Molecular determinants for selective recognition of antidepressants in the human serotonin and norepinephrine transporters

Abstract

Inhibitors of the serotonin transporter (SERT) and norepinephrine transporter (NET) are widely used in the treatment of major depressive disorder. Although SERT/NET selectivity is a key determinant for the therapeutic properties of these drugs, the molecular determinants defining SERT/NET selectivity are poorly understood. In this study, the structural basis for selectivity of the SERT selective inhibitor citalopram and the structurally closely related NET selective inhibitor talopram is delineated. A systematic structure-activity relationship study allowed identification of the substituents that control activity and selectivity toward SERT and NET and revealed a common pattern showing that SERT and NET have opposite preference for the stereochemical configuration of these inhibitors. Mutational analysis of nonconserved SERT/NET residues within the central substrate binding site was performed to determine the molecular basis for inhibitor selectivity. Changing only five residues in NET to the complementary residues in SERT transferred a SERT-like affinity profile for R- and S-citalopram into NET, showing that the selectivity of these compounds is determined by amino acid differences in the central binding site of the transporters. In contrast, the activity of R- and S-talopram was largely unaffected by any mutations within the central substrate binding site of SERT and NET and in the outer vestibule of NET, suggesting that citalopram and talopram bind to distinct sites on SERT and NET. Together, these findings provide important insight into the molecular basis for SERT/NET selectivity of antidepressants, which can be used to guide rational development of unique transporter inhibitors with fine-tuned transporter selectivity.

Imbalances in neurotransmission involving the monoamines serotonin (5-hydroxytryptamine; 5-HT) and norepinephrine (NE) are implicated in depression and anxiety disorders (1). In the brain, specific monoamine transporters, the 5-HT transporter (SERT) and the NE transporter (NET), curtail the lifetime of extracellular monoamines by performing active uptake (or reuptake) from the extracellular space into neurons. Medications for the treatment of depression and anxiety disorders act by increasing the extracellular concentration of 5-HT and/or NE by inhibiting SERT and/or NET mediated transmitter reuptake (2). SERT and NET belong to the solute carrier 6 (SLC6) transporter family, and they are integral membrane proteins that use cotransport of sodium as an energy source to convey neurotransmitters from the extracellular space to the cytoplasm (3). The first generation of drugs targeting SERT and NET were the tricyclic antidepressants (TCAs), but their activity across a variety of other neurotransmitter receptor systems (4) associate their use with severe side effects. Development of newer generations of monoamine transporter drugs have focused on compounds with an improved selectivity toward SERT and/or NET, exemplified by the selective 5-HT reuptake inhibitors (SSRIs), the selective NE reuptake inhibitors (NRIs), and the dual inhibitors of SERT and NET (SNRIs). These drugs are among the most commonly prescribed for treatment of depression and anxiety disorders and are being used in an increasing number of unique therapeutic applications (5).

Balancing the inhibitory activity at SERT and NET is a key determinant for the therapeutic properties of drugs targeting monoamine transporters (6). Importantly, delicate modifications of the same chemical skeleton can tweak selectivity between the two transporters and have in several cases been used to develop both SSRIs, SNRIs and NRIs, from the same chemical scaffold (5). Although the clinical significance of compounds with distinct SERT/NET selectivity profiles has been recognized for more than 30 years, the molecular basis for SERT/NET selectivity is poorly understood, mainly due to limited insight into the localization and structure of the inhibitor binding pockets in SERT and NET. High-resolution X-ray structures of a bacterial homolog to SLC6 transporters (LeuT) have been reported (7–11), providing insight into the tertiary structure of this class of proteins. The structures revealed a topology of 12 transmembrane (TM) spanning regions connected by short intra- and extracellular loops with a high-affinity substrate binding site (denoted the S1 site) centrally located in the core of the transporter protein (12). LeuT has proved to be an excellent structural template for construction of homology models of SERT and NET, facilitating identification of the location and molecular structure of binding pockets for substrates, ions, and inhibitors (13–18).

In this study, we delineate the structural basis for SERT/NET selectivity of the SSRI citalopram and the structurally closely related NRI talopram. Through systematic structure-activity relationship studies, we identify specific substituents as key determinants for inhibitory activity and selectivity toward SERT and NET. Furthermore, we find that switching nonconserved SERT/NET residues within the S1 site fully transfers citalopram sensitivity to NET and renders SERT insensitive to citalopram, thereby demonstrating that the selectivity of citalopram is determined solely by the nature of the central substrate binding pocket in SERT and NET. In contrast, we find that the NRI properties of talopram are remarkably unaffected both by perturbations of the S1 pockets in SERT and NET, as well as in the outer vestibule in NET, which has been proposed to harbor an inhibitor binding site (denoted the S2 site) (10, 11), suggesting that talopram is accommodated at a site distinct from the S1 and S2 binding sites. Thus, we demonstrate that two structurally closely related compounds possibly can have distinct binding sites on the same transporter protein.

Results

Structural Features of Inhibitors Underlying Activity and Selectivity.

Citalopram is among the most selective SERT inhibitors and the structurally related counterpart, talopram, is a potent and selective inhibitor of NET (Fig. 1). The binding affinity of citalopram and talopram was determined by displacement of ¹²⁵I-labeled (-)-2β-carbomethoxy-3β-(4-iodophenyl)tropane (β-CIT) binding to recombinantly expressed human SERT or NET and, as expected, citalopram had high affinity and selectivity toward SERT over NET (4 nM versus 1,414 nM), whereas talopram had high affinity and selectivity toward NET over SERT (9 nM versus 719 nM) (Fig. 1 and Table S1). The two chiral compounds share a phenyl-substituted phthalane skeleton and a propylamine moiety, and they are distinguished by four chemical substituents only (Fig. 1). To delineate the role of these four diverging structural elements for activity at SERT and NET, we used a previously prepared set of 16 compounds comprising all possible combinations of the differing substituents (19) (Fig. 1) and determined the inhibitory potency (K_i) of each compound at SERT and NET in a functional uptake inhibition assay (Fig. 1, Fig. S1, Table S2, and SI Experimental Procedures). The systematic structure-activity analysis allowed us to dissect the role of each of the four substituents for inhibitory activity toward SERT and NET (Fig. S1). Furthermore, we found a striking relationship between the substitution pattern of the 16 compounds and their SERT/NET selectivity profile (Fig. 1C), leading to succinct information on the structural determinants for selectivity. Citalopram was the most potent and selective SERT inhibitor, whereas talopram was the most potent and selective NET inhibitor of the 16 compounds (Fig. 1, Fig. S1, and Table S2). We observe that the cyano- and phthalane methyl groups are key determinants for inhibitory activity and selectivity toward SERT and NET, respectively. Notably, compound 10 is a SNRI with high affinity for both SERT and NET (32 nM versus 44 nM) (Table S1), which reaffirms that by subtle perturbations of the same chemical scaffold, the selectivity ratio between SERT and NET can be controlled.

Fig. 1.

(A) Chemical structures of citalopram, talopram, and the 14 compounds (2–15) comprising all possible combinations of the four substituents that distinguish citalopram from talopram. The differing substituents are colored blue (citalopram substituents) or green (talopram substituents), respectively. (B) Citalopram (Left), compound 10 (Center) and talopram (Right) were assessed for their ability to displace binding of [¹²⁵I]β-CIT to SERT and NET. Data points represent the mean ± SEM from triplicate determinations. (C) The inhibitory potency (K_i) was determined for the 16 inhibitors in a functional uptake inhibition assay (Table S2), and the SERT/NET selectivity ratio were calculated as K_i(NET)/K_i(SERT) or -K_i(SERT)/K_i(NET) for SERT and NET preferring inhibitors, respectively. Inhibitors with >10-fold selectivity are highlighted in blue (SERT selective) or green (NET selective).

SERT and NET Have Opposite Stereochemical Requirements.

Citalopram and talopram are racemic mixtures, and it is well-established that the inhibitory potency of citalopram toward SERT resides in the S-enantiomer (16, 20, 21), as exemplified by Lexapro. Accordingly, we found that the affinity of S-citalopram was 35-fold higher for SERT compared with R-citalopram (4 nM versus 136 nM), whereas both enantiomers displayed low affinity binding to NET (3,025 nM versus 1,516 nM) (Fig. 2 and Table S1). The stereochemical properties of talopram had not been examined. We therefore resolved talopram into the R- and S-enantiomers by preparative supercritical fluid chromatography (SFC), crystallized one of the enantiomers as a hydrotosylate salt, and used X-ray analysis to unequivocally identify this isomer as the R-enantiomer (Fig. 2 and SI Experimental Procedures). Subsequent characterization of R- and S-talopram showed that the R-enantiomer has a remarkable 685-fold higher affinity for NET compared with S-talopram (3 nM versus 1,986 nM), whereas both enantiomers displayed low affinity binding to SERT (1,052 nM versus 2,752 nM) (Fig. 2 and Table S1). Interestingly, SERT and NET display opposite enantioselectivity, because S-citalopram has a 775-fold preference for SERT over NET (4 nM versus 3,025 nM) and R-talopram has almost 1,000-fold preference for NET over SERT (3 nM versus 2,752 nM). In contrast, both R-citalopram and S-talopram are low affinity SNRI compounds with <12-fold selectivity between SERT and NET (Fig. 2 and Table S1). Hence, the R/S-configuration of the inhibitors is a critical determinant for SERT/NET selectivity, which had not been appreciated. To explore the generality of this finding for this compound series, we resolved the R- and S-enantiomers of compound 10 by preparative SFC, crystallized one of the enantiomers as a hydrotosylate salt and identified this isomer as the S-enantiomer by X-ray analysis (Fig. 2 and SI Experimental Procedures). Subsequent characterization showed that S-10 was a high affinity SSRI, with 47-fold selectivity for SERT over NET (28 nM versus 1,288 nM), whereas R-10 was a high affinity SNRI compound, with sixfold selectivity for NET over SERT (29 nM versus 171 nM). These results suggest a pattern for this set of citalopram/talopram analogs, where SERT inhibition resides in the S-enantiomer and NET inhibition resides in the R-enantiomer (Fig. 2). This difference in enantioselectivity between SERT and NET has recently also been observed for other citalopram analogs (21).

Fig. 2.

(A) The binding affinity (K_i) of pure S- and R-enantiomers (black and white squares, respectively) of citalopram, talopram, and compound 10 was determined at SERT and NET (Table S1). (B) Single crystal X-ray structures of hydrotosylate salts of S-10 (Left) and R-talopram (Right). C, gray; H, white; F, green; N, blue; O, red. Displacement ellipsoids are shown at 30% probability level for non-H atoms, and the counterion has been omitted for clarity.

Residues in the Central Substrate Binding Site Control SERT/NET Inhibitor Selectivity.

It has been demonstrated that S-citalopram has a competitive mode of inhibition in SERT (22, 23), which is consistent with overlapping binding sites for 5-HT and S-citalopram (13, 16). To elucidate the mode of inhibition by R-talopram, we performed uptake-saturation experiments with SERT and NET in the absence or presence of inhibitors (SI Experimental Procedures), which revealed a competitive mode of inhibition for S-citalopram and R-talopram in both SERT and NET (Table S3). Hence, it is tempting to speculate that the determinants for inhibitor selectivity are located within or in close proximity of the substrate binding site (S1). From a LeuT-based homology model of SERT (13), we identified 47 residues located within 6 Å of the S1 site, whereof 15 are nonconserved between SERT and NET (Fig. 3 and Fig. S2). Initially, we mutated all of the 15 residues in SERT and NET to the corresponding residues in the other transporter, leading to a 15-fold SERT mutant (SERT-15) and a 15-fold NET mutant (NET-15) and envisioned that these mutations would generate a NET-like S1 binding pocket in SERT and vice versa. SERT-15 and NET-15 were transfected into COS-7 cells, but did not display measurable uptake. However, [¹²⁵I]β-CIT was able to bind SERT-15 in membrane preparations from transfected cells with K_D comparable to wild-type (WT) SERT (3.9 nM versus 5.2 nM), whereas no measurable binding was observed in similar preparations of NET-15. Citalopram and its enantiomers showed up to 753-fold loss of affinity (as observed for S-citalopram) at SERT-15 compared with WT SERT (Table S1). Specifically, the affinity for R-citalopram decreased toward the affinity observed in WT NET (750 nM versus 1,516 nM), whereas the affinities for citalopram and S-citalopram were completely reversed to those observed in WT NET (1,305 nM versus 1,414 nM for citalopram; 2,938 nM versus 3,025 nM for S-citalopram). In contrast, SERT-15 induced <sevenfold changes in the affinity of talopram and its enantiomers (Table S1). Hence, inserting a NET-like binding pocket in SERT did not render the transporter sensitive to talopram, suggesting that talopram binds to a site distinct from the S1 site or that talopram binds in the central binding site of SERT without important interactions to any of the 15 mutated residues.

Fig. 3.

(A) Amino acid sequence alignment of nonconserved residues located within 6 Å of the central substrate binding site in human SERT and NET (24). (B) Homology model of NET (Left) and SERT (Right). The substrate is shown in yellow, sodium ions in magenta, and nonconserved residues are highlighted in green (NET) or blue (SERT). (C) The inhibitory potency of talopram (black bars) and citalopram (gray bars) was determined at single point SERT and NET mutants (Table S4). Data represent the mean ± SEM from at least three independent experiments each performed in triplicate.The stipulated line indicates the K_i value for the inhibitors at WT transporters. Asterisks denote significant different K_i value compared with WT transporters (Student's t test; P < 0.05).

Identification of Specific Residues That Control Citalopram Selectivity.

To test whether inhibitor selectivity is conferred by a single residue among the 15 nonconserved residues in the S1 pockets of SERT and NET, we mutated each of the nonconserved residues to the corresponding residue in the other transporter (Fig. 3). The 30 point mutants showed activity ranging from 10% to 97% compared with WT transporters, and substrate K_M were generally comparable to WT transporters (Table S4). The point mutants in SERT generally had a small effect on K_i for citalopram (<fivefold changes), except for A173G and A441G, which induced a 14- and a sevenfold increase in citalopram potency, respectively. None of the point mutants in SERT had a significant effect on the K_i of talopram. Similarly, the point mutants in NET generally induced <fivefold changes in the potency of citalopram and talopram, except for A77G and A426G, which induced a 16- and a 10-fold decrease in K_i for citalopram, respectively, and V148I and F150S, which increased the K_i of talopram by nine- and sixfold, respectively (Fig. 3 and Table S4).

Because single substitutions of nonconserved residues could not reverse the SERT/NET selectivity of citalopram and talopram, we tested whether inhibitor selectivity could be changed by combining the existing single mutants into 10 multiple SERT mutants (designated S1–S10), and 8 multiple NET mutants (designated N1–N8) (Fig. 4). Mutant design was based on an iterative process directed by results from the single mutants and later combined with results from multiple mutants. Accordingly, a fivefold decrease in citalopram potency was observed by the double SERT mutant S1 (Y95F-G100A). Further elaboration of this mutant led to S7 (Y95F-G100A-I172V-A173G-S174F) and S10 (Y95F-G100A-L443M-G445A-C473T), which induced a 38- and 16-fold increase in K_i for citalopram, respectively. However, none of the multiple SERT mutants were able to completely shift the K_i of citalopram to that found in WT NET. Talopram potency was only moderately affected by the multiple SERT mutants (<fivefold changes in K_i), consistent with the lack of effect on talopram from the SERT-15 mutant.

Fig. 4.

Topology diagram of SERT (blue) and NET (green) and graphical representation of multiple mutants. The identity, TM location, and numbering of the 15 nonconserved SERT/NET residues within the central substrate binding site are shown. SERT mutants are shown on a blue background, with mutations indicated in green (Upper), and NET mutants are shown on a green background, with mutations indicated in blue (Lower). The inhibitory potency (K_i) for citalopram (gray bars) and talopram (black bars) were determined in a functional uptake inhibition assay (Table S4). The stipulated line indicates the K_i value for the inhibitors at WT transporters. Asterisks denote significantly different K_i value compared with WT transporters (Student's t test; P < 0.05).

Compared with WT NET, the double NET mutant N1 (F72Y-A77G) induced a 23-fold decrease in citalopram K_i, and elaboration of this mutant led to N5 (F72Y-A77G-M424L-A426G), which further decreased the K_i of citalopram (46-fold compared with WT NET). Remarkably, introduction of T453C into N5 (N7; F72Y-A77G-M424L-A426G-T453C) promoted a 515-fold increase in the potency of citalopram, leading to a K_i value equivalent to that observed in WT SERT (39 nM versus 42 nM). Characterization of the enantiomers of citalopram at N7 showed that K_i values of both R- and S-citalopram were reversed to those observed in WT SERT (Table S2). Thus, by introducing five SERT-residues into the S1 site of NET, we completely reversed the selectivity for citalopram and its enantiomers. In contrast, talopram was largely unaffected by any of the multiple NET mutations (<threefold changes in K_i value) (Fig. 4 and Table S4). To further probe the role of the S1 site in NET for talopram binding, we introduced three additional mutations (N153S, F323Y, and S419T), corresponding to three SERT mutants (N177S, F341Y, and S438T) that perturb binding of S-citalopram in SERT with up to two orders of magnitude (13, 16). These mutations induced <threefold changes in the inhibitory potency of R-talopram (Table S5), further supporting that talopram does not bind to the S1 site in NET.

To confirm that gain of citalopram potency induced by the N7 mutant reflects a concomitant gain in binding affinity, we determined the binding affinity of citalopram and its enantiomers at N7. Compared with WT NET, N7 induced a 40- and a 65-fold gain in binding affinity of citalopram and S-citalopram, respectively (1,414 nM versus 34 nM for citalopram; 3,025 nM versus 46 nM for S-citalopram), and the affinity of R-citalopram was increased from 1,516 nM at WT NET to 306 nM at N7, which is within same range as the 136 nM affinity displayed at WT SERT (Table S1). The binding affinity of talopram and its enantiomers was only moderately affected by N7 (Table S1).

Mutations Within the Central Substrate Binding Site Selectively Affect SERT/NET Inhibitors.

To further elucidate the apparent differences in molecular mode of action between citalopram and talopram, we determined K_i of compounds 2–15 at N7 and found a correlation between the SERT/NET selectivity of the inhibitors and their sensitivity toward N7 (Fig. 5). SERT selective inhibitors were more sensitive to N7 compared with NET selective inhibitors. Because N7 contains point mutations within the central S1 pocket, this correlation could indicate that SERT selective inhibitors bind in the S1 site, whereas inhibitors with NET selectivity are accommodated at a site distinct from the central binding site.

Fig. 5.

(A) The inhibitory potency of citalopram, talopram, and compounds 2–15 was determined at N7 (Table S4), and the mutation induced loss of potency (calculated as log[K_i(WT NET)/K_i(N7)]) were plotted against the SERT/NET selectivity ratio of the inhibitors (calculated as log[K_i(WT SERT)/K_i(WT NET)]). (B) Cross-sectional illustration showing the location of the central substrate binding site (S1) and the vestibular S2 site in a homology model of human NET in complex with the substrate norepinephrine (shown as yellow spheres). (C) Overlay of the S1 binding site in NET with S-citalopram from a homology model in complex with SERT (13) viewed parallel to the membrane (Left) and from extracellular side (Right). S-citalopram is shown in yellow, and the five key residues that dictate the selectivity of citalopram and its enantiomers are shown in green.

Next, we studied the three prototypical SSRIs, sertraline, fluoxetine, and paroxetine. If the SSRIs showed similar sensitivity as citalopram to N7, it would indicate that the inhibitors share mutual important contact points with citalopram in the S1 site. For sertraline and paroxetine, we observed a fourfold gain in binding affinity at N7 compared with WT NET, whereas fluoxetine was unaffected by N7 (Table S1). We also examined the prototypical NRIs nisoxetine, reboxetine, and atomoxetine. Reboxetine was unaffected by N7, whereas a threefold decrease in affinity for nisoxetine was observed (Table S1). For atomoxetine, the moderate selectivity for NET over SERT was completely reversed by N7 (Table S1), indicating that recognition of citalopram and atomoxetine is controlled by the same determinants within the S1 site.

Talopram Is Largely Unaffected by Mutations in the Extracellular Vestibule.

The finding that the SERT/NET selectivity of citalopram is controlled by S1 residues supports previous work showing that the binding site for of R- and S-citalopram is overlapping the central S1 site in SERT (13, 16, 24). In contrast, talopram activity is generally unaffected by any mutations in the S1 site in SERT and NET (Table S4 and Table S5), indicating that talopram binds to a site distinct from the central binding site. Crystal structures of LeuT in complex with inhibitors have suggested the existence of an inhibitor binding site (S2) in the outer vestibule of the transporter (Fig. 5 and Fig. S3) (8, 10, 11). Several studies have shown that the equivalent site in monoamine transporters can accommodate inhibitors (17, 18, 25, 26), and mutations of S2 residues in SERT and NET have been found to perturb inhibitor affinity (10, 11). To explore whether talopram binds within the S2 site in NET, we introduced 32 single-point mutations into 16 different residues within or in close proximity of the S2 site (Fig. S4 and Table S5). Eighteen of the 32 mutants displayed functional activity that allowed determination of inhibitory potency, and we found that all tested mutations induced <fourfold change in R-talopram K_i (Table S5), indicating that the inhibitor does not bind in the S2 site in NET.

Discussion

The clinical significance of the SERT/NET activity profile of antidepressants has been recognized for decades. However, although molecular determinants defining inhibitor selectivity among related SLC6 transporters have been identified (27–30), the structural determinants within SERT and NET that control inhibitor selectivity have remained poorly understood. The advent of X-ray crystal structures of LeuT has greatly enhanced the understanding of the molecular architecture of the transporters, which allowed us to examine the specific role of nonconserved SERT/NET residues within or in close proximity of the central S1 site in determining selectivity for the SSRI citalopram and the NRI talopram. We find that the selectivity of citalopram, but not talopram, is strictly defined by residues centrally located in SERT and NET. Specifically, a high affinity binding site for citalopram and its enantiomers was generated in NET by switching five residues to the corresponding residues in SERT (N7 mutant). Four of the five key residues are positioned in TM1 (F72 and A77) and TM8 (M424 and A426), which form the substrate binding site together with residues on TM3 and TM6 (9, 31), whereas the fifth residue (T453) is located in TM9 (Fig. 5). F72 in NET and the aligning residue in SERT (Y95) had been found to be important for recognition of citalopram (13, 16, 24, 32), whereas the importance of the remaining four residues has not been recognized to our knowledge. In a LeuT-based homology model of NET, the five key residues encircle the S1 site (Fig. 5), thereby substantiating that the SERT/NET selectivity of citalopram and its enantiomers is determined entirely by residues within the central binding site.

Several recent studies have implied that the binding site for the SSRIs sertraline, paroxetine, and fluoxetine, and the NRI nisoxetine, overlaps the S1 site (24, 25, 33, 34). Generally, these prototypical SERT/NET inhibitors have low sensitivity toward N7 (Table S1), indicating that they have distinct binding modes compared with citalopram within the S1 site, or that they bind to a distinct site (11, 17, 35). Reboxetine was unaffected by N7 (Table S1), and combined with a recent study showing that reboxetine possesses a noncompetitive mode of NET inhibition (23), this finding indicates that the NRI binds to a site different from the S1 site.

The finding that talopram was largely unaffected by any mutations in the S1 site of SERT and NET (Table S4 and Table S5) indicates that talopram binds to a site distinct from the central binding site. We therefore performed a mutational analysis of the S2 site in NET and found that all of the introduced S2 point mutations induced <fourfold changes in the inhibitory potency of R-talopram (Fig. S4 and Table S5), suggesting that R-talopram binds to a yet unidentified binding site that is different from the S1 and S2 sites. Taken together, our results corroborate the proposed presence of distinctive inhibitor binding sites in SERT and NET and highlight the structural and mechanistic diversity of SLC6 transporter inhibition that is emerging (7, 8, 10, 11, 36).

The systematic structure-activity relationship study of citalopram/talopram analogs provided a unique insight into how small modifications of the same chemical scaffold can shape affinity and selectivity at SERT and NET. We found that the two methyl groups on the phthalane ring were essential for activity and selectivity toward NET as proposed (37), whereas the cyano group was the principal determinant for selectivity and affinity toward SERT (Fig. 1). Furthermore, a common pattern showing that SERT and NET have opposite preference for the stereochemical configuration of these inhibitiors was found (Fig. 2). Similar patterns have been observed for sertraline and venlafaxine (38, 39), whereas SERT and NET inhibition resides in the same enantiomer for milnacipran (40), thus demonstrating that chirality is generally a key determinant for both inhibitory activity and selectivity toward monoamine transporters. Interestingly, 3D superimposition of the R- and S-enantiomers of citalopram and analogs display a considerable structural overlap (21, 37), which likely reflect that the observed enantioselectivity at SERT is primarily due to subtle differences in binding of the two aromatic ring systems as suggested recently (16).

In summary, we have demonstrated how systematic and subtle changes on the same chemical scaffold can tweak selectivity between SERT and NET, possibly by controlling where the inhibitor binds on the transporter proteins. Taken together with the identification of molecular determinants in SERT and NET that control selectivity of these compounds, new insight is obtained into the principal molecular features that govern SERT/NET selectivity for structurally closely related inhibitors. These findings provide an important framework to guide rational development of unique inhibitors with tailor-made transporter selectivity and, possibly, an improved clinical efficacy.

Experimental Procedures

Molecular Biology, Transport Assays, and Radioligand Binding Experiments.

Generation of point-mutations in human SERT and NET was performed by site-directed mutagenesis using the QuikChange mutagenesis kit (SI Experimental Procedures). Transport measurements were performed essentially as described (13) (SI Experimental Procedures). Binding of [¹²⁵I]β-CIT to membrane preparations of COS-7 cells expressing SERT or NET mutants was performed as described in SI Experimental Procedures.

Resolution of R- and S-Enantiomers of Talopram and Compound 10.

The R- and S-enantiomers of talopram and compound 10 were resolved by preparative SFC and crystallized as hydrotosylate salts (SI Experimental Procedures). The absolute configuration of the crystallized enantiomers was determined by single crystal X-ray analysis (SI Experimental Procedures).