Rapid and sustained antidepressant properties of an NMDA antagonist/monoamine reuptake inhibitor identified via transporter-based virtual screening

Abstract

Rational design of lead compounds targeting monoamine transporters (MATs) is critical to developing novel therapeutics to treat psychiatric disorders including depression and substance abuse. A 3-D dopamine transporter (DAT) computer model was used to virtually screen a commercially available small molecules library for high DAT affinity drug-like compounds. One hit, coded "MI-4," inhibited human dopamine, norepinephrine, and serotonin transporters in vitro. In vivo administration in mice induced robust, dose-dependent antidepressant-like behaviors in learned helplessness models (tail suspension and forced swim tests). Moreover, chronic administration (21 day, 10 mg/kg, bid) reduced drinking latencies comparable to fluoxetine (10 mg/kg, bid) in the novelty-induced hypophagia test, which requires chronic treatment to produce antidepressant-like effects. MI-4 (10 mg/kg, bid) produced rapid (three-day) antidepressant-like effects in the social-avoidance test following 10 days of social-defeat stress. Unlike ketamine, chronic administration of MI-4 increased social interaction scores while improving resiliency to the mood-altering effects of stress to over 70%. Importantly, MI-4 exhibited minimal abuse liability in behavioral and neurological models (conditioned place preference and dopamine in vivo microdialysis). MI-4 was found to be Ro-25-6981, an ifenprodil analog and reputed NMDA antagonist. The data suggest that Ro-25-6981, previously known for rapid-acting glutamatergic antidepressant actions, may also functionally inhibit monoamine reuptake and produces sustained antidepressant effects in vivo. This demonstrates, as proof of principle, the viability of combining these mechanisms to produce rapid and sustained antidepressant-like effects. Overall, these findings suggest MAT computational model-based virtual screening is a viable method for identifying antidepressant lead compounds of unique scaffold.

Graphical Abstract

1. Introduction

Most FDA-approved antidepressants increase synaptic serotonin, norepinephrine, and/or dopamine by blocking their respective transporters, SERT, NET and DAT. The resulting increase in post-synaptic signal transduction occurs in minutes, though alleviation of clinical symptoms is not observed for weeks (Stahl et al., 2013;Immadisetty et al., 2013). This is a tremendous problem for patients experiencing suicidal thoughts, self-harm urges or social self-exclusion who need immediate relief from depressive symptoms. Unfortunately, no rapid-acting FDA-approved therapies currently exist. It is therefore critical to develop rapid-acting antidepressants that still retain the long-term benefits of the plasma membrane monoamine transporter (MAT) inhibitors currently available.

Efforts to discover new medications for psychiatric conditions require synthesis of lead compounds (and their derivatives) and high-throughput pharmacologic testing in CNS or behavioral models, which require considerable resources. Alternatively, costs could be drastically curtailed by virtual (in silico) screening (VS) of candidate lead compounds if a computational model of the drug target is available. Screening of thousands to millions of potential ligands in a structural library using a drug receptor molecular model is a powerful, rapid and inexpensive tool for discovery of lead compounds (Cheng et al., 2012). Fortunately, the breakthrough crystallization of the leucine transporter LeuT (Yamashita et al., 2005), a MAT homolog, and more recently the Drosophila DAT (Penmatsa et al., 2013) provide templates for generation of the first reliable SERT and DAT computational models for MAT-directed drug discovery.

At least two ligand binding sites within the MAT proteins have been proposed. The primary binding site, S1, is at the approximate midpoint of the lipid bilayer and is flanked by gating residues that alternate access between the intracellular and extracellular regions (Yamashita et al., 2005;Forrest et al., 2008;Beuming et al., 2008). One or more binding sites have been proposed for the extracellular side of the external gate of S1, in what is termed the extracellular vestibule (Indarte et al., 2008;Beuming et al., 2008). Among these, a second substrate site, S2, has been proposed to bind the substrate before its translocation to the interior S1 site (Shi et al., 2008). The need for a secondary substrate pocket is in dispute, however (Piscitelli et al., 2010). Psychostimulants, such as cocaine, and antidepressants, including the selective serotonin reuptake inhibitors (SSRIs) fluoxetine and citalopram and the tricyclic antidepressant (TCA) nortriptyline, have been shown to reach the S1 pocket (Beuming et al., 2008;Sinning et al., 2010;Plenge et al., 2012;Sorensen et al., 2012;Wang et al., 2012;Penmatsa et al., 2013;Dahal et al., 2014). TCA drugs can also noncompetitively inhibit MAT substrate binding (Plenge and Mellerup, 1997), consistent with a MAT binding site outside S1. TCA drugs co-crystallized with LeuT were bound in the extracellular vestibule (Zhou et al., 2007); however, the Drosophila DAT crystals place TCA drugs in the S1 pocket (Penmatsa et al., 2013). The S2 site or similarly located vestibular binding pocket may serve as the MAT allosteric inhibitor binding site (Plenge et al., 2012); (Kortagere et al., 2013).

The original goal of the present work was to screen for compounds that could block DAT binding of cocaine without potent inhibition of dopamine uptake. At that time, it was decided to avoid the DAT S1 pocket and focus on S2 (Indarte et al., 2010). A DAT computational model (Indarte et al., 2008) generated using the original LeuT-leucine cocrystal structure as a template (Yamashita et al., 2005) was used for virtual screening (VS) of a structural library of small molecule, drug-like compounds. Superposition of this DAT homology model with the newer Drosophila DAT protein crystal structure (Penmatsa et al., 2013) confirmed a similar spatial location of key interacting residues (Fig 1). This suggests that despite natural and expected differences between the computational model and x-ray crystal data, the in silico approach employed was credible, retrieving active compounds beyond random chance.

Figure 1.

MOE-docked ligand poses within the rDAT model or the dDAT x-ray structure (4M48). Left: The spatial proximity of conserved and similar S1 and S2 residues for the rDAT (blue sticks, annotated) model and dDAT (green sticks) x-ray structure can be seen via superposition. The DAT vestibular S2 pocket is delineated by the Connolly surface (magenta) of rDAT-docked MI-4 (Ro-25-6981; magenta, ball-and-stick). The S1 substrate/inhibitor pocket is outlined by the Connolly surface (atom colored) of docked dopamine (white, ball-and-stick) within rDAT and co-crystallized nortriptyline (yellow, ball-and-stick) within dDAT. Right: The position of the occupied S1 and S2 pockets is shown relative to transmembrane and other α-helical regions (cylinders) for rDAT (green) and dDAT (blue).

The top 20 VS "hit" compounds with respect to predicted DAT binding affinity were ranked, and most of these were tested for affinity to the three MATs. This approach identified the known NMDA glutamate receptor antagonist Ro-25-6981 to be an inhibitor of all three monoamine transporters. Ro-25-6981 showed antidepressant-like responses in behavioral models of mood that provide predictive and constructive validity (tail suspension test, forced swim test, novelty-induced hypophagia and social avoidance test, respectively) comparable to established inhibitors of serotonin and norepinephrine reuptake such as fluoxetine and desipramine. More importantly, the findings described herein suggest that Ro-25-6981, previously characterized as a rapid-acting antidepressant (Li et al., 2011), may also functionally inhibit monoamine reuptake and, in contrast to ketamine, produce sustained antidepressant effects in vivo. These data are the first to show, as proof of principle, the viability of combining monoaminergic and glutamatergic mechanisms that produce both rapid and sustained antidepressant properties.

2. Materials and methods

2.1. Molecular modeling, screening and pharmacology.

An in silico 3-D DAT model based primarily on its alignment with the bacterial leucine transporter protein LeuT_Aa (2A65) (Yamashita et al., 2005) was created as described (Indarte et al., 2008). The model was used to screen a small molecule structural library for compounds with predicted high affinity at the DAT. The library contained 140,000 compounds selected from the Sigma-Aldrich catalog. The selected compounds were determined by eliminating those compounds that did not pass the Lipinski rules for drug-like compounds or contained toxic functional groups. The virtual screen and subsequent in vitro pharmacology were conducted as described (Indarte et al., 2010).

2.2. Behavioral testing.

C57BL/6J mice were bred in house using stock breeders obtained from Jackson Laboratories (Bar Harbor, Maine). At weaning, male mice were group housed with same-sex littermates and behavioral testing was conducted during the light phase of the light/dark cycle between 9:00 am and 12:00 pm. Animals were typically between 8 and 12 weeks of age at the time of testing. All procedures were IACUC-approved and followed the NIH guidelines outlined in “Using Animals in Intramural Research.”

2.3. Tail suspension test.

Using adhesive tape, mice were individually suspended by the tail (approximately 2 cm from the tip) from a metal bar elevated 30 cm above the laboratory bench (Talbot et al., 2010;Steru et al., 1985). Behavior was videotaped for 6 min. Videos were later scored for immobility time (sec) using a stopwatch by a blinded observer. Immobility behavior was defined as the absence of limb movement.

2.4. Forced swim test.

Mice were evaluated in the Porsolt forced swim test with modifications using a pre-test/test paradigm as described (Hirani et al., 2002;Talbot et al., 2002). Twenty-four hours prior to testing, mice were subjected to a 15 min pre-swim in approximately 10 cm of 25°C water in a 20 cm x 13 cm glass cylinder. On the day of testing, mice were allowed to swim for 5 min and behavior was videotaped and scored for immobility time (sec) by a blinded observer. Immobility behavior was defined as the absence of limb movements except minor movements necessary to keep the animal afloat.

2.5. Conditioned place preference test.

Mice were tested using an unbiased design with a two-compartment apparatus differing in visual and tactile cues (Kelz et al., 1999) where vehicle and drug administration were randomly assigned to compartments. On Day 1, animals freely explored both compartments (30 min). On Days 2-4, animals were restricted to the vehicle-paired compartment (30 min) immediately after vehicle (sterile water) administration (1 ml/kg, i.p., 30 min) followed four hours later by i.p. drug administration and immediate placement in the drug-paired compartment (30 min). Pairings were counter-balanced so that equal numbers of animals received vehicle or drug administration paired with opposing compartments. On Day 5, mice were allowed access to both compartments (30 min); time spent in the drug-paired minus saline-paired compartment was reported as the place preference score.

2.6. Locomotor activity.

Mouse general locomotor activity was assessed in a non-invasive manner using an Opto-Varimax 4 Activity Analyzer (Columbus Instruments, Columbus, OH) that measures animal activity in a cage (44.5 cm x 44.5 cm x 20.3 cm) intersected with photocells detecting infrared beams from opposing sides, 2.5 cm apart and 2 cm above the floor (Talbot et al., 2010;Nolan et al., 2011;Desai et al., 2014). After 30 min of habituation to the testing room, and approximately 30 min following drug administration, animals were placed in the apparatus and activity (distance traveled) was measured for 20 min.

2.7. Novelty-induced hypophagia.

Mice were evaluated in the novelty-induced hypophagia test as described (Talbot et al., 2010;Dulawa and Hen, 2005) using a five-day protocol with non-fasted animals in home- and novel-cage environments. Animals had ad lib access to food and water with the exception of the training/testing sessions. A commercially available, sucrose-sweetened nutritional supplement (Ensure™; Abbott Laboratories; Chicago, IL) was delivered by serological pipette. During the first three days subjects were given access to 1.0 ml of Ensure in a 2.0 mL serological pipette attached to a sipper tube during once daily 60-min training sessions. Home-cage consumption was measured on Day 4 for 30 min. Consumption in a novel cage was assessed on Day 5 by placing each animal in a new cage with no bedding in full light for 30 min. Drinking latency times and volumes were recorded for both home- and novel-cage testing.

2.8. Social defeat stress and social avoidance test.

Studies were performed essentially as described (Golden et al., 2011). Briefly, chronic stress was induced by repeated daily exposure of C57BL6/J mice for 10 consecutive days (Days 1-10) to an “aggressor” animal (CD-1 male mouse age 4-6 mo., Jackson Labs). Exposure consisted of direct contact for a period of either 10 min or five bouts of social defeat (defined as repetitive, combative interactions lasting approximately 5-10 sec resulting in retreat from the CD-1 mouse). For the remainder of each 24-hr period, the aggressor and study animal were co-housed separated by a perforated Plexiglas divider, allowing for constant exposure to social cues. Negative (non-defeated) controls were pair-housed for 10 days in defeat boxes with one C57BL6/J mouse per side with no physical contact between cage mates.

On Day 11, stressed and non-stressed mice were subjected to the social avoidance test to assess depression-like behaviors. Animals were placed within the open field of the Motor Activity Analyzer for two three-min trials, in the absence (No target) and presence (Target) of a novel CD-1 aggressor mouse contained in perforated Plexiglas to allow for social interaction. Depressive-like behaviors are associated with social avoidance, which is indicated when an animal spends less time in the interaction zone (14x24 cm area immediately surrounding the Plexiglas enclosure) and more time in the comer zones (9x9 cm areas projecting from both corners opposite the enclosure) in the presence of the CD-1 aggressor. Social interaction is quantified as the time spent (sec) in the interaction zone in Trial 2 minus Trial 1 or as the social interaction ratio (times spent in the zone for “target” divided by “no target”). Social “resiliency” or “susceptibility” is defined as the social interaction ratio > 1.0 or < 1.0, respectively. For experiments assessing TST behaviors following social defeat stress, the TST was performed at least 24 hrs after the conclusion of social defeat and approximately two hours after the social avoidance test (see Timeline of Experimental Protocol).

2.9. Dopamine (DA) microdialysis.

Swiss Webster mice (30–40 g) were surgically implanted with concentric dialysis probes prepared with AN69 fibers (dialyzing surface = 1.0 mm) aimed at the nucleus accumbens shell, as described (Tanda et al., 2009). Within 42-46 h of implanting the probe, dialysate samples (10 μl) were analyzed with HPLC to detect DA (Mereu et al., 2013). Results are expressed as a percentage of basal DA values, calculated as means of 2–4 consecutive samples (differing by no more than 10%) immediately preceding the test drug injection.

2.10. Protection of neuroblastoma cells.

N2a cells were pretreated with the proteasome inhibitor MG132 at an IC₅₀ concentration (2 μM for the ATP assay and 1 μM for the cell number assay) in the absence or presence of a range of MI-4 concentrations (0.05 – 6.4 μM) dissolved in DMSO. Control cells were treated with DMSO. Two days later, viability was assessed by measuring ATP (Cell Titer Glo, Promega). On parallel plates, cells were stained with the nuclear stain DRAQ5 (Biostatus) in conjunction with the cytosolic stain Sapphire (LI-COR) or the Hoechst reagent (Sigma-Aldrich), as previously published (Posimo et al., 2014). Infrared DRAQ5+Sapphire signal was quantified on an Odyssey Imager (LI-COR) and Hoechst-stained cells were photographed on an EVOS epifluorescent microscope (Life Technologies).

2.11. Drugs.

Ro-25-6981 was obtained from Axon Medchem (Groningen, Netherlands). Fluoxetine was obtained from Biotrend Chemicals (Destin, FL) and was used as a prototypical inhibitor of serotonin reuptake at 10 mg/kg. Desipramine (3 mg/kg) and ketamine (10 mg/kg) were purchased from Sigma-Aldrich (St. Louis, MO) and were used as prototypical antidepressant agents acting via inhibition of norepinephrine reuptake and glutamatergic (NMDA receptor) antagonism, respectively. Morphine sulfate (30 mg/kg) and cocaine (10 mg/kg) were also obtained from Sigma-Aldrich (St. Louis, MO) and were used as opioid agonist and dopaminergic positive controls, respectively, for behavioral reinforcement in the conditioned place preference assay. Unless indicated otherwise, all drugs were dissolved in sterile water and administered in vivo by intraperitoneal (i.p.) injection at a volume of 0.1 ml/kg. For acute administration, drugs were administered 30 min prior to behavioral testing. Chronic administration consisted of twice daily (bid) injections (7 AM / 7 PM), unless indicated otherwise.

2.12. Data Analysis.

All data were analyzed using GraphPad Prism 5 Software (San Diego, CA). Data from the TST and FST were averaged within each treatment. Totals for immobility times were analyzed by one-way ANOVA and Tukey-Kramer post hoc tests were used to compare drug-treated groups to vehicle-treated animals. Drinking latency data from the novelty-induced hypophagia test were analyzed by two-way ANOVA accounting for variables of environment (home vs. novel cage) and drug treatment (vehicle vs. drug: Ro-25-6981, fluoxetine) with Tukey-Kramer post hoc analysis. Differences between treatment groups in volume of solution consumed in the novel cage environment were analyzed by one-way ANOVA with Tukey-Kramer post hoc analysis. Following 10 days of social defeat stress, data from the social avoidance test were presented as the social interaction ratio averaged between groups and the mean of the time spent in the interaction zone. Data were analyzed by two-way ANOVA accounting for variables of environment (stress vs. no stress; target vs. no target) and drug treatment (vehicle vs. drug: Ro-25-6981, ketamine, fluoxetine) with Tukey-Kramer post hoc analysis. Drug-induced conditioned place preference was assessed by differences in the mean place preference scores between treatment groups and analyzed by one-way ANOVA with Tukey-Kramer post hoc analysis. Locomotor activity data were summed across 20 min approximately 30 min following injection with varying doses of Ro-25-6981. Locomotor activity counts were averaged across mice in each treatment group and were calculated as distance traveled. Values were compared for each dose by one-way ANOVA with Tukey-Kramer post hoc test.

3. Results

Using the vestibular S2 ligand binding pocket of a DAT computational model (Indarte et al., 2008), virtual screening (VS) of a structural library of small molecule, drug-like compounds yielded a hit compound coded “MI-4” (Fig 1). MI-4 was subsequently determined to be the known NMDA receptor antagonist Ro-25-6981, and is henceforth referred to as such. Ro-25-6981 was found to have binding affinities for hNET (K_i = 365 nM) and hSERT (K_i = 670 nM) as well as the expected affinity for hDAT (K_i = 3460 nM). Ro-25-6981 also inhibited substrate uptake at each transporter using in vitro models (Indarte et al., 2010). This hit compound was pharmacologically tested at 100 nM and 10 μM final concentrations by a screen involving 63 radioligand/enzyme assays. A screening result of at least 50% inhibition by 10 μM Ro-25-6981 was followed by generation of complete inhibition curves to obtain K_i values (Table S1).

Ro-25-6981 was tested for in vivo antidepressant potential using the tail suspension test (TST) and forced swim test (FST), which have strong predictive value for antidepressant-like activity in humans (Cryan and Holmes, 2005). A single injection (30 min) of Ro-25-6981 dose-dependently reduced immobility times in both the TST (Fig 2a; [F(4,35) = 11.9, p < 0.0001]) and FST (Fig 2b; [F(2,23) = 9.42, p = 0.0010]) to a similar degree as that observed with established monoaminergic antidepressants including the SSRI fluoxetine (Prozac; 10 mg/kg, i.p.) or the TCA desipramine (Norpramin; 10 mg/kg, i.p.) (Fig 2c; [F(2,24) = 39.7, p < 0.0001]). Ro-25-6981 also exhibited substantial efficacy in the novelty-induced hypophagia (NIH) test, in which hyponeophagia behaviors are reduced by chronic but not acute or sub-chronic drug administration, mimicking the temporal delay associated with antidepressant efficacy in humans (Talbot et al., 2010). Chronic (21 day) administration of Ro-25-6981 (10 mg/kg, bid) decreased latency to drink in the novel cage environment to half that obtained in vehicle-treated subjects (563±129 sec vs. 204±53 sec) and similar to that obtained with chronic (21 day) administration (10 mg/kg, bid) of the SSRI fluoxetine (292±43 sec) (Fig 2d; effect of drug treatment on latency to drink [F(2,45) = 3.84, p = 0.0289]. Consumption measured by volume increased, though not significantly, in both Ro-25-6981 and fluoxetine treatment groups (Fig 2e).

Figure 2.

Ro-25-6981 dose-dependently decreased immobility times compared to vehicle-treated (Veh) animals in the tail suspension test (a) and the forced-swim test (b), indicating antidepressant-like effects (n=8-11). Data are presented as the mean ± SEM and were analyzed by one-way ANOVA with Tukey-Kramer post hoc test (*p<0.05, and ***p<0.001 vs. the appropriate Veh control). c) Acute administration (30 min) of the SSRI fluvoxamine (FLVX; 10 mg/kg, i.p.) or the TCA desipramine (DSP; 3 mg/kg, i.p.) decreased immobility times compared to water vehicle-treated (Veh) control animals (n=7-9). Data are presented as the mean ± SEM (n=6-11) analyzed by one-way ANOVA with Tukey-Kramer post hoc test (*p≤0.05, **p≤0.01 vs. Veh control). d) Chronic (21 day) administration of Ro-25-6981 (10 mg/kg, i.p., bid) reduced drinking latency times in the novelty-induced hypophagia test compared to vehicle-treated (Veh) animals and to a similar degree as the SSRI fluoxetine (FLX; 10 mg/kg, i.p., bid). Data are the mean ± SEM (n=10-12) analyzed by two-way ANOVA with Tukey-Kramer post hoc test with ***p≤0.01 and *p≤0.05. (e) The volume of Ensure consumed was also measured following chronic administration of vehicle, Ro-25-6981 or fluoxetine (10 mg/kg, i.p., bid). Data are shown as the mean ± SEM (n=10-12) and analyzed by one-way ANOVA with Tukey-Kramer post hoc analysis. Differences in volume consumed were not found to be statistically significant.

In recent years, chronic social stress has been recognized for superior etiological, predictive and discriminative properties as well as face validity relative to the clinical condition observed in depressed humans (Golden et al., 2011). In this model, animals are subjected to 10 days of repeated bouts of social defeat. Following chronic stress, the social avoidance test assesses social “resilience,” a response associated with antidepressant efficacy, or “susceptibility” behavior characteristic of a depressive-like state (Golden et al., 2011), where resiliency or susceptibility is defined as the extent of social interaction with a novel social aggressor (Fig S1, Panels a-d). Ro-25-6981 induced robust and rapid antidepressant-like behaviors in the social avoidance test following three days of administration (10 mg/kg, i.p., bid). Ro-25-6981 significantly increased the social interaction ratio (Fig 3a; [F(1,50) = 4.31, p = 0.0432]) and the time spent in the interaction zone (Fig 3b; [F(1,51) = 4.91, p = 0.0312]) relative to vehicle-treated, stressed mice. Importantly, Ro-25-6981 did not appear to affect behavioral responses in the social avoidance test in non-stressed control animals (Fig 3a) nor in the interaction zone time with the aggressor animal absent (Fig 3b, “No Target”).

Figure 3.

Sub-chronic administration of Ro-25-6981 for 3 days (10 mg/kg, i.p., bid) significantly increased the social interaction ratio (a) and time spent in the interaction zone (b) in the social avoidance test following 10 consecutive days of social defeat stress. Data are expressed as the mean ± SEM (n=18) analyzed by two-way ANOVA with Tukey-Kramer post hoc test (*p≤0.05 and **p≤0.01). Chronic administration of Ro-25-6981 (21 days, 10 mg/kg, i.p., bid) similarly increased the social interaction ratio (c) and time spent in the interaction zone (d), whereas chronic administration of ketamine (10 mg/kg, i.p., bid) did not. Data are expressed as the mean ± SEM (n=16) analyzed by two-way ANOVA with Tukey-Kramer post hoc test (*p≤0.05). e) Chronic administration (21-day) of Ro-25-6981 increased resiliency to 75% (defined as social interaction ratios > 1.0) compared to vehicle-treated (45%) and ketamine-treated (40%) animals (n= 16 each treatment group). f) Both sub-chronic (3-day) and chronic (21-day) administration of Ro-25-6981 reduced immobility scores in the TST following 10 days of social defeat stress to a comparable degree as 21-day administration of fluoxetine (FLX; 10 mg/kg, i.p., bid). Data are expressed as the mean ± SEM (n=6-10) analyzed by one-way ANOVA with Tukey-Kramer post hoc test (**p≤0.01).

Ro-25-6981 and ketamine were also directly compared for sustained antidepressant-like properties. Each drug was administered chronically (21 days at 10 mg/kg, i.p., bid) to animals subjected to social defeat stress and then evaluated in the social avoidance test. Unlike ketamine, Ro-25-6981 significantly increased the social interaction ratio (4.1±1.3) compared to vehicle-treated control animals (1.6±0.3) when chronically administered (Fig 3c; [F(2,78) = 3.71, p = 0.0289]). Likewise, chronic Ro-25-6981 increased the overall time spent in the interaction zone in the presence (53±10 sec) vs. absence (24±3.7 sec) of the aggressor target animal, whereas chronic ketamine had no effect (Fig 3d; [F(1,55) = 4.91, p = 0.0309]). The percentage of Ro-25-6981-treated animals found to be resilient was 75%, compared to 44% resiliency for the vehicle treatment group and 40% resiliency for the ketamine treatment group (Fig 3e), suggesting that Ro-25-6981 may confer resistance to the mood-altering effects of stress. Furthermore, animals evaluated by TST 24 hours following the completion of social defeat stress showed significant decreases in immobility scores in response to both three day and 21 day administration of Ro-25-6981 and to a similar degree as chronic (21 day) administration of fluoxetine (Fig 3f; [F(4,42) = 17.31, p < 0.0001]). Together, these data support a model in which Ro-25-6981 provides both rapid and sustained antidepressant-like effects in several animal models predictive of antidepressant activity.

The potential neuroprotective effects of Ro-25-6981 were explored using a cellular model of neurodegeneration, given the potential utility of neuroprotective agents in treating depression (Enache et al., 2011) and the putative role of neurodegeneration in the pathogenesis of depression (Swaab et al., 2005). As expected, Ro-25-6981 protected neuronal N2A cells against toxicity induced by the proteasome inhibitor MG132 when assessing both ATP levels and nuclear and cytosolic staining (Fig. S2). Remarkably, the protection observed in the ATP assay was almost 100%. These data demonstrate that Ro-25-6981 can provide both structural and functional protection against proteinopathic stress. Previous studies have also shown that Ro-25-6981 can robustly prevent cell death in response to excitotoxicity and ischemia (Fischer et al., 1997;Liu and Zhao, 2013), supporting the view that Ro-25-6981 may possess neuroprotective mechanisms.

An important consideration for compounds targeting the DAT is whether or not they have abuse liability (Kelz et al., 1999). Therefore, the conditioned place preference (CPP) procedure was used, as it exploits the commonality between rodents and humans of displaying a preference for environments associated with receiving a reinforcing substance, such as cocaine. As expected, mice that received cocaine (10 mg/kg) or morphine sulfate (30 mg/kg) spent significantly more time in the drug-paired compartment compared to animals injected with sterile water alone. In contrast, Ro-25-6981 at doses of 3.0 or 10.0 mg/kg produced no increase in time spent in the drug-paired compartment (Fig 4a; [F(4,48) = 5.17, p = 0.0015]). Further, no statistically significant change in nucleus accumbens shell DA levels was detected upon acute administration of 3, 10 or 30 mg/kg Ro-25-6981 (Fig 4b). The nucleus accumbens is a brain region associated with reinforcement that typically shows a rapid surge of DA release from mesolimbic efferents when a drug of abuse is administered (Di Chiara G. et al., 1999) Administration of Ro-25-6981 alone at lower doses had minimal effects on locomotor activity, with only moderately increased distance traveled during the treatment period (Fig S3) . These data are characteristic of monoaminergic antidepressants in their ability to acutely increase DA in the cortex but not in the accumbens shell (Tanda et al., 1994) and suggest that Ro-25-6981 is unlikely to have abuse potential.

Figure 4.

a) Ro-25-6981 (10 mg/kg, i.p.) had no effect on the place preference score (time spent in the drug-paired chamber post- vs. pre-conditioning) compared to morphine sulfate (MS; 30 mg/kg, i.p.) or cocaine (10 mg/kg, i.p.). Data are presented as the mean ± SEM (n=9-14) analyzed by one-way ANOVA with Tukey-Kramer post hoc test (**p≤0.01). b) Ro-25-6981 administration (3–30 mg/kg, i.p.) had no effect on in vivo release of dopamine in the nucleus accumbens (NAc) shell as measured by microdialysis. Data are presented as the mean ± SEM (n=5).

4. Discussion

The present study validates the use of in silico MAT models in antidepressant drug discovery efforts and provides new evidence suggesting that Ro-25-6981 may contribute a monoamine-based influence to the antidepressant response that mirrors in timing and character that observed with traditional antidepressants when chronically administered. More importantly, these data suggest as proof of principle the possibility of combining monoaminergic and glutamatergic mechanisms as a mechanistic strategy that affords both rapid and sustained antidepressant effects.

The current findings suggest the NMDA antagonist and rapid-acting antidepressant-like agent Ro-25-6981 possesses inhibitory effects on monoamine reuptake with a possible contribution to sustained antidepressant activity. Ro-25-6981 exhibited significant though modest affinities for the human monoamine transporters (SERT, NET, and DAT) classically associated with traditional antidepressants and the more recently developed triple reuptake inhibitors. These data are consistent with a previous observation from a large-scale screen for off-target drug binding that Ro-25-6981 interacted with MATs in vitro (Keiser et al., 2009). In vivo, Ro-25-6981 induced antidepressant-like behaviors similar to those observed for traditional serotonin and norepinephrine reuptake inhibitors using multiple animal models: TST, FST, novelty-induced hypophagia, and chronic social stress.

Recent attention in antidepressant drug discovery has focused on the manipulation of glutamate signaling in the CNS. Early studies established a clear link between antagonism of NMDA receptors and the induction of antidepressant behaviors in rodents and humans, including rapid effects that appear within hours of drug administration (CRANE, 1959;Trullas and Skolnick, 1990;Berman et al., 2000;Zarate, Jr. et al., 2006;Li et al., 2011). While ketamine has emerged as a prototypical glutamatergic antidepressant, its utility as a viable long-term antidepressant agent is less clear. The drug’s adverse effect profile, oral bioavailability, and duration of antidepressant action are less than ideal, and it has potential for abuse (Salvadore and Singh, 2013). However, its rapid antidepressant effects and potential for efficacy in treatment-refractory disease have led to intense investigation of drugs with ketamine-like pharmacology. Similar to ketamine, Ro-25-6981 has been shown to induce rapid antidepressant behaviors in animal models following single dose administration (Maeng et al., 2008;Li et al., 2011), likely through stimulation of mammalian target of rapamycin (mTOR)-dependent synaptogenesis. However, Ro-25-6981 may afford potential mechanistic as well as functional advantages.

Ro-25-6981 exhibits high-affinity binding selective for the NR2B subunit of the NMDA receptor (Fischer et al., 1997), whereas ketamine binds noncompetitively to open channel and allosteric sites at the NMDA receptor with relatively low (>1000 nM) affinity (Orser et al., 1997). In addition, conditioned place preference and microdialysis models suggest Ro-25-6981 exhibits relatively low risk for abuse similar to other NR2B-acting agents such as ifenprodil (Boyce-Rustay and Cunningham, 2004). Indeed, emerging evidence suggests selective targeting of the NR2B subunit of the NMDA receptor affords the rapid antidepressant effects of glutamatergic antagonism with diminished potential for abuse. For example, two NR2B-selective agents, CP-101606 (Preskorn et al., 2008) and MK-0657 (Ibrahim et al., 2012), exhibit rapid antidepressant effects in humans with no apparent abuse liability. In addition, morphological neurotoxicity and altered cortical function associated with the dissociative effects of some NMDA receptor antagonists appear to be absent or reduced following Ro-25-6981 administration (Lima-Ojeda et al., 2013;Kocsis, 2012). Taken together, these data suggest the potential for a more favorable adverse effect profile of Ro-25-6981 relative to non-selective antagonists such as ketamine.

Nearly all in vivo studies to date have evaluated glutamatergic agents such as ketamine using intermittent serial infusions (Rasmussen et al., 2013;Murrough et al., 2013) or single dose administration (Katalinic et al., 2013;Caddy et al., 2014). The current study utilizes chronic administration of Ro-25-6981 or ketamine using a dosing paradigm (21 day, bid) suited for monoaminergic-related antidepressants. Reasons for the lack of sustained response to ketamine following this chronic dosing regimen are unclear, though it is possible that either receptor tolerance or dissociative behavioral effects may play a role (Morgan and Curran, 2012). Also, the activity of Ro-25-6981 on other CNS targets should be resolved, as activation of α_1A-adrenergic receptors promotes neurogenesis (Gupta et al., 2009) and antidepressant-like behaviors (Doze et al., 2009), and κ-opioid receptor antagonists exhibit antidepressant efficacy in animal models (Mague et al., 2003). Ro-25-6981 also exhibits relatively high affinity for the dopamine D4 receptor (120 nM K_i); however D4-selective agents lack antidepressant effects in vivo (Serretti et al., 1999;Basso et al., 2005). Regardless, the distinct molecular targets and binding characteristics of Ro-25-6981, particularly NR2B-selective antagonism and the inhibition of monoamine reuptake, likely underlie its functional differences from ketamine observed in vivo. For example, Zanos et al. (2016) recently reported that ketamine’s rapid antidepressant-like effects may be the direct result of rapid and sustained AMPA receptor activation by the ketamine metabolite hydroxynorketamine rather than by NMDA receptor antagonism, per se. Together, these data highlight the potential role for Ro-25-6981, as a putative dual-acting monoaminergic/glutamatergic agent, to be a lead compound in antidepressant drug discovery, and add to a growing body of evidence pointing to glutamatergic targets for novel antidepressant drug development.

Highlights:

• In silico screening identified Ro-25-6981 (coded "MI-4”) as a dopamine transporter ligand

• The NMDA receptor antagonist Ro-25-6981 also inhibits monoamine transporters (MATs)

• Ro-25-6981 exhibits rapid and sustained antidepressant-like effects in animal models

• MAT virtual screening is a viable method to identify antidepressant lead compounds

• Dual monoaminergic/glutamatergic agents are a novel strategy to treat depression

Abbreviations:

MAT

monoamine transporters

DAT

dopamine transporter

SERT

serotonin transporter

NET

norepinephrine transporter

SSRI

serotonin selective reuptake inhibitor

TCA

tricyclic antidepressant

NMDA

n-methyl-D-aspartate

TST

tail suspension test

FST

forced swim test

CPP

conditioned place preference