1996; Bismuth et al., 1997; MacAulay et al., 2001; Zomot and Kanner, 2003; Zhou et al., 2004, 2006; Zomot et al., 2005; Rosenberg and Kanner, 2008; Ben-Yona and Kanner, 2009). For analysis of proximity relationships between specific residues, the primary approach has been to take advantage of the structural constraints required for construction of binding sites for divalent cations in proteins. Engineering of His and/or Cys residues to generate artificial binding sites for Zn²⁺ or Cd²⁺ in monoamine and GABA transporters has been used to establish distances between specific residues (Loland et al., 1999, 2003; Norregaard et al., 2000, 2003; MacAulay et al., 2001; Goldberg et al., 2003; Torres et al., 2003b; Volz and Schenk, 2005; Zomot et al., 2005; Henry et al., 2006b; Kanner, 2006; White et al., 2006; Kanner and Zomot, 2008). Detailed accounts on the elucidation of NTT topology and mutational analysis are provided in several reviews (Povlock and Amara, 1996; Chen and Reith, 2000, 2002; Lin and Uhl, 2002; Torres et al., 2003b; Uhl and Lin, 2003; Rudnick, 2006; Kanner and Zomot, 2008).

1. Structure of the Leucine Transporter.

The X-ray crystal structure of LeuT showed that the protein adopted a unique fold compared with the few other secondary active transporters for which structures were available, such as the glycerol-3-phosphate/phosphate antiporter GlpT (Huang et al., 2003) and the proton-coupled lactose symporter LacY (Abramson et al., 2003). The structure confirmed many predictions made for SLC6 transporter architecture by revealing 12 TM regions connected by short intra- and extracellular loops with intracellular N and C termini (Fig. 4). The predicted secondary structure of the TM regions was confirmed, because all 12 TMs are almost exclusively α-helical, except short unwound segments in TM1 and TM6. Furthermore, several α-helical elements are also present in the intra- and extracellular loops (Fig. 4). The TMs pack as an intertwined helical bundle into a cylindrical shape that can be divided into an interior and outer part (Fig. 4). The inner ring is formed almost exclusively by TM1, TM3, TM6, and TM8 and holds a central substrate binding site (S1) in which the substrate (Leu) and two Na⁺ ions are accommodated (Figs. 4 and 5). In all substrate-bound LeuT structures, this central substrate binding pocket is occluded from both the external and internal medium (Yamashita et al., 2005; Singh et al., 2007, 2008; Zhou et al., 2007, 2009; Quick et al., 2009). Access to the S1 binding pocket from the extracellular medium is blocked by an interaction network formed between side chains from TM1, TM3, TM6, and TM10 (Fig. 5). The aromatic side chains of Tyr108 on TM3 and Phe253 on TM6 form a hydrophobic lid across the top of the pocket (Figs. 5B and 8C). Just above this external “lid,” the guanidium group of Arg30 on TM1 interacts via a pair of water molecules with the carboxylate group of Asp404 on TM10 (Fig. 5). In later structures of LeuT in complex with the noncompetitive inhibitor Trp, a direct interaction between the α-substituents of the inhibitor and the side chains of Arg30 and Asp404 is observed, an observation that is potentially significant for the mechanism of noncompetitive inhibition in LeuT (Singh et al., 2007; Zhou et al., 2007, 2009) (section III). Access from the substrate binding pocket toward the intracellular medium is blocked by a much larger protein layer containing a network of interactions formed mainly by tight packing of the intracellular halves of TM1, TM6 and TM8 (Yamashita et al., 2005; Singh et al., 2007, 2008; Zhou et al., 2007, 2009), resulting in a layer of ∼20-Å ordered protein structure being situated directly beneath the substrate binding pocket. This structure seems to be stabilized mainly by an interaction network formed between Arg5, Asp369, Ser267, and Tyr268 (Fig. 8D). Adjacent to these residues is Trp8, which has also been suggested to stabilize the substructure. Together, these regions prevent access to the central pocket and are proposed to constitute the gating regions in LeuT; e.g., they act as external and internal gates toward the extra- or intracellular medium by undergoing structural rearrangement as the protein shuttle through a series of conformational states during the substrate translocation mechanism (Yamashita et al., 2005; Kanner, 2008; Singh, 2008) (section III). This crucial role is supported by the observation that all residues participating in the two networks are strictly conserved across the SLC6 NTTs except for Asp404, which is substituted by a glutamate residue in SERT (Beuming et al., 2006), indicating important roles for both networks for the function of LeuT and the SLC6 NTTs. A notable feature of the “occluded” substrate-bound structures is the presence of an extracellular pathway, which leads from the occluded binding pocket to the extracellular surface of LeuT (Fig. 5), whereas a similar pathway leading from the substrate binding site toward the intracellular medium is less obvious.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

X-ray crystal structure and topology of LeuT, a bacterial homolog of SLC6 NTTs. A, structure of LeuT in the substrate-bound, outward-facing occluded conformation (Protein Data Bank ID 2A65). The substrate binding site is located at the core of the transporter, and the substrate and two Na⁺ ions are shown as van der Waals spheres in yellow and purple, respectively. B, topology of LeuT with the inner ring (TM1, TM3, TM6, and TM8) that forms the substrate binding site shown in blue. The red triangles highlight the 5 + 5 inverted repeats, formed by TMs 1 to 5 and TMs 6 to 10, that are related by an apparent two-fold symmetry. C, structure of LeuT with the TM domains that form the inner ring highlighted in blue seen from the side (left; TM10 is omitted for clarity) and the top (right; EL1 and EL4 are omitted for clarity).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Cross-sectional illustration of LeuT in an outward-facing occluded conformation. A, leucine is located in the substrate binding site in the core of the transporter and shown as van der Waals spheres in yellow. The extracellular and intracellular gating regions are highlighted with dashed lines, and the water-accessible pathway from the extracellular space to the substrate binding site is denoted with an orange arrow. B, close-up view of the S1 and S2 regions. The inner ring TM domains (TM1, TM3, TM6, and TM8), in addition to TM10, which lines the extracellular vestibule, are shown. The extracellular gate, formed by a water-mediated salt bridge between Arg30 and Asp404 in addition to the two aromatic residues Tyr108 and Phe253, separate the substrate binding site from the extracellular vestibule. Figures constructed on basis of the crystal structure of LeuT in complex with leucine (Protein Data Bank ID 2A65).

Although the overall LeuT structure is asymmetric, two similar structural motifs are found to be arranged by a pseudosymmetric inverted-repeat architecture: a five-helix bundle, containing TM1 to TM5, forms a V-shaped structural motif that is related by an apparent 2-fold rotational symmetry around an axis through the center of the membrane plane to a similar motif formed by TM6 to TM10 (Yamashita et al., 2005) (Fig. 4). The fold of this structural motif, denoted the “5 + 5 inverted repeat” fold, has subsequently been identified in several structures of secondary active transporters from prokaryotes, belonging to transporter families with no sequence relationship to LeuT or other SLC6 transporters and thus were not expected to be structurally or mechanistically related (Forrest and Rudnick, 2009; for review, see Abramson and Wright, 2009). These structurally related transporters include three Na⁺-coupled symporters for galactose (vSGLT; Faham et al., 2008), benzylhydantoin (Mhp1; Weyand et al., 2008), and betaine (BetP, Ressl et al., 2009), the arginine-agmatine antiporter AdiC (Gao et al., 2009, 2010) as well as the proton-coupled amino acid symporter ApcT (Shaffer et al., 2009) and the Na⁺-independent carnitine-butyrobetaine antiporter CaiT (Schulze et al., 2010). Consequently, the 5 + 5 inverted repeat fold first observed in LeuT is emerging as the structural hallmark of a large family of secondary active transporters that potentially operate via a conserved structural mechanisms (Lolkema and Slotboom, 2008; Forrest and Rudnick, 2009; Krishnamurthy et al., 2009; for review, see Abramson and Wright, 2009) (section II.C).

2. Binding Pockets in the Leucine Transporter and Solute Carrier 6 Neurotransmitter Transporters.

If sufficient sequence conservation is present, structures of bacterial proteins are often excellent templates for construction of three-dimensional structural models of homologous mammalian proteins. LeuT and the SLC6 NTTs share a remarkably high sequence similarity (55–67%) in the regions that are thought to constitute the core transport machinery, especially within the four-helix bundle, which house the central substrate binding site (S1) as well as the Na⁺ binding sites (Yamashita et al., 2005; Beuming et al., 2006) (Fig. 6). Therefore, the LeuT structures were predicted to provide a solid basis for construction of homology models for NTTs (Yamashita et al., 2005; Beuming et al., 2006; Henry et al., 2006b; Rudnick, 2006), which has not previously been possible because of the lack of appropriate template structures (Ravna and Edvardsen, 2001; Ravna et al., 2003; Pratuangdejkul et al., 2008). Accordingly, LeuT-based homology modeling is emerging as a valuable tool in studies of the NTT members, both in purely computational studies (Ravna et al., 2006; Huang and Zhan, 2007; Jørgensen et al., 2007a,b; Indarte et al., 2008; Xhaard et al., 2008; Kardos et al., 2010; Wein and Wanner, 2010) and as a complementary tool in functional studies (Dodd and Christie, 2007; Forrest et al., 2007, 2008; Paczkowski et al., 2007; Vandenberg et al., 2007; Zomot et al., 2007; Beuming et al., 2008; Celik et al., 2008b; Kniazeff et al., 2008; Andersen et al., 2009b, 2010; Kaufmann et al., 2009; Tavoulari et al., 2009; Field et al., 2010; Koldsø et al., 2010; Sinning et al., 2010) (section III). Modeling of DAT and SERT have so far received the most attention, which probably reflects the important role of these transporters as drug targets and results in generation of several three-dimensional models of human DAT (Beuming et al., 2006, 2008; Ravna, 2006; Indarte et al., 2008) and human SERT (Ravna et al., 2006; Forrest et al., 2007; Jørgensen et al., 2007a,b; Celik et al., 2008b; Forrest et al., 2008). However, models of NET (Paczkowski et al., 2007; Xhaard et al., 2008), GABA (Meinild et al., 2009; Kardos et al., 2010; Skovstrup et al., 2010; Wein and Wanner, 2010), and glycine transporters (Beuming et al., 2006; Edington et al., 2009) have also been generated. In general, the high degree of sequence similarity observed between the core region of LeuT and the SLC6 NTTs produce models with very similar overall structure of the inner and outer ring regions. In contrast, the intra- and extracellular loop regions are deviating more between the SLC6 NTTs and LeuT and are therefore more difficult to align with the LeuT template and confined with higher inaccuracy in the models (Beuming et al., 2006). Furthermore, the N- and C-termini of LeuT are much shorter and have no sequence similarity to the corresponding regions in the SLC6 NTTs counterparts (section V) (Fig. 10), thus these cannot be modeled.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Substrate binding pocket in LeuT and SLC6 NTTs. A, cross-sectional view of the substrate binding site in LeuT (Protein Data Bank ID 2A65), DAT (Beuming et al., 2008), GAT1 (Skovstrup et al., 2010), and GLYT1 (Beuming et al., 2006). TM domains that form the substrate binding site (TM1, TM3, TM6, and TM8) are shown as blue helices, Na⁺ ions as purple spheres, residue side chains as gray stick representations, and the substrates in yellow ball-and-stick representation. SLC6 residues conserved in LeuT are indicated in green. B, alignment of amino acid sequences of the inner ring helices (TM1, TM3, TM6, and TM8) of LeuT and human SLC6 NTTs: SERT (P31645), DAT (Q01959), NET (P23975), GAT1 (P30531), BGT1 (P48065), GAT2 (Q9NSD5), GAT3 (P48066), GLYT1 (GLYT1c isoform; P48067-1), and GLYT2 (Q9Y345) (UniProt-SwissProt human accession numbers are given in parentheses). Stars indicate the positions involved in substrate binding in LeuT (Yamashita et al., 2005); green and red indicate whether LeuT residue is conserved or nonconserved, respectively, in human NTTs.

3. The S1 Binding Site in the Central Binding Pocket.

In all substrate-bound LeuT structures reported so far, a single substrate molecule is accommodated in a common binding site, designated S1, that is located in the central cavity of the inner TM ring (Fig. 4). The conformations adapted by LeuT in these substrate-bound structures are highly similar and thought to represent the transporter in an “outward-facing occluded” conformational state occurring early in the translocation process. Here, the substrate and ions have triggered a conformational change that has closed access to the pocket from the external medium (section II.C). The S1 pocket surface is lined by polar, aromatic, and aliphatic amino acid side chains all four TMs (TM1, TM3, TM6, and TM8) surrounding the binding site, in addition to backbone amide groups from the unwound regions of TM1 and TM6 (Yamashita et al., 2005; Singh et al., 2008) (Fig. 6). The S1 pocket can be divided into two regions: a polar region formed exclusively by the unwound regions of TM1 and TM6 that accommodates the α-amino and α-carboxylate groups of the amino acid substrates, and a hydrophobic pocket formed by aliphatic side chains from TM1, TM3, and TM6 accommodating the hydrophobic substrate side chains (Yamashita et al., 2005; Singh et al., 2008) (Fig. 6). Within the polar region, the α-substituents of the substrate form hydrogen bonds with exposed backbone amide groups of TM1 and TM6. In addition to providing hydrogen-bonding partners for the substrate within the backbone, the helical breaks have been suggested to allow the substrate to interact directly with the ends of the helical segments, thus maximizing α-helical dipole moments (Yamashita et al., 2005). The α-amino group of the substrates interacts with backbone carbonyls from Ala22 in TM1, Phe253 and Thr254 in TM6, and with a side-chain hydroxyl from Ser256 in TM6 (Fig. 6). The α-carboxyl group of the substrate interacts with one of the two Na⁺ and backbone amide nitrogen from Leu25 and Gly26 in TM1, and a phenolic hydroxyl moiety from Tyr108 in TM3 (Yamashita et al., 2005). The hydrophobic region accommodating the aliphatic substrate side chain is formed by the side chains of Val104 and Tyr108 in TM3, Phe253, Ser256 and Phe259 in TM6, and Ser355 and Ile359 in TM8 (Fig. 6). As will be discussed, this region is a major determinant for substrate specificity of the SLC6 NTT binding pockets. It is noteworthy that Tyr108 and Phe253 are part of the predicted extracellular gate that separates the substrate from the extracellular vestibule as their aromatic side chain moieties align to form a barrier across at the top of the pocket (Figs. 5, 6, and 8C) (Singh, 2008).

4. The S2 Binding Site in the Extracellular Vestibule.

It has been suggested that the extracellular pathway (often referred to as the extracellular vestibule) that forms the solvent-accessible path from the extracellular medium toward the S1 site harbors a second substrate binding site in LeuT, designated the S2 site (Shi et al., 2008). The S2 site is located at the bottom of the extracellular vestibule, separated from the S1 site by the extracellular gate (Figs. 5 and 9). Substrate binding to the S2 site was first suggested by Shi et al. (2008), who used molecular dynamics (MD) simulations in combination with biochemical experiments on reconstituted LeuT to propose that occupation of the S2 site is required to trigger conformational changes that releases substrate from the S1 to intracellular side. Although the S2 site has not been found to occupy substrates in any structure of LeuT crystallized in presence of substrates, other structures of LeuT crystallized in presence of certain LeuT inhibitors or detergents have found that these bind in the S2 site (Zhou et al., 2007, 2009; Singh et al., 2008; Quick et al., 2009), hereby preventing conformational changes necessary for substrate translocation (Singh et al., 2007; Quick et al., 2009; Zhou et al., 2009), a mechanism that might be relevant for the mechanism of action of SLC6 NTT inhibitors (Zhou et al., 2007, 2009) (section III.B).

5. Structural Basis for Substrate Recognition in Mammalian Neurotransmitter Transporters.

Although the overall amino acid identity between mammalian NTT transporters and LeuT is 20 to 25%, the TM regions forming the S1 binding site are much more conserved with 55 to 67% sequence similarity (Fig. 6). Moreover, 7 of the 11 residues in LeuT that have direct interactions with the substrate are conserved across all nine mammalian NTTs (Fig. 6). This strongly indicates that the S1 site is structurally conserved between LeuT and SLC6 NTTs, consistent with a large body of experimental evidence that have identified S1 residues to control substrate affinity in the SLC6 NTTs (Beuming et al., 2006; Rudnick, 2006; Henry et al., 2007; Kanner and Zomot, 2008).

6. Structure of the S1 Pocket in Neurotransmitter Transporters and Implications for Substrate Selectivity.

For the monoamine transporters, docking 5-HT and dopamine into models of the S1 site in SERT and DAT, respectively, have obtained strikingly similar poses of the substrates (Huang and Zhan, 2007; Beuming et al., 2008; Celik et al., 2008b; Indarte et al., 2008). The aromatic moieties of the substrates are accommodated in a hydrophobic region of the pocket, formed by hydrophobic and aliphatic residues in TM1, TM3, and TM6 (Fig. 6), whereas the alkylamine side chains of 5-HT and dopamine both occupy a region equivalent to the polar region in LeuT that accommodates the substrate α-carboxyl group (Fig. 6). An important feature in this region of SERT, DAT, and NET is the presence of an Asp residue close to the TM1 helical break at a position where all other NTTs contain a Gly residue (Fig. 6). The acidic side chain in this position is strictly required for the function of the monoamine NTTs (Kitayama et al., 1992; Barker et al., 1999; Celik et al., 2008b; Andersen et al., 2009b), leading to the early suggestion that this residue is involved in a critical interaction with the monoamine substrate, possibly by a direct interaction with the amino group of the substrate (Kitayama et al., 1992; Barker et al., 1999). This idea is supported by recent models of monoamine substrates docked into LeuT-based homology models of DAT (Beuming et al., 2008) and SERT (Celik et al., 2008b; Kaufmann et al., 2009) as well as mutational studies of SERT showing that the loss of affinity observed when shortening the alkyl-amine side chain of 5-HT derivatives by one methylene can be compensated for by extending the amino acid side chain by one methylene via an Asp-to-Glu mutation in this position (Barker et al., 1999; Celik et al., 2008b). Moreover, as the monoamine substrates lack a negatively charged carboxylate group, the unique presence of an Asp residue in this position in the monoamine transporters is believed to compensate for the inability of the monoamine substrates to coordinate with one of the Na⁺-ions (section II.B). Sequence analysis show that the polar S1 region in LeuT accommodating the α-amino and α-carboxyl groups of the amino acid substrates are highly conserved in both GLYT1 and GLYT2 transporters, indicating a conserved binding mode of this substrate moiety (Beuming et al., 2006). By contrast, the residues forming the hydrophobic pocket surrounding the substrate side chains in LeuT are substituted with residues of larger size or different shape in GLYT1 and GLYT2 (Fig. 6). Hereby, the volume of the S1 pocket is reduced, so that only amino acids with no or small side chains such as Gly and Ala can be accommodated in the glycine transporters. In contrast, the equivalent positions in the S1 site in SERT, DAT, and NET contain amino acids with smaller-sized side chains in accordance with accommodation of larger substrates (Figure 6) (Pratuangdejkul et al., 2005; Celik et al., 2008b).

For the GABA transporters, several recent studies have employed LeuT-based homology modeling and ligand docking to generate models of the potential binding mode of GABA and other ligands in the S1 site (Kardos et al., 2010; Skovstrup et al., 2010; Wein and Wanner, 2010). The models display very similar orientations of GABA in the S1 site and show that the GABA transporter displays the highest degree of similarity with LeuT in terms of the substrate binding mode. Accordingly, when GABA is docked into models of GAT1, GAT2, and GAT3, the substrate is predicted to adopt an extended conformation, where the α-carboxyl group of GABA is accommodated equivalent to the α-carboxyl groups of the amino acid substrates in LeuT. The amino propyl chain of GABA is accommodated similarly to the alkyl side chain of Leu in LeuT, with formation of direct hydrogen bonds between the γ-amino group of GABA and the side chains of Tyr60, Ser396, Thr400, and the backbone carbonyl groups of Tyr60 and Ser396 (human GAT1 numbering) (Kardos et al., 2010; Skovstrup et al., 2010; Wein and Wanner, 2010) (Fig. 6). It has been suggested that GABA can assume a “cyclic” conformation in the binding pocket where the γ-amino group of GABA form an intermolecular interaction with the α-carboxyl group (Kanner and Zomot, 2008). This was recently supported by MD simulations of GABA binding in GAT1, although the majority of GABA-GAT1 interactions were conserved after rearrangement of the GABA molecule into a “cyclic” binding conformation (Skovstrup et al., 2010; Wein and Wanner, 2010).

7. Ion Binding Sites.

A fundamental feature of all SLC6 transporters is the cotransport of Na⁺ ions with the substrate molecule. Because Na⁺ is transported along the electrochemical gradient, this provides the energy required for the thermodynamically unfavorable transport of substrates. In addition, binding and, in some cases, cotransport of Cl⁻, K⁺, and H⁺ ions is necessary for the function of most SLC6 transporters (Fig. 2; Table 2). Considering these similarities among the NTTs, there seem to be surprisingly large differences in terms of the transport stoichiometry between substrate and Na⁺ and Cl⁻. Even among the individual members of the monoamine, GABA, and glycine transporter subclasses, different ionic stoichiometries are observed (Fig. 2; Table 2). For example, the monoamine transporters DAT and NET are believed to translocate one substrate molecule with two Na⁺ and one Cl⁻, whereas SERT translocate 5-HT with one Na⁺ and one Cl⁻, whereas performing antiport of one K⁺ and maybe H⁺ (Keyes and Rudnick, 1982; Talvenheimo et al., 1983; Gu et al., 1994, 1996b; Quick, 2003). Similar differences are also observed among GABA transporters with variation in the number of Na⁺ and Cl⁻ translocated per substrate molecule (Keynan and Kanner, 1988; Kavanaugh et al., 1992), whereas GLYT1 and GLYT2 cotransport glycine with two and three Na⁺, respectively, as well as one Cl⁻ (López-Corcuera et al., 1998; Roux and Supplisson, 2000; Supplisson and Roux, 2002). Essentially, all knowledge on the structural basis for ion binding to the NTTs has been provided by studies starting from LeuT structures, because previous mutagenesis studies had identified only a few residues that seemed involved in ion dependence and selectivity (Keshet et al., 1995; Mager et al., 1996; Penado et al., 1998; Ponce et al., 2000; Kristensen et al., 2004).

a. Na⁺ binding sites.

In all LeuT structures, two distinct sites in the S1 pocket are unambiguously being occupied by a Na⁺ ion, designated Na1 and Na2 (Yamashita et al., 2005; Zhou et al., 2007, 2009; Singh et al., 2008). Both Na⁺ binding sites are believed to have a key role in stabilizing in particular the unwound structures of TM1 and TM6, in the presence of substrate (Figs. 6 and 7). Furthermore, the Na⁺ ion located in Na1 interacts directly with the substrate α-carboxyl group, thereby providing a potential structural basis for the coupled Na⁺ and substrate translocation. Figure 7 compares the LeuT residues that form the Na1 and Na2 sites with the corresponding residues in the mammalian NTTs. In the Na1 and Na2 sites, Na⁺ is coordinated by five and eight backbone carbonyls or side-chain oxygens, respectively (Fig. 7). The residues forming the Na1 site are highly conserved across all SLC6 NTT members, strongly indicating that an equivalent site is present in these transporters (Rudnick, 2006; Singh, 2008; Gouaux, 2009). The only major difference among these residues is found in the monoamine transporters, which contain an Asp in the position equivalent to Gly24 in LeuT. Homology models of SERT (Henry et al., 2006b; Ravna et al., 2006; Celik et al., 2008b; Forrest et al., 2008; Andersen et al., 2009b) and DAT (Huang and Zhan, 2007; Beuming et al., 2008; Jin et al., 2008b) suggest that the β-carboxyl group of Asp compensates for the inability of the monoamine substrates to interact with Na⁺ in the Na1 site because these lack a negatively charged carboxyl group, in contrast to GABA and glycine as mentioned previously (Figs. 5 and 7). This is supported by the observation that mutation of this Asp residue in all monoamine transporters has detrimental effects on transport function (Kitayama et al., 1992; Barker et al., 1999; Henry et al., 2003; Celik et al., 2008b; Andersen et al., 2010). Of the five residues that coordinate Na2 in LeuT, one is identical and three are highly similar to the equivalent residues in the majority of NTTs (Fig. 7). The single nonconserved residue is a Thr in LeuT and an Asp in the NTTs, except for GLYT1, where it is a Gly. Assuming that the Asp β-carboxyl group could coordinate Na⁺ similar to the β-hydroxy group in Thr present in LeuT, it is feasible that the Na2 site is also conserved in all SLC6 NTTs, even for those that only translocate one Na⁺ with the substrate (Rudnick, 2006; Singh, 2008). Although Na⁺ binding sites similar to Na1 and Na2 exist in the mammalian NTTs, the precise role of occupation of these sites by Na⁺ for substrate binding and translocation remains largely unknown, partly because of a lack of functional data addressing the role of each site in the individual NTT members. For example, it has been suggested that NTTs translocating one Na⁺ (SERT and NET) uses only the Na1 site, whereas the transporters that translocate at least two Na⁺ per substrate also use the Na2 site (Rudnick, 2006; Singh, 2008).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Ion-binding site in LeuT and SLC6 NTTs. Na⁺ and putative Cl⁻ binding sites in LeuT (Protein Data Bank ID 2A65), GAT1 (Skovstrup et al., 2010), and SERT (Celik et al., 2008b). Na⁺ and Cl⁻ ions are shown as purple and green spheres, respectively, and the substrate is shown as yellow stick-and-ball representation. Putative direct interactions between ions and specific residues are indicated. In contrast to SLC6 transporters, LeuT is Cl⁻-independent. The Cl⁻ binding site in SLC6 transporters has been suggested to be located at a site equivalent to negatively charged carboxylate group (highlighted with red mesh) of Glu290 in LeuT (Forrest et al., 2007; Zomot et al., 2007). Similar to leucine, GABA and glycine are believed to coordinate Na1 directly through the α-carboxylate group. In the monoamine transporters, an Asp residue located in close proximity to the Na1 has been suggested to compensate for the lacking carboxylate group in monoamine neurotransmitters.

1. Structures of Prokaryotic Transporters in Various Conformational States.

The classic alternating access model implies that the transporter protein shuttle through at least three principal conformational states during the transport cycle: 1) the outward-facing conformation, where the substrate binding pocket is accessible to the extracellular medium; 2) the occluded conformation, where access to the pocket is blocked from either side; and 3) inward-facing, where the pocket is open to the intracellular medium. As a close SLC6 homolog, LeuT remains the best representative structural template for studying putative SLC6 NTT conformations; however, the available LeuT structures represent only outward-facing or outward-facing occluded conformations, whereas no inward-facing conformation is available. Nonetheless, some of the non-SLC6 transporters sharing the 5 + 5 inverted repeat fold have been crystallized in both occluded and inward-facing conformations (Fig. 8), thus offering alternative templates for modeling of SLC6 NTTs in the occluded and inward-facing conformations. These include structures of the sodium-benzylhydantoin symporter Mhp1 in occluded, outward-facing, and inward-facing conformation (Weyand et al., 2008; Shimamura et al., 2010), the arginine-agmatine symporter AdiC in occluded and outward-facing conformation (Fang et al., 2009; Gao et al., 2009, 2010), and an occluded structure of the sodium-betaine symporter BetP (Ressl et al., 2009). Furthermore, inward-facing structures are available for the sodium-galactose symporter vSGLT in the occluded (Faham et al., 2008) as well as the open form (Watanabe et al., 2010), the H⁺-coupled amino acid symporter ApcT (Shaffer et al., 2009), and the carnitine-butyrobetaine antiporter CaiT (Schulze et al., 2010). It is noteworthy that virtually all of the key structural features of the transporter core formed by the 5 + 5 motif are conserved across these transporter structures (Abramson and Wright, 2009; Forrest and Rudnick, 2009; Krishnamurthy et al., 2009). Specifically, the architecture and the apparent structural role of the 10 TM helices in the 5 + 5 motif are found to be strictly maintained in these seven prokaryotic transporters, including location of substrate and ion binding pockets. This remarkable structural homology among seven different transporters from five unrelated transporter families has led to the general proposal that the core structure operates by a highly conserved mechanism, in which movements of the individual parts are determined by common molecular principles. It seems likely that this is applicable to all members of this emerging transporter superfamily, including the NTT members of the SCL6 family (Singh, 2008; Weyand et al., 2008; Forrest and Rudnick, 2009; Krishnamurthy et al., 2009).

2. Models of Alternating Access Transport.

Structural analysis of the prokaryotic transporter structures has led to construction of several models for the structural mechanism of alternating access transport by members of the 5 + 5 inverted repeat superfamily. The initial step involves binding of substrate and ions to an outward-open apo state in which the S1 site is accessible through an open extracellular permeation pathway (Fig. 8). Representative conformations for this state are provided by structures of LeuT bound with a competitive inhibitor (Singh et al., 2008) and of substrate-free Mhp1 (Weyand et al., 2008) and AdiC (Fang et al., 2009; Gao et al., 2009). Occupation of the S1 site triggers transition into the outward-occluded state where the substrate is trapped in S1 site, exemplified by structures of substrate-bound LeuT (Yamashita et al., 2005; Singh et al., 2008), Mhp1 (Weyand et al., 2008), and AdiC (Gao et al., 2010) (Fig. 8). Comparative analysis of the outward-open and outward-occluded structures show that the occlusion is achieved mainly by rearrangement of the extracellular parts of the TM regions that contribute to formation of the extracellular pathway [i.e., inner ring helices TM1, TM3, TM6, and TM8 as well as TM2 and TM10 from the outer ring (LeuT numbering)]. This is achieved either by formation of interaction networks between side chains across the extracellular pathway that act as an extracellular gate (as observed for LeuT, vSGLT, and AdiC) or by steric occlusion of the extracellular pathway (as observed for Mhp1 and BetP). In contrast, the subsequent transitions leading to the inward-facing states seem, not surprisingly, to require far more substantial conformational changes. Comparison of the inward- and outward-facing structures show that the extracellular pathway retained in the outward-facing occluded structures have collapsed into a compact layer, whereas a pathway that exposes the S1 site to intracellular side has formed between the intracellular halves of TM1, TM6, and TM8 of the inner ring (LeuT numbering) as observed for the inward-facing structures of ApcT, Mhp1, and vSGLT (Faham et al., 2008; Shaffer et al., 2009; Shimamura et al., 2010). After the transition into the inward-facing conformation, ions and the substrate are then released from the S1 site by diffusion through the intracellular pathway.

Several proposals have been made for the principal movements of individual elements in the transporter core during the outward-to-inward transition. Major discrepancies in these models include the type of motions of the inner ring TM1 and TM6 helices. In one model, the unwound segments in the middle of TM1 and TM6 act as flexible “hinges” around which the extra- and intracellular halves of TM1 and TM6 can move independently in sequential transitions to open and close the intra- and extracellular pathways (Yamashita et al., 2005; Singh et al., 2008; Krishnamurthy et al., 2009). In contrast, Forrest et al. (2008) and Forrest and Rudnick (2009) have proposed that the helical bundle formed by TM1 and TM6 together with TM2 and TM7 constitutes a rigid body that by a single “rocking” motion can account for the transition between the outward- and inward-facing conformations. On the basis of the structure of BetP, Ressl et al. (2009) have proposed a third type of motion for the inner ring helices during a set of sequential transitions into the inward-facing conformation via an intermediate fully occluded conformation (exemplified by the unique conformation observed for BetP), stating that TM1 and TM6 together with TM8 carry out a stepwise, iris-like rearrangement by performing concerted anticlockwise rotations.

3. Structural Correlates of Alternating Access Transport in Solute Carrier 6 Neurotransmitter Transporters.

The above-mentioned models constitute a quantum leap forward in our understanding of the alternating access transport mechanism. However, many functional aspects of the transporters will remain difficult to correlate with the proposed structural mechanisms, because experimental data describing the protein dynamics of the transporters are lacking. In this respect, it is encouraging that 1) an increasing number of studies on transporter dynamics are emerging with focus on the fundamental ideas and concepts regarding transporter structure-function relationships proposed by these models and 2) many of these studies have focused on the SLC6 NTTs and their bacterial homolog LeuT (Beuming et al., 2008; Kniazeff et al., 2008; Shi et al., 2008; Zhao et al., 2010a,b).

A burst of activity within the SLC6 NTT research field has followed the growing number of bacterial transporter structures, many studies focusing on testing predictions made from the models described above, thereby often uncovering new details regarding key aspects of SLC6 NTT structure-function relationships. MD simulations in combination with biochemical and biophysical experiments have been used to study conformational changes taking place during substrate and ion binding, and the subsequent translocation processes in both LeuT and SLC6 (Forrest et al., 2007, 2008; Jørgensen et al., 2007a; Beuming et al., 2008; Caplan et al., 2008; Celik et al., 2008a; Kniazeff et al., 2008; Noskov and Roux, 2008; Noskov, 2008; Shi et al., 2008; Tavoulari et al., 2009; Zhao et al., 2010a,b). However, the time scale of typical MD simulations (10–100 ns) is far from the turnover rates of membrane transporters (0.1–10 s), and MD simulations are unlikely to be able to simulate the structural rearrangements during a complete translocation cycle. Still, MD simulations have played a pivotal role in studying the proposed changes of the interaction networks that form the extra- and intracellular gates (Fig. 5 and 8). Celik et al. (2008) followed the formation of the proposed extracellular gate in LeuT during a 100-ns MD simulation of LeuT in the outward-facing conformation (Fig. 5B), which is proposed to be a key event during the transition from the outward-facing open to the outward-facing occluded conformation (Fig. 8). As mentioned previously, the residues in LeuT that form the extracellular gate are conserved across all SLC6 NTTs and have been shown to be critical for the function of several of the SLC6 NTTs (Pantanowitz et al., 1993; Bismuth et al., 1997; Cao et al., 1998; Barker et al., 1999; Chen and Rudnick, 2000; Ponce et al., 2000; Beuming et al., 2008). In addition, MD simulation was used by Kniazeff et al. (2008) to study the effect of disrupting the interaction network that is thought to stabilize the tight packing of the intracellular halves of the inner ring helices TM1, TM6, and TM8 when the transporters reside in outward-facing states. Disruption of this tight network of interactions is predicted to be a major determinant for the transition into the inward-facing state (Yamashita et al., 2005; Singh, 2008). In the structures of LeuT, this network includes interactions between Arg5, Ser267, Tyr268, and Asp369 (Fig. 8D), all of which are strictly conserved in the SLC6 NTTs. Specifically, Arg5 forms a salt bridge with Asp369 that is stabilized by a cation-π interaction between Tyr268 and Arg5 (Yamashita et al., 2005; Singh, 2008). The role of this network in determining the conformational equilibrium of the transporter was demonstrated in DAT by mutation of key residues in the network, which shifted the transporter into an inward-facing conformation (Loland et al., 2002; Kniazeff et al., 2008). The conclusion was supported by rescue in the mutants of transport by Zn²⁺, which stabilizes the outward-facing conformation of DAT (Norregaard et al., 1998; Loland et al., 1999 2004; Norgaard-Nielsen and Gether, 2006). Furthermore, the mutants displayed dramatically decreased affinity for cocaine, which preferentially binds the outward-facing conformation (section III) and an increased accessibility of conformationally sensitive cysteine inserted in TM3 (Kniazeff et al., 2008). MD simulations of homology models of the DAT mutants corroborated the experimental findings by showing increased flexibility in the intracellular gating region and, moreover, that disruption of the network propagated toward the extracellular region and affected the global dynamics involved in transition of the transporter between the outward- and inward-facing conformations (Kniazeff et al., 2008). All together, these results demonstrated principal conformational changes in DAT predicted from the recent alternating access models.

Several other studies have demonstrated that transport-associated conformational changes in SLC6 NTTs are consistent with predictions from alternating access models. These include the use of voltage-clamp fluorometry to show movement of TM1 during transport (Meinild et al., 2009) and SCAM analysis to confirm that TM8 lines the intracellular pathway of GAT1 (Ben-Yona and Kanner, 2009). Furthermore, SCAM analysis of TM5 in SERT shows that the intracellular region of this domain becomes exposed to the intracellular side when SERT is locked in an inward-facing conformation; consistent with predictions that this region lines the intracellular pathway that becomes accessible in the predicted inward-facing state (Zhang and Rudnick, 2006). It is noteworthy that Forrest et al. (2008) used SERT as a model protein for SCAM experiments to validate the principle motions of the “rocking bundle” model.

A milestone study achieving real-time monitoring of protein dynamics of specific structural elements within LeuT has been reported (Zhao et al., 2010b). Single-molecule fluorescence resonance energy transfer (FRET) (Roy et al., 2008) was used to measure time-resolved dynamic behavior of the intracellular gating region in single LeuT molecules during the application of substrates and inhibitors. The study showed molecular details of intracellular gating of LeuT that might be masked by ensemble averaging or suppressed under crystallographic conditions. The data were interpreted in the context of an allosteric model for LeuT transport developed by the same group (Shi et al., 2008). This model incorporates the role of the S2 site into the original model proposed for alternating access transport in LeuT (Yamashita et al., 2005; Singh, 2008) (Fig. 8) and proposes that binding of a second substrate molecule to the S2 site is required to trigger the transition from the outward-facing occluded state into the inward-facing state (Shi et al., 2008).

In summary, mechanistic models based on the pool of structures obtained for prokaryotic transporters have provided a new framework for establishing hypotheses for structural dynamics of the SLC6 NTTs. The results obtained so far have largely validated several of the proposed mechanistic principles for alternating access to be valid for the SLC6 NTTs. Given the short time period over which these advances have been made, it must be expected that this progress will accelerate in the future years and will probably continue to be fuelled by increasingly detailed transport models for transporters of 5 + 5 structural repeat fold. However, requirements for continued progress will probably include not only development of new types of functional experiments but also structures of higher resolution, further snapshots of alternative conformations, and structures of other members of the transporter family, preferably the first structure of a SLC6 NTT protein.

1. Ligands for the Monoamine Transporters.

The recognition, now decades ago, of monoamine transporters as important drug targets for treatment of mood disorders initiated extensive drug discovery efforts in the pharmaceutical industry, focusing on design and synthesis of compounds targeting SERT, NET, and DAT. These endeavors have provided a plethora of ligands for these transporters, including compounds with high affinity and selectivity for each of the three transporters (Table 3). The tricyclic antidepressants (TCAs), including imipramine and clomipramine, were developed in the 1950s as the first generation of monoamine transporter drugs (Moltzen and Bang-Andersen, 2006; Andersen et al., 2009a). However, the TCAs display activity across a variety of different receptors (Gillman, 2007), and new generations of monoamine transporter inhibitors with little or no affinity for other proteins have subsequently been developed. According to their selectivity profile, these inhibitors are classified as selective serotonin-reuptake inhibitors (SSRIs), such as fluoxetine and paroxetine; selective norepinephrine-reuptake inhibitors (NRIs), such as reboxetine and atomoxetine; and selective dopamine-reuptake inhibitors, such as bupropion (Table 3). Dual-acting inhibitors with affinity for two of the three monoamine transporters include the classes of serotonin- and norepinephrine-reuptake inhibitors (SNRIs), such as duloxetine and desvenlafaxine, and NET/DAT inhibitors, such as nomifensine. Finally, compounds inhibiting all three monoamine transporters have also been identified, exemplified by the novel triple reuptake inhibitor tesofensine (Astrup et al., 2008).

Reflecting the central role of monoamine transporters in cocaine abuse and addiction, the tropane class of competitive inhibitors, including cocaine, has been extensively studied as monoamine transporter inhibitors. Cocaine itself is a potent and nonselective inhibitor of SERT, NET, and DAT, but several closely related analogs have been developed with improved affinity and selectivity across the monoamine transporters (Carroll, 2003; Jin et al., 2008a). Important examples include 2β-carbomethoxy-3β-(4-fluorophenyl)tropane (WIN 35,428), which has selectivity for DAT over NET and SERT and (−)-2β-carbomethoxy-3β-(4-iodophenyl)tropane (RTI-55), which has selectivity for SERT and DAT over NET (Table 3). Substantial evidence suggests that DAT is the primary target for the rewarding properties of cocaine (Giros et al., 1996b; Chen et al., 2006). This has increased the focus on development of DAT inhibitors as potential medications for cocaine addiction, and analogs of benztropine (BZT), which generally have higher affinity and selectivity for DAT compared with cocaine, have received most attention (Newman and Kulkarni, 2002; Dutta et al., 2003; Loland et al., 2008). It is noteworthy that some of these compounds do not possess the rewarding properties associated with cocaine. Indeed, the BZT derivative N-(n-butyl)-(bis-fluorophenyl)methoxytropane (JHW 007) has been found to efficiently antagonize the behavioral effects of cocaine (Newman et al., 1995; Katz et al., 2004; Desai et al., 2005).

The use of monoamine transporters as biomarkers for imaging of monoaminergic circuits (Fowler et al., 1999; Laakso and Hietala, 2000; Huang et al., 2002; Meyer, 2007) has prompted development of several highly selective compounds for use as radiolabeled tracers in in vivo imaging techniques, such as positron emission tomography and single photon emission computed tomography (Houle et al., 2000; Wilson et al., 2000). These ligands include some of the most potent and selective compounds targeting the monoamine transporters, such as 2-(2-dimethylaminomethylphenylsulfanyl)-5-methyl-phenylamine (MADAM) and 3-amino-4-(2-dimethylaminomethyl-phenylsulfanyl)benzonitrile (DASB) for SERT (Houle et al., 2000; Tarkiainen et al., 2001; Meyer et al., 2004), 2β-carbomethoxy-3β-(4-chlorophenyl)-8-(2-fluoroethyl)nortropane (FECNT) for DAT (Goodman et al., 2000), and 2-[(2-methoxyphenoxy)phenylmethyl]morpholine (MeNER) and methylreboxetine for NET (Ding et al., 2003; Schou et al., 2003, 2004; Wilson et al., 2003) (Table 3).

Several synthetic substrates of pharmacological interest are available for all three monoamine transporters. Prominent examples are amphetamine and amphetamine derivates such as methamphetamine and methylphenidate, which are being used as therapeutics against attention deficit hyperactivity disorder (ADHD), in addition to the widely used drug of abuse, 3,4-methylenedioxymethamphetamine (also known as “ecstasy”) (Green et al., 2003; Sulzer et al., 2005; Rothman and Baumann, 2006; Rothman et al., 2006). 1-Methyl-4-phenylpyridinium, the neurotoxic metabolite of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, is a substrate for all three monoamine transporters and has been used as tracer in functional studies of SERT, NET, and DAT (Buck and Amara, 1994; Wall et al., 1995; Scholze et al., 2000; Sitte et al., 2000, 2001). MPP⁺ has been further developed into the fluorescent analog 4-(4-(dimethylamino)-styryl)-N-methylpyridinium (ASP+), which has enabled real-time kinetic analysis of transport mediated by SERT (Oz et al., 2010), NET (Schwartz et al., 2003, 2005), and DAT (Bolan et al., 2007; Zapata et al., 2007) in addition to high-throughput assays for the measurement of monoamine transport uptake activity (Fowler et al., 2006; Jørgensen et al., 2008). A fluorescent cocaine analog has been developed with binding properties that enabled specific labeling of DAT in living neurons for real-time monitoring of DAT trafficking by confocal microscopy imaging (Eriksen et al., 2009).

2. Ligands for the GABA Transporters.

Inhibition of GABA reuptake has been recognized as a therapeutic strategy for the treatment of epileptic disorders for many years (Krogsgaard-Larsen et al., 1987, 2000). Extensive medicinal chemistry efforts have been devoted to the synthesis of GAT inhibitors with emphasis on development of compounds that discriminate between the different GAT subtypes (Høg et al., 2006). However, truly selective compounds have not yet been achieved, and only GAT1-selective inhibitors, including the antiepileptic drug tiagabine, have been identified so far (Table 4). Hence, there is a considerable lack of selective pharmacological tool compounds targeting the three remaining GAT subtypes (BGT-1, GAT2, and GAT3), and the functional role and the therapeutic potential of these subtypes remains to be established.

Pioneering work in the 1970s led to the discovery of guvacine and nipecotic acid as substrates for GABA transporters (Johnston et al., 1975; Krogsgaard-Larsen and Johnston, 1975; Krogsgaard-Larsen et al., 1975). After the cloning of the four GAT subtypes, it was realized that guvacine and nipecotic acid have selectivity for GAT1 over GAT3 with little or no affinity for BGT-1 and GAT2 (Thomsen et al., 1997; Liu et al., 2003; Kragler et al., 2005; Kvist et al., 2009). Although guvacine and nipecotic acid do not readily cross the blood-brain barrier (Krogsgaard-Larsen et al., 1987), lipophilic derivatives of these substrates represent the most important biologically active GAT inhibitors developed. Important examples include tiagabine and 1-(4,4-diphenyl-3-butenyl)-3-piperidinecarboxylic acid (SKF89976A), which are derivatives of nipecotic acid, and 1-(2-(((diphenylmethylene)amino)oxy)ethyl)-1,2,5,6-tetrahydro-3-pyridinecarboxylic acid (NNC-711) and 1-(2-(bis(4-(trifluoromethyl)phenyl)methox)ethyl)-1,2,5,6-tetrahydro-3-pyridinecarboxylic acid (CI-966), which are derivatives of guvacine. Still, these compounds are some of the most potent and selective GAT1 inhibitors identified (Nielsen et al., 1991; Andersen et al., 1993; Borden et al., 1994; Dhar et al., 1996; White et al., 2002; Kragler et al., 2005; Kvist et al., 2009) (Table 4).

GAT inhibitors with activity on GAT2, GAT3, and/or BGT1 include (S)-1-(2-(tris(4-methoxyphenyl)methoxy)ethyl)-3-piperidinecarboxylic acid (SNAP-5114), which has selectivity for GAT2 and GAT3 over BGT-1 but little or no activity at GAT1 (Dhar et al., 1994; Borden, 1996; Kvist et al., 2009), and (R)-N-(4,4-bis(3-methyl-2-thienyl)-3-butenyl)-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo(d)isoxazol-3-ol (EF1502), which possesses dual-acting noncompetitive inhibitory activity at the BGT-1 and GAT1 (Clausen et al., 2005; White et al., 2005) (Table 4). Assessment of these compounds in animal models of partial and generalized epilepsy indicates that BGT-1 is a potential antiepileptic drug target (White et al., 2005; Madsen et al., 2009). However, selective BGT-1 inhibitors are needed to support this hypothesis, which clearly illustrates the unmet need for subtype-selective pharmacological tool compounds.

3. Ligands for the Glycine Transporters.

Several ligands are available for the two glycine transporters GLYT1 and GLYT2, including competitive inhibitors and synthetic substrates (Table 5). Their identification has been driven by their potential use in the treatment of psychotic diseases and, besides being candidates for clinical studies, the ligands have excellent properties as pharmacological tools (Lechner, 2006; for review, see Dohi et al., 2009). The prototypic GLYT1-selective substrate N-methyl glycine (sarcosine) has formed the basis for generations of highly potent and selective derivatives, including N-[3-(4′-fluorophenyl)-3-(4′-phenylphenoxy)propyl]sarcosine (NFPS), (R)-N[3-phenyl-3-(4′-(4-toluoyl)phenoxy)-propyl]sarcosine [(R)-NPTS], and (R)-(−)-N-[3-[(4-triflouromethyl)phenoxy]-3-phenyl-propyl]glycine (ORG 24598) (Bergeron et al., 1998; Atkinson et al., 2001; Aubrey and Vandenberg, 2001; Brown et al., 2001; Lowe et al., 2003; Mezler et al., 2008; Perry et al., 2008) (Table 5). For nonsarcosine GLYT1 inhibitors, several structurally diverse compounds have emerged from drug discovery programs, such as ((2-(4-benzo(1,3)dioxol-5-yl-2-tert-butylphenoxy)ethyl)methylamino)acetic acid (LY2365109) (Perry et al., 2008), 2-chloro-N-((S)-phenyl((2S)-piperidin-2-yl)methyl)-3-trifluoromethyl benzamide (SSR504734) (Depoortère et al., 2005), SSR130800 (Boulay et al., 2008), N-[3-(4-chlorophenyl)-3-[4-(2-thiazolylcarbonyl)phenoxy]propyl]-N-methyl-glycine (CP-802,079) (Martina et al., 2004), and (R)-4-[5-chloro-2-(4-methoxy-phenylsulfanyl)-phenyl]-2-methyl-piperazin-1-yl-acetic acid (Lu AA20465) (Smith et al., 2004), in addition to a series of cyclic tetrapeptides that was recently isolated from a soil bacteria (Terui et al., 2008). These compounds are excellent tools for studying the modulatory role of GLYT1 in neurotransmission [for example, in N-methyl-d-aspartate (NMDA) receptor-mediated signaling and plasticity at excitatory synapses (Bergeron et al., 1998; Kinney et al., 2003) as well as glycinergic signaling at inhibitory synapses (Bradaïa et al., 2004)].

The first GLYT2-selective inhibitor was the competitive inhibitor 4-benzyloxy-3,5-dimethoxy-N-[(1-dimethylaminocyclopentyl)methyl]benzamide (ORG 25543) (Caulfield et al., 2001), and other GLYT2-selective inhibitors belonging to different chemical classes have subsequently been developed (Isaac et al., 2001). GLYT1 and GLYT2 have several isoforms, thus increasing Gly transporter heterogeneity in the CNS (Aragón and López-Corcuera, 2003). However, because the isoforms exclusively involve changes in the intracellular N- and C-terminal domains, which have limited contribution to transporter function, it appears unlikely that differential targeting of individual isoforms by ligands will be possible; accordingly, all Gly transporter ligands examined across isoforms show identical pharmacological properties (Mezler et al., 2008).

1. Location of Inhibitor Binding Pockets.

So far, two binding sites have been identified in the crystal structures of inhibitor-bound LeuT: the centrally located substrate S1 binding pocket and the extracellular vestibule, which contains the putative S2 binding site (Singh et al., 2007, 2008; Zhou et al., 2007, 2009) (Fig. 9). The equivalent regions in the SLC6 NTTs are natural candidates for harboring ligand binding sites. Indeed, re-evaluation of extensive mutational analysis shows that the majority of residues that are critical for recognition of inhibitors are located in the TM and loop regions that contribute to the formation of the extracellular permeation pathway and the substrate binding pocket (Chen and Reith, 2000, 2002; Norregaard et al., 2000; Mortensen et al., 2001; Norregaard and Gether, 2001; Goldberg et al., 2003; Gu et al., 2006; Henry et al., 2006b; Neubauer et al., 2006; Andersen et al., 2009b, 2010; Koldsø et al., 2010; Sinning et al., 2010).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

Small-molecule inhibition of LeuT and SLC6 NTTs. A, cross-sectional view of crystal structures of LeuT in complex with competitive (left, Trp; Protein Data Bank ID 3F3A) and noncompetitive (right, imipramine; Protein Data Bank ID 2Q72) inhibitors. Trp binds in the S1 site and stabilizes LeuT in an outward-facing open conformation, whereas imipramine binds in the S2 site in the extracellular vestibule and stabilizes LeuT in an outward-facing occluded conformation. B, schematic illustration of putative transporter conformations stabilized by inhibitors. C, cross-sectional illustrations of ligand-docking models of competitive inhibitor binding in SLC6 transporters. TM domains forming the binding site (TM1, TM3, TM6, and TM8) are shown as blue helices, Na⁺ ions as purple spheres, residue side chains as gray stick representations, and the inhibitors in yellow ball-and-stick representation. Conserved residues are highlighted with green labels. Left, cocaine binding in DAT (Beuming et al., 2008). Middle, imipramine binding in SERT (Sinning et al., 2010). Right, tiagabine binding in GAT1 (Skovstrup et al., 2010).

2. Structural Basis for Competitive Inhibition.

The central substrate binding pocket is an obvious candidate site for binding of competitive inhibitors, because inhibitor binding sites overlapping with the S1 substrate site offer a straightforward structural mechanism for competitive inhibition and are in agreement with pharmacological data showing that competitive inhibitors can be displaced in a concentration-dependent manner by substrates (Talvenheimo et al., 1979; Humphreys et al., 1988; Marcusson and Tiger, 1988; Graham et al., 1989; Apparsundaram et al., 2008). Although mutational studies on the SLC6 NTTs have identified residues in virtually all TM domains to be important for recognition of competitive inhibitors, the majority of the most sensitive residues reside within the regions that form the extracellular permeation pathway and the S1 substrate binding pocket (Barker et al., 1994, 1998; Barker and Blakely, 1996; Mortensen et al., 1999; Adkins et al., 2001; Larsen et al., 2004; Henry et al., 2006b; Beuming et al., 2008; Severinsen et al., 2008; Walline et al., 2008; Andersen et al., 2009b, 2010; Field et al., 2010). For example, in the monoamine transporters, sensitivity toward a wide range of competitive inhibitors, including fluoxetine, paroxetine, citalopram, and imipramine, as well as psychostimulants such as cocaine and amphetamines, can be decreased by several orders of magnitude by mutation of S1 residues (Barker et al., 1999; Henry et al., 2006b; Beuming et al., 2008; Walline et al., 2008; Andersen et al., 2009b, 2010; Koldsø et al., 2010; Sinning et al., 2010).

LeuT structures have been used as templates for construction of monoamine transporter models in complex with several competitive inhibitors, including the prototypical antidepressants imipramine (Sinning et al., 2010), escitalopram (Andersen et al., 2010; Koldsø et al., 2010), and fluoxetine (Tavoulari et al., 2009) in SERT and cocaine in DAT (Beuming et al., 2008) (Fig. 9). These experimentally validated models have revealed a common structural model for competitive inhibition of monoamine NTTs: Inhibitors compete with substrate for binding in the S1 site and thereby trap the transporter in an outward-facing occluded or outward-facing open conformational state with local structural differences in the binding site. The inhibitors are accommodated within the inner ring helices TM1, TM3, TM6, and TM8, and the essential amino group found in virtually all SERT, NET, and DAT inhibitors is coordinated by an Asp (Asp79 in DAT and Asp98 in SERT) located in the helical break of TM1 (Fig. 9). In SERT, imipramine was found to stabilize a conformation with a slightly more open external gate region compared with fluoxetine and escitalopram, probably because of the more bulky nature of the inhibitor molecule (Tavoulari et al., 2009; Andersen et al., 2010; Koldsø et al., 2010; Sinning et al., 2010). Beuming et al. (2008) have constructed experimentally validated models of the binding sites for cocaine, amphetamines, and BZTs in DAT. It is noteworthy that direct support for a binding site overlapping with that of dopamine was obtained by trapping the radiolabeled cocaine analog [³H]WIN 35,428 in the transporter through cross-linking of engineered cysteines or by engineering of a Zn²⁺ binding site that was extracellular to the predicted common binding pocket within the vestibule corresponding to S2 (Beuming et al., 2008). In the same study, evidence was obtained that cocaine stabilizes a more outward-facing open conformation of DAT compared with amphetamine, MDMA, BZT, and BZT analogs, which were suggested to stabilize the transporter in an outward-facing occluded conformation (Beuming et al., 2008; Bisgaard et al., 2011). It is noteworthy that this has been linked to the reduced stimulant effect in vivo of the BZT analogs compared with cocaine (Reith et al., 2001; Katz et al., 2004; Loland et al., 2008). The precise orientation and binding mode of BZT and BZT analogs in S1 was demonstrated in a later study (Bisgaard et al., 2011).

Only a few LeuT-based homology models of GABA and glycine transporters have emerged, and the structural details underlying inhibition of GATs and GLYTs are therefore not as well characterized. However, the first model of inhibitor binding to GAT was recently constructed, thereby providing insight into the structural details of the tiagabine binding site in GAT1 (Skovstrup et al., 2010). The hydrophilic moiety of tiagabine was found to bind in the S1 substrate site, whereas the lipophilic “anchor” of tiagabine protrudes out of the extracellular lid and partially binds in the extracellular vestibule site and thereby stabilizes an outward-open conformation of GAT1 (Skovstrup et al., 2010) (Fig. 9). Hence, the tiagabine binding site overlaps with both the S1 and S2 sites, which might clarify the mixed-type mechanism of GABA uptake inhibition found in early pharmacological characterization of tiagabine (Braestrup et al., 1990). The binding site for the competitive GAT1 inhibitor SKF89976A was recently suggested to overlap the substrate binding site in GAT1, thereby stabilizing an outward-facing conformation of the transporter (Zhou et al., 2004; Dodd and Christie, 2007; Rosenberg and Kanner, 2008; Ben-Yona and Kanner, 2009), substantiating that the competitive inhibitor binding site in GATs overlaps the S1 substrate binding site as found for SLC6 monoamine transporters.

3. Structural Basis for Noncompetitive Inhibition.

LeuT has been cocrystallized with a set of noncompetitive inhibitors, including TCAs and SSRIs (Singh et al., 2007; Zhou et al., 2007, 2009). Although these compounds have very low affinity for LeuT, they are capable of inhibiting transport at high micromolar concentrations. Cocrystallization of LeuT with both TCAs and SSRIs revealed that these compounds bind in the extracellular vestibule (Fig. 9), suggesting that the equivalent vestibule in the SLC6 NTTs have a similar function for accommodation of noncompetitive inhibitors (Zhou et al., 2007, 2009). However, it should be noted that a large body of evidence indicates that the primary high-affinity binding site for TCAs and SSRIs, including those cocrystallized with LeuT, likely occurs in the S1 site in monoamine transporters (for review, see Henry et al., 2006a; Rudnick, 2007; Singh, 2008; Gouaux, 2009). Several studies have found that TCAs and SSRIs, in addition to binding to a high-affinity binding site, also bind to a low-affinity allosteric site in SERT (Sette et al., 1983; Wennogle and Meyerson, 1983, 1985; Plenge and Mellerup, 1985; Segonzac et al., 1985; O'Riordan et al., 1990; Plenge et al., 1991; Chen et al., 2005a,b). Binding to this allosteric site markedly increases the off-rate of the inhibitor bound to the high-affinity binding site. It has been suggested that the allosteric site is situated corresponding to a suggested oligomeric interface in the transporter (Neubauer et al., 2006); however, it is tempting to speculate that the allosteric site could be situated in the S2 vestibule outside the primary S1 binding pocket. Further experimental efforts are required to investigate this possibility.

Compared with the large number of competitive inhibitors available for SLC6 NTTs, very few inhibitors exhibit a noncompetitive mode of inhibition, in which inhibition is entirely independent of substrate concentrations. In a structural context, noncompetitive inhibitors must bind to a site whose occupation is mutually exclusive with substrate site occupation, which may be achieved by stabilizing a distinct conformational state of the transporter in which the substrate cannot bind. An example of such a mechanism is provided by the SERT inhibitor ibogaine, which has been suggested to trap SERT in an inward-facing conformation (Zhang and Rudnick, 2006; Jacobs et al., 2007). When ibogaine promotes the inward-facing conformation, the substrate site becomes mutually exclusive with the ibogaine binding site, thereby providing an explanation for the noncompetitive mechanism of action in a structural context (Jacobs et al., 2007). Another structural mechanism underlying noncompetitive mode of inhibition has been suggested for the χ-conotoxin MrIA. The 13-amino acid χ-conotoxin acts as a noncompetitive inhibitor of NET (Sharpe et al., 2001, 2003), presumably by binding in the external permeation pathway and stabilizing the transporter in an outward-facing conformation in which access of substrate to the S1 site is efficiently blocked (Paczkowski et al., 2007). Although several noncompetitive GLYT1 inhibitors have been reported, including NFPS and Org 24598 (Aubrey and Vandenberg, 2001; Mallorga et al., 2003; Mezler et al., 2008), the structural details underlying noncompetitive inhibition of GLYTs remains elusive. However, a recent study identified EL2 and EL4 in GLYT2 to be important for recognition of the noncompetitive GLYT2 inhibitor, N-arachidonylglycine (Edington et al., 2009). These extracellular loop regions contribute to formation of the vestibule region, including the S2 binding site, suggesting that N-arachidonylglycine binds in the S2 site of GLYT2. Residues in EL2 and EL4 have previously been shown to be critical for binding of Zn²⁺, which acts as a noncompetitive inhibitor of GLYT1 (Ju et al., 2004), suggesting that the S2 site in GLYTs forms a common noncompetitive binding site equivalent to the S2 site found in LeuT.

1. Glycosylation.

The SLC6 NTTs each possess two to four consensus sites for N-linked glycosylation (N-X-S/T) within EL2 that are subject to N-linked glycosylation (Table 6) (Tate and Blakely, 1994; Olivares et al., 1995; Bennett and Kanner, 1997; Torres et al., 2003a; Li et al., 2004; Rasmussen et al., 2009). Differences in the degree of glycosylation are observed among different tissue, brain regions, and cell types and during development (Lew et al., 1991; Patel et al., 1994; Qian et al., 1995; Nguyen and Amara, 1996). Furthermore, glycosylation patterns also differ between species (Patel et al., 1993). In general, removal of the glycosylation sites by mutagenesis or by enzymatic deglycosylation leads to markedly reduced uptake activity as a consequence of a reduced number of transporters at the cell surface. This has been attributed to instability of the nonglycosylated transporter (DAT and NET) (Melikian et al., 1996; Li et al., 2004), disruption of proper trafficking to the plasma membrane (GLYT1 and GLYT2) (Olivares et al., 1995; Martínez-Maza et al., 2001), or a combination of both (GAT1) (Cai et al., 2005). In general, removal of glycosylation sites did not affect ligand binding affinities or intrinsic transporter function (Tate and Blakely, 1994; Nguyen and Amara, 1996; Torres et al., 2003a), although reduced turnover rates have been suggested in some studies (Melikian et al., 1996; Li et al., 2004; Cai et al., 2005). For GAT1, a mechanism involving a reduced apparent affinity of the transporter for extracellular Na⁺ were proposed to contribute to the reduced transport activity observed in glycosylation deficient mutants (Cai et al., 2005).

2. Phosphorylation

1. Soluble N-Ethylmaleimide-Sensitive Factor Attachment Protein Receptors and Synaptic Vesicle Proteins.

One interaction that seems to be general for all SLC6 NTTs is an interaction with the SNARE protein syntaxin1A (Beckman et al., 1998; Geerlings et al., 2000, 2001; Haase et al., 2001; Sung et al., 2003; Lee et al., 2004). Syntaxin1A is a neuronal membrane protein belonging to the family of Gln-SNAREs involved in the vesicle docking process during exocytosis of synaptic vesicles (Sudhof, 2004). An additional role of syntaxin1A has emerged with the findings that syntaxin1A binds to and regulates the activity and cell surface availability of several neuronal membrane proteins. Compelling evidence suggests that the N termini of SLC6 NTTs interact directly with syntaxin1A, possibly through an interaction with the SNARE motif H3 (Deken et al., 2000; Quick, 2003; Lee et al., 2004). For DAT, the interaction site is contained within the first 33 residues of the N terminus (Binda et al., 2008), whereas residues 30 to 54 of the GAT1 N-terminal domain were shown to interact directly with syntaxin1A (Deken et al., 2000) (Fig. 10). Further characterization identified three Asp in the GAT1 N terminus (Asp40, Asp43, and Asp45) as critical for syntaxin1A binding. Likewise, mutation of negatively charged residues between positions 11 and 30 of SERT disrupted syntaxin1A binding (Quick, 2003). Knockdown of syntaxin1A by antisense oligonucleotide or inactivation by cleavage of syntaxin1A with botulinum toxin C1 (BoTx C1) have substantiated that syntaxin1A plays a role in regulating the level of SLC6 NTTs at the cell surface (Beckman et al., 1998; Deken et al., 2000; Geerlings et al., 2000, 2001; Quick, 2002; Sung et al., 2003; Cervinski et al., 2010). In addition, the interaction with syntaxin1A has been shown to reduce the catalytic activity of NET and GAT (Deken et al., 2000; Sung et al., 2003). For GAT1, the interaction with syntaxin1A was reported to be regulated by GABA as prolonged incubation with substrate increased the catalytic activity of GAT1 when in the presence of syntaxin1A. Furthermore, the syntaxin1A interaction with GAT1 was found to decrease after transport activity. This suggests that syntaxin1A is involved in a regulatory mechanism in which GABA transport capacity can be up-regulated after periods with high transport activity (Quick, 2002). Finally, the electrophysiological properties of the transporters have been found to be influenced by the interaction with syntaxin1A. For SERT, interaction with syntaxin1A abolished the transport-associated current that is a result of an electrogenic nonstoichiometric transport mode as well as the Na⁺ leak current observed in absence of substrate, resulting in a electroneutral transport mode in which substrate and ion transport are strictly coupled (Quick, 2003). Likewise, studies performed in Caenorhabditis elegans suggested that the interaction of DAT with the C. elegans syntaxin1A homolog changes the electrophysiological properties of the transporter (Carvelli et al., 2008).

By use of the mating-based split ubiquitin system, the synaptic vesicle protein synaptogyrin-3 has been identified as a potential interacting partner for DAT (Egaña et al., 2009). The synaptogyrin-3/DAT interaction was further confirmed by FRET experiments and GST-pulldowns and coimmunoprecipitations that further showed that the interaction was mediated between the N terminus of DAT and the cytoplasmic N terminus of synaptogyrin. In the catecholaminergic cell lines PC12 and MD9D, which both contain vesicular monoamine transporter-positive vesicles, overexpression of synaptogyrin-3 with DAT increased DAT uptake capacity, whereas siRNA-mediated knockdown of synaptogyrin-3 expression reduced DAT activity, effects that both were absent in non-neuronal HEK293 cells. Furthermore, DAT activity was reduced by 40% in the presence of the vesicular monoamine transporter 2 inhibitor reserpine in PC12 cells but not in HEK293 cells. Because the DAT N terminus was found to be capable of pulling down purified synaptic vesicles, the authors suggest the challenging hypothesis that the interaction between synaptogyrin-3 and DAT physically couples DAT to synaptic vesicles and that this coupling leads to an increase in DAT activity and presumably a more efficient filling of synaptic vesicles (Egaña et al., 2009). Synaptogyrin-3 neither bound NET nor affected NET function, suggesting that the interaction is DAT-specific. Another SNARE protein, soluble N-ethylmaleimide-sensitive factor attachment protein-25, and the synaptic vesicle membrane protein synaptophysin were found to colocalize with NET in axonal varicosities; however, a direct interaction was not confirmed (Schroeter et al., 2000).

2. Ca²⁺/Calmodulin-Dependent Protein Kinase.

There is compelling evidence that the DAT C terminus, but not the C termini of NET and SERT, binds the α subunit of the serine/threonine protein kinase CaMKII. This interaction was suggested to facilitate CaMKII-dependent phosphorylation of N-terminal Ser that in turn stimulates amphetamine-induced dopamine efflux (Fog et al., 2006). The in vivo role of this interaction was studied by in vivo chronoampermetry measurements showing that CaMKII activity was required for amphetamine-induced dopamine efflux in mouse striatum (Fog et al., 2006). It is noteworthy that amphetamine-induced dopamine efflux might also be regulated by the interaction of DAT with syntaxin1A (Binda et al., 2008). Amphetamine was found to increase the interaction between DAT and syntaxin1A in a CaMKII-dependent manner and, thus, a mechanism was proposed in which amphetamine activation of CaMKII strengthens the DAT/syntaxin1A interaction, resulting in a mode of DAT in which efflux is possible (Binda et al., 2008). It is noteworthy that CaMKII-dependent regulation of a syntaxin1A/transporter interaction has also been demonstrated for SERT (Ciccone et al., 2008) and NET (Dipace et al., 2007).

3. Postsynaptic Density 95/Discs-Large/Zona Occludens Domain Proteins.

Although the SLC6 NTTs differ in their extreme C terminus, they all contain putative PDZ (postsynaptic density 95/discs-large/zona occludens) binding sequences. PDZ domains represent one of the most widespread protein recognition domains in the human genome and are characterized by binding short linear consensus motifs mostly located at the C terminus of their interaction partners (Nourry et al., 2003; Kim and Sheng, 2004). Most PDZ domain-containing proteins play key roles in regulating targeting and trafficking of their interaction partners and are critical in protein scaffolding processes, where they organize proteins into supramolecular complexes and bring different cellular components in close proximity to ensure spatial and temporal control of cellular processes.

The synaptic PDZ domain-containing protein PICK1 was initially identified as a binding partner to the PDZ recognition motif in the extreme C termini of DAT and NET (Torres et al., 2001) (Fig. 10). This interaction was proposed to enhance DAT surface targeting and induce a clustering phenotype in transfected cells (Torres et al., 2001). However, the PICK1 interaction does not seem to play a role in ER export and surface targeting of the transporter. DAT C-terminal residues are important for appropriate membrane targeting of DAT, but mutations in DAT were identified that specifically disrupted PDZ domain interactions without affecting surface targeting, and mutations were identified that disrupted surface targeting without affecting PICK1 binding (Bjerggaard et al., 2004). Thus, the functional significance of the DAT/PICK1 interaction remains to be settled.

Additional PDZ domain-containing proteins have been identified as SLC6 NTT interaction partners mainly by use of yeast two-hybrid systems. Syntenin-1 was found to interact with the C terminus of GLYT2 (Ohno et al., 2004). Native syntenin-1 and GLYT2 colocalize in neuronal tissue and can be immunoprecipitated together (Ohno et al., 2004). Removal of the PDZ binding motif in GLYT2 result in reduced synaptic localization in neuronal cells (Armsen et al., 2007), suggesting that the syntenin-1/GLYT2 or another PDZ domain-containing protein interaction are involved in recruitment to and/or stabilization of GLYT2 at synaptic sites. It is noteworthy that syntenin-1 also binds to syntaxin1A, but GLYT2 and syntaxin1A cannot bind to syntenin-1 simultaneously, indicating that syntenin-1 does not participate directly in syntaxin1A-mediated trafficking of GLYT2 (Ohno et al., 2004).

Using a proteomics approach, neuronal nitric-oxide synthase (nNOS) was identified as a PDZ domain-containing protein interacting with the PDZ binding motif in the C terminus of SERT (Chanrion et al., 2007) (Fig. 10). Coexpression of SERT with nNOS decreased 5-HT uptake capacity in transfected cells by decreasing SERT cell surface expression, and this effect could be abolished by removing the PDZ binding motif in SERT. Furthermore, 5-HT transport in synaptosomes from nNOS knockout mice was enhanced compared with wild-type animals. Inhibition of the SERT/nNOS interaction in wild-type animals by injection of a SERT C-terminal peptide fused to the cell membrane penetrating HIV-1 Tat peptide also increased SERT acitivty in synaptosomes (Chanrion et al., 2007).

The PDZ domain of Pals1, a member of the membrane-associated guanylate kinase family, binds the C terminus of GAT1 (Fig. 10). Coexpression of GAT1 with Pals1 increased GABA uptake because of an increase in total GAT1 protein, which led to the proposal that Pals1 might slow the turnover time for GAT1 and thereby increase GAT1 cell-surface levels (McHugh et al., 2004). Another member of the membrane-associated guanylate kinase family, PSD-95, was found to interact with GLYT1 and stabilize the transporter at the cell surface (Cubelos et al., 2005b). Gly is a coagonist at the NMDA receptor, a well characterized interaction partner for PSD-95, and the GLYT1/PSD-95 interaction was proposed to bring GLYT1 and NMDA receptors together to enable tight regulation of extracellular Gly concentration around NMDA receptors. This hypothesis correlates with the observation that GLYT1, in addition to its presence in glial cells, is present at postsynaptic sites of glutamatergic neurons within complexes containing both GLYT1 and NMDA receptors (Cubelos et al., 2005b). Finally, BGT1 was found to bind the PDZ protein LIN7 (Massari et al., 2005). It is noteworthy that the interaction with LIN7 was regulated in a PKC-dependent manner; i.e., PKC activation promoted phosphorylation of Thr612 in the LIN7 association motif, causing decreased LIN7 binding to BGT1 concomitantly with transporter internalization (Massari et al., 2004).

4. Hydrogen Peroxide-Inducible Clone-5.

The LIM (Lin11, Isl-1, and Mec-3) domain-containing focal adhesion protein hydrogen peroxide-inducible clone-5 (Hic-5) is another scaffolding protein found to interact with the C terminus of DAT, NET, and SERT (Carneiro et al., 2002; Carneiro and Blakely, 2006). On the basis of results from differential extraction and sedimentation procedures with human and mouse platelets, it was suggested that the interaction of Hic-5 with SERT facilitates SERT internalization by promoting interaction with the actin cytoskeleton (Carneiro and Blakely, 2006). Furthermore, the Hic-5/SERT association was linked to the PKC-mediated regulatory pathway, because stimulation of PKC positively modulated the Hic-5/SERT interaction (Carneiro and Blakely, 2006). The protein MacMARCKS (homolog of myristoylated alanine-rich C kinase substrate) has also been proposed to be involved in PKC-mediated internalization of SERT via interaction with the SERT C terminus (Jess et al., 2002). MacMARKS is a PKC substrate and coexpression of MacMARKS and rat SERT in HEK293 cells altered the PKC-dependent regulation of SERT transport activity compared with cells expressing SERT alone (Jess et al., 2002).

5. Proteins in the Early Secretory Pathway.

As mentioned before, the GAT1 C terminus binds Sec24D, a component of the COPII coat complex that facilitates transport from ER to Golgi (Farhan et al., 2004) (section V.B). The binding of Sec24D, which depends on Arg566 and Leu567 in the GAT1 C terminus, was shown to facilitate concentrative ER export of GAT1 (Farhan et al., 2007). This motif is conserved in the SLC6 family and consistently the corresponding motif in DAT and SERT was shown to mediate Sec24D binding with these transporters (Farhan et al., 2007). The C terminus of GLYT1 interacts directly with another component of the secretory pathway, namely Sec3, which is part of the exocyst complex implicated in targeting post-Golgi secretory vesicles to the plasma membrane. Accordingly, coexpression of Sec3 increased the uptake activity of GLYT1, suggesting that the exocyst complex augment insertion of GLYT1 into the plasma membrane (Cubelos et al., 2005a).

6. The Dopamine D₂ Receptor.

Two independent studies have shown that the dopamine D₂ receptor short variant, which is the primary autoinhibitory receptor expressed in presynaptic dopaminergic neurons, interacts with DAT and causes an increase in dopamine uptake by increasing in DAT surface expression (Bolan et al., 2007; Lee et al., 2007) (Fig. 10). The potential biological significance of physical and functional coupling between D₂ receptors and DAT activity seems highly plausible considering the shared attenuating role of DAT and D₂ receptors in control of dopaminergic transmission. However, the molecular mechanism underlying the effect of D₂ receptors on the function of DAT is still not fully understood. A direct interaction between the N terminus of DAT and the third intracellular loop of the D₂ receptors was suggested to be responsible for enhanced DAT transport capacity (Lee et al., 2007). Furthermore, up-regulation of DAT activity by coexpression of D₂ receptors has been found to be dependent on receptor activation as well as downstream ERK1 signaling (Bolan et al., 2007). Further studies are required to fully understand the significance of the interaction between DAT and the D₂ receptor. It is highly interesting, however, that a correlation between schizophrenia and potential loss of the DAT/D₂ receptor interaction recently has been reported. Coimmunoprecipitation of DAT with D₂ receptors was reduced to 60% in samples from the striata of patients with schizophrenia compared with the control group, and in samples from patients with bipolar disease, no difference was observed (Lee et al., 2009).

7. α-Synuclein.

DAT, NET, and SERT have all been found to interact with α-synuclein, a presynaptic protein that is implicated in the pathogenesis of Parkinson's disease (Lee et al., 2001; Wersinger and Sidhu, 2005; Wersinger et al., 2006a,b; Jeannotte and Sidhu, 2007). Coexpression studies revealed that α-synuclein modulated transporter activity by regulating cell surface expression; however, it is not clear whether the regulation is stimulatory or inhibitory. Recent studies in human neuronal cell lines revealed ∼50% reduction in DAT activity upon siRNA knockdown of α-synuclein (Fountaine and Wade-Martins, 2007), suggesting that under native conditions, endogenous α-synuclein promotes transporter activity. It is noteworthy that it was shown that parkin, a protein believed to be involved in a familial form of Parkinson's disease, disrupts the α-synuclein/DAT interaction and abolish α-synuclein-induced enhancement of DAT transport in transfected cells (Moszczynska et al., 2007). It was suggested that this mechanism could account for the protective effect of parkin with respects to dopaminergic cell death in Parkinson's disease (Moszczynska et al., 2007). The importance of these interaction effects remains to be determined in vivo. So far, α-synuclein KO mice have failed to show any alterations in DAT function (Dauer et al., 2002; Chandra et al., 2004).

Additional SLC6 NTT interacting proteins are listed in Table 7. In general, the plethora of proteins so far found to interact with the SLC6 NTTs clearly implies that the transporter function and subcellular distribution is tightly regulated by protein-protein interactions. However, the in vivo significance and mechanistic role of most of these interactions is still far from fully understood and require further validation and exploration.

1. Monoamine Transporter Drugs.

As described in section IV, the monoamine neurotransmitter systems are crucially involved in control of human behavior and emotional states, and diseases related to dysfunctions in monoaminergic activity are typically of affective and behavioral nature, such as depression, anxiety disorders (including OCD), and ADHD.

2. GABA Transporters

1. Premature Ejaculation.

Male sexual dysfunction is a common side effect associated with monoamine transporter inhibitors, in particular SSRIs, and includes polymorphic alterations in libido, arousal, erection, and orgasm as well as erectile dysfunction (Ferguson, 2001; Clayton et al., 2002; Labbate et al., 2003; Nurnberg et al., 2003; Balon, 2006). Although SSRI treatment seems to negatively affect all steps in the male sexual response cycle (Corona et al., 2009), the effect of delayed ejaculation is considered a potential treatment of premature ejaculation (PE), illustrated by the off-label usage of antidepressants for this condition (Waldinger et al., 2004; Fallon, 2008). PE is the most common male form of sexual dysfunction (Giuliano and Hellstrom, 2008) and was historically considered a psychological disorder but has recently been suggested to have a neurobiological component related to disturbance in monoaminergic neurotransmission (Waldinger, 2006; Giuliano, 2007). The general side-effect profile and long half-life of conventional SSRIs make these nonoptimal for PE treatment and have prompted development of short-acting SSRIs as treatments for PE. One example hereof is dapoxetine (Table 8; Fig. 11), an SSRI with rapid absorption and shorter half-life than conventional SSRIs; after favorable performance in several clinical trials, dapoxetine has been approved (Hellstrom, 2009).

2. Obesity.

Obesity is associated with numerous medical complications including type 2 diabetes, cardiovascular disease, hypertension, depression, coronary artery disease and stroke, and therefore represents a serious health problem (Flegal et al., 2007). All three monoamines are implicated in regulation of appetite and energy homeostasis (Ramos et al., 2005), and the SNRI sibutramine has been one of the most widely used pharmacological treatments against obesity (Finer, 2002). Unfortunately, long-term sibutramine treatment are associated with severe cardiovascular side effects (James et al., 2010), and it has therefore been withdrawn from several markets, including Europe and the United States. The future potential of this compound remains uncertain, but a recent study found that coadministration of the 5-HT_1A receptor agonist eptapirone (F11440) attenuates the sibutramine-induced side-effects without compromising the effect on weight loss in animal models (Thomas et al., 2009), suggesting coadministration of 5-HT_1A receptor agonists with sibutramine as a potential weight loss therapy (Thomas et al., 2009). Sibutramine is rapidly metabolized in vivo to its desmethyl and didesmethyl congeners, which are potent inhibitors of all three monoamine transporters (Glick et al., 2000), indicating that combined SERT, NET, and DAT inhibition might offer an effective treatment against obesity. The potential of triple uptake inhibitors as weight-loss therapy has been demonstrated by the triple SERT/NET/DAT inhibitor tesofensine (Fig. 11), which produces significant weight loss in humans, presumably through a mechanism that involves indirect stimulation of α₁ adrenoceptor and dopamine D₁ receptor pathways (Axel et al., 2010), and is currently in clinical development for the treatment of obesity (Astrup et al., 2008). The combination of bupropion (NET/DAT inhibitor) with zonisamide, an anticonvulsant modulator of voltage-gated Na⁺ and Ca²⁺ ion channels that has been found to also modulate SERT and DAT activity, has further implied that modulation of all three monoamine transporters is beneficial in the pharmacological treatment of obesity. When administered alone, bupropion and zonisamide each induce modest weight loss (Gadde et al., 2001, 2003; Anderson et al., 2002), whereas an additive effect is achieved when they are administered in combination (Gadde et al., 2007). Bupropion has also shown promising results in combination with the opioid receptor antagonist naltrexone, which blocks the inhibitory feedback loop through pro-opiomelanocortin neurons, believed to limit sustained weight loss. The bupropion/naltrexone combination therapy is currently in late-stage phase III trials as a novel treatment for obesity (Greenway et al., 2009).

3. Schizophrenia.

Compounds that block the NMDA receptors induce schizophrenia-like psychotic symptoms, indicating that glutamatergic hypofunction is involved in the etiology of schizophrenia (Tsai and Coyle, 2002). The use of NMDA receptor agonists in the treatment of schizophrenia is associated with severe side-effects, including brain damage and seizures, and indirect modulation of NMDA receptors is considered a more promising strategy (Tsai and Coyle, 2002). Gly is a coagonist of NMDA receptors; by controlling the synaptic clearance of Gly, GLYT1 is involved in the regulation of glutamatergic neurotransmission (Tsai et al., 2004; Gabernet et al., 2005). Selective inhibitors of GLYT1 enhance glutamatergic signaling by increasing the Gly occupancy at NMDA receptors; hence, such compounds might be beneficial in the pharmacological treatment of patients with schizophrenia (Kinney et al., 2003; Sur and Kinney, 2004). Several selective GLYT1 inhibitors have been identified (section III), and some of these are currently in clinical development against schizophrenia, including (4-(3-fluoro-5-trifluoromethylpyridin-2-yl)piperazin-1-yl)(5-methanesulfonyl-2-(2,2,2-trifluoro-1-methylethoxy)phenyl)methanone (RG1678) (Fig. 11), which recently entered phase II trials (Pinard et al., 2010). Existing pharmacological treatments generally reduce the positive symptoms (such as hallucinations, delusions, and disturbance of thought) of schizophrenia, whereas GLYT1 inhibitors reduce the negative symptoms (such as apathy and social incompetence) and cognitive symptoms (impairment of attention, memory, and executive functions) (Tsai and Coyle, 2002; Thomsen, 2006).

4. Pain.

TCAs and dual-acting SNRIs are widely used in the treatment of pain (Micó et al., 2006). This establishes an important role for 5-HT and norepinephrine in modulation of pain, although the mechanisms by which antidepressant drugs modulate nociceptive signals remain unclear. Bicifadine (Fig. 11) is a potent inhibitor of SERT and NET and is currently in development against neuropathic pain (Krieter et al., 2008). It is noteworthy that bicifadine also possess inhibitory activity against DAT (Basile et al., 2007), which indicates that inhibition of all three monoamine transporters might provide a novel and efficacious strategy for the treatment of neuropathic pain. Tapentadol [(-)-(1R,2R)-3-(3-dimethylamino-1-ethyl-2-methyl-propyl)-phenol] combines NRI activity with μ-opioid receptor agonism in a single molecule. Although tapentadol has low affinity for NET, both mechanisms of action contribute to the broad-spectrum analgesic properties of the compound (Schröder et al., 2011), and tapentadol is approved for the treatment of moderate to severe acute pain in adults.

The χ-conotoxin MrIA is a 13-residue peptide isolated from the venom of the mollusk-hunting marine snail Conus marmoreus. MrIA is selective and noncompetitive inhibitor of NET (Sharpe et al., 2001, 2003) that has been shown to suppress neuropathic pain upon intrathecal administration to rodents (McIntosh et al., 2000). Further development of MrIA led to the identification of an analog with increased in vivo stability, Xen2174 (Sec-Gly-Val-Cys-Cys-Gly-Tyr-Lys-Leu-Cys-His-Pyl-Cys), which has proved effective against both nociceptive and neuropathic pain (Nielsen et al., 2005; Brust et al., 2009). Xen2174 have recently entered phase II clinical trials against postoperative pain (Brust et al., 2009).