Introduction

Escitalopram is a widely used antidepressant prescribed to more than 240 million patients since it was approved in 2001. The primary molecular target for escitalopram is the serotonin transporter (SERT), which is responsible for serotonin (also called 5-hydroxytryptamine [5-HT]) reuptake at the terminals and cell bodies of serotonergic neurons, and is also the target for selective serotonin reuptake inhibitor (SSRI) antidepressants (Blakely et al. 1994). Serotonin is a neurotransmitter in the brain with important roles in mood homeostasis, and alterations in 5-HT levels are implicated in the pathophysiology of depression (Owens and Nemeroff 1994). Escitalopram and SSRIs inhibit the function of the SERT by binding to its orthosteric binding site (also referred to as the “primary site” in previous reports), where the substrate 5-HT binds. Inhibition of the SERT results in increased 5-HT levels in the synapses in the brain and thereby an enhanced serotonergic neurotransmission which is thought to be the basis for the antidepressant effects of these drugs. In addition to the orthosteric binding site, one or more allosteric sites may exist on the SERT (Chen et al. 2005a, b; Sanchez 2006). A compound such as escitalopram that binds to the allosteric site can modulate the properties of the orthosteric binding site, without directly affecting 5-HT uptake. In this sense, escitalopram is an allosteric serotonin reuptake inhibitor (ASRI).

Escitalopram is the S-enantiomer of racemic citalopram (Sanchez et al. 2004). Citalopram also contains an equal amount of the nontherapeutic enantiomer, R-citalopram. Escitalopram interacts with both the orthosteric and allosteric sites of the SERT, whereas R-citalopram has much weaker binding to the orthosteric site, although its affinity for the allosteric site is comparable to that of escitalopram (Chen et al. 2005a, b; Sanchez 2006). R-citalopram was originally assumed to be inactive, but it is now known to counteract the action of escitalopram without causing pharmacokinetic interactions (El Mansari et al. 2005; Mnie-Filali et al. 2007; Sanchez et al. 2004; Sanchez 2006).

Results of clinical trials and preclinical investigations indicate that escitalopram is more efficacious than citalopram and many other antidepressants (Kennedy et al. 2009; Montgomery et al. 2007; Sanchez et al. 2003, 2004; Wade et al. 2007). Escitalopram also has a fast onset of action (Kasper et al. 2006; Wade and Friis 2006). More recently, a number of reviews and meta-analyses of the clinical efficacy of escitalopram in depression have been published. Escitalopram is more effective compared to citalopram (Ali and Lam 2011; Cipriani et al. 2009; Leonard and Taylor 2010; Montgomery et al. 2011; Montgomery and Moller 2009) and to paroxetine and duloxetine (Cipriani et al. 2009; Kasper et al. 2009a; Montgomery and Moller 2009). For example, in the nonsponsored study by Cipriani et al. (2009), a multiple treatment meta-analysis of 117 randomized controlled trials (25,928 participants) from 1991 to 2007 showed that escitalopram is one of the two favored treatments in both efficacy and acceptability. When compared to serotonin and norepinephrine reuptake inhibitors (SNRIs) in patients with major depressive disorder (MDD), escitalopram is associated with a significantly lower duration of sick leave (Wade et al. 2008), as well as a better efficacy and tolerability profile (Kornstein et al. 2009; Lam et al. 2010). In addition, escitalopram has demonstrated a statistically significant dose–response relationship in severely depressed patients on multiple outcome scales (Bech et al. 2004), which is not the case for other antidepressants. Escitalopram is metabolized by multiple cytochrome P450 (CYP) enzymes and has little CYP or P-glycoprotein inhibition (Rao 2007), thus having a low potential for drug–drug interactions. Thus, from these clinical observations, escitalopram has demonstrated at least a quantitative clinical difference compared to other therapies, and this difference does not seem to be limited to certain specific subsets of patients.

The actions and mechanism of action of escitalopram have been described in previous reviews (Sanchez et al. 2004; Sanchez 2006). Since then, a large amount of preclinical and clinical research in this field has been published. This review discusses current knowledge about the mechanism of action of escitalopram. The binding interactions of escitalopram with the SERT, as well as the neurochemical and neurophysiological effects of escitalopram in the brain, are covered.

Preclinical and clinical mechanisms of escitalopram

Allosteric interactions at the SERT

Allosteric interactions at the SERT were first described in imipramine binding experiments (Wennogle and Meyerson 1982) and were subsequently reported for the dopamine and norepinephrine transporters (Plenge and Mellerup 1997). Although other compounds have also been reported to have allosteric activities for SERT (Boos et al. 2006; Nandi et al. 2004; Nightingale et al. 2005), the most thoroughly characterized is escitalopram. Much of the historical background leading to the understanding of the allosteric mechanism of escitalopram has been described (Nutt and Feetam 2010; Sanchez 2006; Zhong et al. 2009). A chronological list of key developments is shown in Table 1. As explained below, the exact loci and constituents of the allosteric site as well as its biological functions are presently under investigation.

Table 1 Progression of discoveries leading to the allosteric mechanism of escitalopram
Full size table

The effects of escitalopram and R-citalopram at the orthosteric and allosteric binding sites on the SERT have been studied in radioligand binding studies using wild-type SERT and mutated SERT, in which one or more residues have been replaced. From studies of wild-type SERT, escitalopram prolongs its own dissociation kinetics, which strongly supports the existence of an allosteric binding site (Chen et al. 2005a, b; Sanchez 2006). The orthosteric binding site is associated with amino acid residues in transmembrane (TM) domains 1 and 3 (Henry et al. 2006). More direct evidence of the existence of an allosteric binding site that is distinct from the orthosteric binding site comes from structural studies. Cross-species comparisons and mutagenesis studies have shown that the allosteric binding properties of escitalopram and R-citalopram are influenced by residues in TM domains 10, 11, and 12 of the human (h) SERT since mutations of these residues disrupted the allosteric activity of escitalopram: ALI → VFL (ALI/VFL), II → VT (II/VT), MS → SN (MS/SN), and SI → TT (SI/TT) (Neubauer et al. 2006). As an allosteric inhibitor, escitalopram elicits a more complete inhibition of 5-HT reuptake by binding to both the orthosteric and allosteric binding sites, leading to higher extracellular 5-HT levels in vivo and faster 5-HT1A autoreceptor desensitization, and hence greater efficacy and/or a faster onset of action (Sanchez et al. 2004; Sanchez 2006).

In addition, in vitro association studies demonstrated that R-citalopram delayed the association binding of escitalopram to the SERT (El Mansari et al. 2007). Furthermore, R-citalopram delayed the association binding of paroxetine but not fluoxetine to SERT in this study. Dissociation binding studies of paroxetine had shown paroxetine to have allosteric properties at the SERT, though of a lower potency than escitalopram, whereas fluoxetine in the same study was devoid of allosteric effects (Chen et al. 2005b). These findings support the hypothesis that the inhibitory effect of R-citalopram on association rates is mediated via an allosteric modulation, since a typical competitive interaction would render a concentration-dependent inhibition of the maximal binding. Using the aforementioned mutant hSERT (ALI/VFL + SI/TT), we confirmed that while its orthosteric site binding and transport function were conserved, the allosteric effect of escitalopram on dissociation kinetics was disrupted (Zhong et al. 2009). In association binding studies using the hSERT mutant (ALI/VFL + SI/TT), R-citalopram inhibited the maximal binding of [3H]escitalopram, but did not affect the association rate, indicating a loss of allosteric interactions in the mutant (Zhong et al. 2009). These data provide key evidence that the interactions between R-citalopram and escitalopram in both dissociation and association binding are mediated through the allosteric mechanism which involves the TM 10 ALI and TM 12 SI amino acid residues, identified previously (Neubauer et al. 2006).

The inhibitory effect of R-citalopram on escitalopram in the in vitro studies mentioned above has important implications in clinical settings. The effect of R-citalopram was seen below 100 nM (El Mansari et al. 2007), which is a clinically relevant concentration (Tanum et al. 2010). In humans, escitalopram and R-citalopram are metabolized by cytochrome P450 isozymes at different rates (Rochat et al. 1998), resulting in different concentrations of the two enantiomers even though citalopram tablets contain equal amounts of both enantiomers. Under steady-state conditions with 40 mg citalopram per day, the R-citalopram concentration was at least twice as high as escitalopram in human serum or cerebrospinal fluid (CSF), with serum or CSF concentrations of both compounds in the 30–100-nM range (Nikisch et al. 2004; Sidhu et al. 1997). In a later clinical study of citalopram using multiple doses under steady-state conditions, it was confirmed that R-citalopram levels were at least twice those of escitalopram at doses from 20- to 100-mg doses (Tanum et al. 2010). Thus, in a clinical setting, R-citalopram would readily exert an inhibitory effect on escitalopram.

To demonstrate this, Kasper and colleagues obtained SERT occupancy data in the midbrain of healthy subjects using a selective radioligand [123I]ADAM (2-([2-([dimethylamino]methyl)phenyl]thio)-5-[123I]-iodophenylamine) in single-photon emission computerized tomography (SPECT) studies (Kasper et al. 2009b). SERT occupancy after 10 days of escitalopram administration (10 mg/day) was 82%, significantly higher than the 64% occupancy after treatment with an equivalent dose of citalopram (20 mg/kg/day) (Kasper et al. 2009b). This shows that R-citalopram in racemic citalopram reduces the maximal binding of escitalopram to SERT (i.e., occupancy). This reduction is not due to simple competitive interaction between R-citalopram and escitalopram, since the unoccupied SERT binding sites were measured by a SPECT ligand of dissimilar structure, [123I]ADAM (Wellsow et al. 2002). Rather, at steady-state conditions, the presence of R-citalopram seems to make the orthosteric binding site of SERT less accessible to escitalopram, consistent with an allosteric mechanism by which R-citalopram decreases the association binding of escitalopram to SERT (El Mansari et al. 2007; Zhong et al. 2009).

Based on these findings, it can be seen that the allosteric mechanism for escitalopram is complex. The allosteric site mediates both the low affinity (e.g., an EC50 of 25 μM for R-citalopram in inhibiting escitalopram dissociation) and the high affinity (e.g., 40 nM is an effective concentration of R-citalopram to inhibit escitalopram association) interaction between escitalopram and R-citalopram. As illustrated in Fig. 1, multiple interactions can take place through the allosteric site of SERT: (1) Escitalopram prolongs its own dissociation from the orthosteric site; (2) escitalopram can potentially influence the modulation of SERT by its associated proteins (see below); (3) R-citalopram can inhibit the association of escitalopram to the orthosteric site; and (4) R-citalopram may interfere with the allosteric function of escitalopram and its ability to inhibit 5-HT reuptake. The fourth interaction was suggested from experiments using other SERT mutants that have an intact allosteric site but with a orthosteric site that does not bind escitalopram (Plenge et al. 2007).

Fig. 1
figure 1

Putative model of the interactions between escitalopram and the orthosteric and allosteric sites of SERT. In presynaptic terminals, the SERT transports 5-HT from the synaptic cleft back into presynaptic neurons. Escitalopram (S) binding to the allosteric site enhances its own binding at the orthosteric site (as suggested by the decreased dissociation rate from the orthosteric site), resulting in increase in extracellular 5-HT levels. This leads to enhanced serotonergic neurotransmission, which elicits effects on neuronal activities and remodeling, neuroadaptation, BDNF levels, and neurogenesis. Escitalopram binding to the allosteric site can also potentially influence the physical interaction between SERT and its interacting proteins, hence the modulation of SERT by its associated proteins. Note R-citalopram (R) can also bind to the allosteric site, probably in a different orientation, and this can interfere with the binding of escitalopram at the orthosteric site (as suggested by the lower association rate of escitalopram). S escitalopram, R R-citalopram

Full size image

As mentioned earlier, additional allosteric compounds besides escitalopram exist for the SERT. Table 2 summarizes some distinct properties of allosteric compounds among common antidepressants. Paroxetine is also allosteric, but its allosteric function is much weaker (Chen et al. 2005b; Sanchez 2006). Other antidepressants (fluoxetine, duloxetine, sertraline, and venlafaxine) do not have allosteric activities. For some of these drugs, e.g., fluoxetine and paroxetine, enantiomers have also been studied (El Mansari et al. 2007; Nutt and Feetam 2010). Interestingly, R-citalopram can inhibit the association of the allosteric paroxetine to the SERT, but not those of the non-allosteric fluoxetine, venlafaxine, or sertraline (El Mansari et al. 2007). The tricyclic antidepressant, imipramine, does not seem to be allosteric, since the only evidence is that 5-HT and R-citalopram slow its dissociation from SERT (Plenge et al. 2007; Wennogle and Meyerson 1982), likely from the orthosteric site, whereas imipramine does not slow the dissociation of [3H]-imipramine. This is different from escitalopram in that escitalopram binds to both the allosteric and orthosteric sites. For the other compounds that reportedly have allosteric activities for SERT (Boos et al. 2006; Nandi et al. 2004; Nightingale et al. 2005), their allosteric effects are visible with a different orthosteric compound, RTI-55. These compounds are of interest in understanding the allosteric function, but so far, no in vivo studies with them have been reported.

Table 2 Allosteric interactions differentiate SERT-related antidepressants
Full size table

Thus, escitalopram is the only SERT-related antidepressant that has demonstrated both allosteric and chiral advantages. This makes escitalopram unique compared to other antidepressants, with more prolonged 5-HT reuptake inhibition mediated through two interacting sites on SERT (Nutt and Feetam 2010). Some clinical studies of antidepressants have suggested that multi-target agents or combined therapies may provide therapeutic benefits by affecting more than one neurotransmitter system and thereby via different points of intervention modulate downstream events that are needed for the antidepressant activity (Millan 2006; Wong et al. 2008). One could argue that the dual mechanism of action of escitalopram at the SERT conceptually fits with this notion, as the enhanced serotonergic neurotransmission is believed to augment and expedite downstream events that are critical for the antidepressant effect. Since the profile of escitalopram was discovered by serendipity, identification of new and optimized ASRIs might be of interest as an alternative to identification of a multi-target drug. However, pursuing one or the other strategy implies significant but different challenges for medicinal chemistry design as well as for ASRI’s screening assay (Morphy and Rankovic 2009).

Escitalopram binding sites on the SERT

The SERT has 12 transmembrane domains. The orthosteric binding site for escitalopram is of high affinity in the nanomolar range. The orthosteric binding site involves Y95 and I172 residues, while R-citalopram seems to bind somewhat differently to the SERT since, unlike escitalopram, it is relatively insensitive to the Y95F mutation (Henry et al. 2006). The importance of I172 in the orthosteric site binding of escitalopram is also clearly demonstrated in the recent report of a citalopram insensitive (I172M knock-in) mouse by Thompson et al. (2011). A crystal structure of the SERT has not been reported, but progress has been made for its bacterial homologue, the leucine transporter (LeuT). Co-crystal structures of the tricyclic antidepressants clomipramine and desipramine with LeuT have provided models for the understanding of inhibitor binding to the SERT (Singh et al. 2007; Zhou et al. 2007). The mutations generated in the studies of Zhou et al. (2007) to impact antidepressant binding produced only small effects as compared to the mutations reported by Henry et al. (2006) in the orthosteric site. This is likely because the binding site reported by Zhou et al. is a potential allosteric site (see below). A homology model of the SERT was constructed using LeuT as the template, and induced-fit docking models were then made for the substrate 5-HT, as well as for both escitalopram and R-citalopram enantiomers (Celik et al. 2008; Koldso et al. 2010). These models together with biochemical validation studies demonstrate that the binding pocket for escitalopram almost completely overlaps with that of the substrate 5-HT. Other computational modeling and mutational studies confirm this by showing that a residue located within the 5-HT binding pocket, S438, is important for the binding of escitalopram (Andersen et al. 2009).

The allosteric binding site on the SERT is distinct from the high affinity orthosteric binding site, as mentioned earlier. An additional example is that a G128A mutation in the SERT disrupted the high affinity binding of escitalopram without changing the allosteric binding (Plenge and Wiborg 2005). The residues in TM 10, 11, and 12 mentioned above as important for allosteric activities of escitalopram and R-citalopram have not been constructed into a binding pocket in modeling studies. It also remains possible that these residues contribute indirectly to the allosteric binding site. Jorgensen and Topiol examined the potential migration paths of the substrate leucine during its transport process by LeuT. It was found that leucine may rely on a series of residues with electrostatic forces near the extracellular surface in order to propagate into the deeper transmembrane transport channel (Jorgensen and Topiol 2008). These residues potentially serve as “stop-over” sites for guiding the substrate into the inner pocket, and the SERT may use an analogous mechanism for 5-HT transport. Zhou et al. (2009) determined the crystal structures for LeuT in a complex with the SSRIs sertraline or fluoxetine. Based on this study, it is suggested that a binding pocket exists between the tip of the extracellular loop EL4 and the extracellular gate for SSRIs (Zhou et al. 2009). This pocket is termed the “vestibule” based on its shape and outer location and coincides with one of the “stop-over” sites proposed by Jorgensen and Topiol. A recent report by Loland and colleagues suggests that the allosteric binding mechanism for escitalopram may be associated with the “vestibule” region (Loland et al. 2010). Despite this interesting progress, caution must be taken in evaluating the binding site visualized using a bacterial leucine transporter in the structural studies. The elucidation of the exact allosteric binding pocket for escitalopram awaits definitive structural studies using the serotonin transporter itself.

Neurophysiological and neurochemical effects of escitalopram

In this section, updated reports of the neurophysiological and neurochemical effects of escitalopram in vivo are summarized. Findings for escitalopram in preclinical and clinical studies in comparison to other antidepressants are also discussed. It is worth noting that although the allosteric interaction by escitalopram has been extensively characterized based on in vitro studies as mentioned earlier, much remains to be learned with respect to this interaction under in vivo conditions. Future or current studies needed to understand the physiological functions of allosterism will be described.

Neuronal activities and neuroadaptation

Antidepressants including SSRIs take several weeks to produce their therapeutic effects, and this delayed onset is likely due to neuroadaptive changes in the brain in addition to the elevation of 5-HT levels that are required for the antidepressant effects (Blier and de Montigny 1999). The effect of SERT inhibition can be studied by in vivo electrophysiological recording of dorsal raphe 5-HT neurons. Acute administration of escitalopram inhibited the spontaneous firing activity of raphe neurons, but the potency of racemic citalopram was 4-fold weaker (Sanchez 2006). The recovery of raphe 5-HT neuronal firing after the desensitization of 5-HT1A autoreceptors is thought to reflect the neuroadaptive process underlying the onset of antidepressant action (Blier and de 1999; El Mansari et al. 2005). For escitalopram, it took 2 weeks before 5-HT neuron firing returned to control levels, but for citalopram, it took at least 3 weeks, suggesting a delayed onset of action of S-citalopram in the presence of R-citalopram (El Mansari et al. 2005). Similar effects were seen in a later study, in which the firing rate after a 2-week treatment was comparable to baseline with escitalopram, but inhibition of firing rate was still significant at the end of treatment in a parallel group treated with both R-citalopram and escitalopram (Mnie-Filali et al. 2007). As mentioned in the previous section, the different effects of escitalopram and citalopram are likely due to the inhibitory effects of R-citalopram through an allosteric mechanism at the SERT. The target mediating the effects of escitalopram and citalopram is clearly the SERT, as demonstrated recently by Thompson et al. (2011). In this report, transgenic mice that bear the I172M mutation, which abolishes citalopram binding without impacting the recognition of 5-HT, possess normal basal 5-HT levels and transport rates but lack the response to citalopram in 5-HT increase and raphe neuron firing modulation.

In addition to the above studies on the firing of 5-HT neurons, the effects of escitalopram and citalopram on dopaminergic and glutamatergic neurons have recently been reported (Schilstrom et al. 2011). In this study, Schilstrom and colleagues tested the acute effects of compounds on the firing of dopamine neurons in the ventral tegmental area (VTA) of anesthetized rats. Escitalopram (40–640 μg/kg i.v.) increased both firing rate and burst firing of dopamine neurons, while citalopram dosed at equivalent S-citalopram levels (80–1,280 μg/kg i.v.) had only a minimal effect on burst firing. R-citalopram (40–640 μg/kg) did not have an effect by itself, but 320 μg/kg of R-citalopram completely blocked the effect of an equal dose of escitalopram (320 μg/kg) (Schilstrom et al. 2011). These results, however, are different from previous reports in which escitalopram and SSRIs such as fluoxetine, citalopram, paroxetine, sertraline, and fluvoxamine seem to inhibit the firing of dopamine neurons in the VTA (Di Mascio et al. 1998; Dremencov et al. 2009; Prisco and Esposito 1995). Thus, this research area warrants additional study, especially those considering different species, methodologies, and dosing regimes.

Schilstrom and colleagues also tested the effect of escitalopram on N-methyl-d-aspartate (NMDA)-mediated currents in pyramidal neurons in medial prefrontal cortical slices (Schilstrom et al. 2011). Escitalopram (5 or 100 nM) potentiated NMDA-induced currents by ~40%. However, neither citalopram (10 or 200 nM) nor R-citalopram (5 or 100 nM) has a significant effect at matching concentrations. The potentiation of NMDA currents by escitalopram is also in line with a recent study on long-term potentiation (LTP), in which hippocampal LTP deficits induced by neonatal clomipramine manipulation were restored after 2 weeks of escitalopram treatment (Bhagya et al. 2011). Additional studies to confirm these results will be valuable, since a possible excitatory effect of escitalopram on dopaminergic and NMDA receptor-mediated neurotransmission may have bearing on cognition (Schilstrom et al. 2011). Further investigations into the contribution of NR2A versus NR2B subtype of NMDA receptors in these effects may also be important, given NR2B is particularly indicated in potential novel treatment strategies for depression (Skolnick et al. 2009).

Neuronal activity and firing studies cannot be readily measured in a clinical setting. However, electroencephalography (EEG) may be able to help identify overall neuroadaptive changes in the human brain. Leuchter et al. reported the use of frontal quantitative EEG (QEEG) to predict escitalopram treatment outcome with good accuracy (Leuchter et al. 2009a, b). For 375 patients with MDD, QEEG measurements were made from four recording sites on the forehead and two on the earlobes before and after 1 week of treatment with escitalopram (10 mg). Using a special algorithm to compare several EEG features before and after 1 week of treatment, the authors ranked the QEEG changes from the frontal lobe for each subject and identified a threshold. Following the initial week of dosing with escitalopram, the patients were randomized to three groups to continue treatment for a total of 49 days: escitalopram (10 mg), bupropion (300 mg), or an escitalopram (10 mg) + bupropion (300 mg) combination. It was found that patients with a QEEG change (at the end of the initial week of treatment with escitalopram) above the threshold were 2.4-fold more likely to show response to continued treatment with escitalopram (Leuchter et al. 2009a, b). In addition, patients with a QEEG change (at the end of the initial week of treatment with escitalopram) below the threshold were almost twice as likely to respond to bupropion instead. Thus, this QEEG method may help guide treatment management as well as the development of additional translational strategies, for example, for use in conjunction with positron emission tomography (PET) and functional magnetic resonance imaging.

Brain-derived neurotrophic factor (BDNF)

Jacobsen and Mork (2004) reported that chronic treatment with escitalopram decreased BDNF levels in the frontal cortex and hippocampus in the rat brain. In comparison, chronic electroconvulsive seizures and lithium administration increased BDNF levels in both brain regions, while desipramine did not affect BDNF levels in either region (Jacobsen and Mork 2004). In a rat chronic stress model with a resident–intruder paradigm, BDNF levels doubled in the cerebral cortex after 5 weeks of stress, and this was prevented by escitalopram administration during the last 4 weeks (Schulte-Herbruggen et al. 2009). More recently, it was reported that 1 week of escitalopram increased BDNF messenger RNA (mRNA) and pro-BDNF protein levels in prefrontal cortical regions in normal rats, while a 3-week escitalopram administration decreased BDNF mRNA levels as well as CREB/BDNF signaling in the hippocampus (Alboni et al. 2010).

Different effects on BDNF levels, however, have been reported with other antidepressants. Sertraline increased brain BDNF levels in a R6/2 mouse Huntington’s disease model that possessed a deficit in BDNF (Peng et al. 2008). In a rat model of chronic unpredictable mild stress, BDNF levels were decreased in the hippocampus, and this was reversed by chronic treatments with antidepressants, including venlafaxine, mirtazapine, and fluoxetine (Zhang et al. 2010). In normal male Sprague–Dawley rats, chronic treatments with fluoxetine, desipramine, or phenelzine increased BDNF levels in the frontal cortex, but not in the hippocampus (Balu et al. 2008). The difference between escitalopram and other antidepressants in the regulation of BDNF in the brain is interesting, although contributing factors may also include confounders such as different animal models or methodology. In depressed patients, abnormal blood BDNF levels have been reported, and the levels are normalized after escitalopram treatment (Aydemir et al. 2006; Serra-Millas et al. 2011). Continued research using animal models will help understand the clinical role and translational potential of BDNF in depression.

Neurogenesis

Adult neurogenesis has been observed during antidepressant treatment in preclinical models and has been considered to be an important feature in the characterization of novel antidepressants (Banasr and Duman 2007). In addition to antidepressants, it is now known that many compounds can induce neurogenesis, and its functional impact and contribution to the etiology of depression and the treatment mechanism remains unclear (Lucassen et al. 2010). Still, it is believed that an increase or a normalization of neuronal proliferation is at least partly involved in antidepressant action (Lucassen et al. 2010).

Many preclinical studies have reported a positive effect of escitalopram on neurogenesis. In the dorsal hippocampus of normal rats, neurogenesis was significantly increased after a 2-week chronic treatment regime with escitalopram and this effect was correlated with the firing recovery of raphe neurons (Mnie-Filali et al. 2007). In a rat chronic mild stress (CMS) model, CMS significantly decreased the hedonic state measured as sucrose consumption (anhedonia) and neurogenesis in the ventral hippocampus (Jayatissa et al. 2006). Chronic treatment with escitalopram for 4 weeks reversed the CMS-induced decrease in sucrose consumption and the decrease of neurogenesis in the ventral hippocampus. In addition, there was a correlation between recovery from anhedonia and the increase in hippocampal neurogenesis (Jayatissa et al. 2006).

Neuronal remodeling and synaptic protein markers

Molecules mediating synaptic transmission and plasticity are important in the hypotheses of the pathophysiology and treatment mechanisms of depression (Racagni and Popoli 2008). In the inbred rat strain Flinders Sensitive Line (FSL), there is an increased vulnerability to environmental factors inducing depressive-like behavior compared with the control Flinders Resistant Line (FRL); thus, this model has some predictive validity for antidepressant effect (Ryan et al. 2009). In FSL rats, synaptic signaling proteins synapsin I and synaptotagmin-1 in the hippocampus are either over-activated (as measured by synapsin I phosphorylation) or over-expressed (Musazzi et al. 2010). After chronic (30 days) treatment with escitalopram, these two alterations in FSL were restored to the levels found in FRL control rats (Musazzi et al. 2010). This effect of escitalopram seems to be consistent with other antidepressants reported in previous studies. After chronic (14 days) treatments with either fluoxetine or reboxetine, there were also dramatic decreases in the levels of total and phosphorylated synapsin I protein in synaptosomes and synaptic membranes (Barbiero et al. 2007), suggesting a converging mechanism for traditional antidepressant actions. A single dose of ketamine, an anesthetic with a very fast onset (in hours) and strong antidepressant efficacy (Berman et al. 2000; Zarate et al. 2006), can induce sustained increases in synaptic synapsin I, GluR1, and PSD95 in the rat prefrontal cortex, probably mediated through AMPA and mTOR signaling (Li et al. 2010). Therefore, further studies of neuroplasticity (such as additional synaptic proteins) will be crucial for understanding the mechanism of action of escitalopram.

The 5-HT1A receptor is an important autoreceptor in the brain that regulates the firing of 5-HT neurons, as mentioned above. In depressed patients, there are reduced levels of 5-HT1A receptors in several brain regions, including the mesiotemporal cortex and raphe (Drevets et al. 2007). Treatment-induced changes of 5-HT1A receptor binding were recently examined using [carbonyl-11C]WAY-100635 PET (Hahn et al. 2010). In patients with anxiety disorders, the 5-HT1A binding potential for dorsal raphe autoreceptors showed a disorganized correlation pattern without area-specific characteristics. After escitalopram treatment for 3 months, the correlation between the 5-HT1A binding potential of dorsal raphe autoreceptors and that of the heteroreceptors within the amygdala and the hippocampus was dramatically enhanced (Hahn et al. 2010). Additional studies comparing the effect of escitalopram with other SSRIs would be interesting. Together with preclinical studies, such as those on neuronal firing mentioned above, these clinical data provide valuable insights into the neuroadaptive processes induced by antidepressant treatment.

Genetic variations of SERT affecting responses to escitalopram

Mutations and polymorphisms of the SERT in human populations have been identified and linked to diseases. From screens for human sequence variations, a rare I425V mutation was identified (Glatt et al. 2001) and was later found to have a gain-of-function phenotype (Kilic et al. 2003; Prasad et al. 2005) as well as an association with obsessive-compulsive disorder (Ozaki et al. 2003). A slightly different mutation of the I425 residue, I425L, together with other naturally occurring variations L550V and G56A, may be linked to autism and rigid-compulsive behaviors (Sutcliffe et al. 2005). Other naturally occurring human SERT variants, such as T4A and G56A, are found to be linked to disrupted SERT regulatory mechanisms. Once these naturally occurring mutations have been identified, it is important to know if they affect the activities of antidepressants. For example, citalopram and fluoxetine have a significant loss of activity for the P339L variant of human SERT (Prasad et al. 2005). Thus, this kind of genetic information from depressed patients will be valuable for evaluating the dosage of and treatment responses to escitalopram.

Another well-studied example is the polymorphism in the length of the promoter region of the human SERT gene, 5-HTTLPR. The long (L) allele of 5-HTTLPR has a 44 base pair insertion of tandem repeat elements in the promoter region. The short (S) allele lacks this 44 base pair insertion and has a 3-fold lower 5-HT uptake function in lymphoblasts compared to the L allele, due to reduced transcriptional efficiency (Lesch et al. 1996). The LL genotype is associated with earlier age at diagnosis among patients with familial pulmonary arterial hypertension (Willers et al. 2006). On the other hand, studies have been reported that the S allele is correlated with higher risk of depression (Lotrich and Pollock 2004), alcohol dependence (McHugh et al. 2010), and their comorbidity (Nellissery et al. 2003). Recently, a study of the treatment outcomes for patients with comorbid depression and alcohol dependence indicated that the LL genotype predicted greater antidepressant responses to escitalopram (3 months of treatment) as measured by the improvement in Montgomery-Åsberg Depression Rating Scale scores (Muhonen et al. 2011).

With an increasing use of genetic characterization of susceptible genes in human disease diagnoses during recent years, it may be possible in the future to find patients bearing mutations that resemble those with impaired orthosteric or allosteric binding sites for escitalopram. Studies of those patients will shed further light on the mechanism of action for escitalopram. To better understand the possible impact of these mutations as well as the physiological functions of the orthosteric or allosteric binding sites, genetically manipulated mouse models can be created. As to the allosteric site for escitalopram, it will be interesting to see the neurochemical and behavioral characteristics in animals that have this binding site disrupted. In this regard, work is ongoing in establishing knock-in mouse models bearing the aforementioned allosteric mutations ALI → VFL (ALI/VFL), II → VT (II/VT), MS → SN (MS/SN), and SI → TT (SI/TT) (Owens 2007; Marc Caron, personal communication). This forthcoming knock-in model can be studied in conjunction with the mouse model lacking the orthosteric binding site for citalopram by Thompson et al. (2011) in order to understand the physiological functions of the orhosteric versus allosteric interactions.

Potential effects mediated through the SERT but independent of 5-HT uptake

The effects of escitalopram on BDNF, neuronal activities and remodeling, neuroadaptation, and neurogenesis summarized above are mediated directly through the uptake function of SERT. All these effects require elevated 5-HT levels as a result of uptake inhibition and are thus 5-HT uptake dependent. As illustrated in Fig. 1, binding by escitalopram to the orthosteric and allosteric site on the SERT results in increased extracellular 5-HT levels and, in turn, enhanced 5-HT transmission. On the other hand, some effects of escitalopram can be potentially mediated by the SERT regardless of its uptake function. The SERT is dynamically regulated in vivo and can interact with other proteins (Blakely et al. 1998; Steiner et al. 2008; Zahniser and Doolen 2001). These SERT-interacting proteins (SIPs) may be potentially modulated as a result of allosteric interactions between escitalopram and the SERT, leading to indirect effects that may be independent of 5-HT uptake, as depicted in Fig. 1. For instance, protein kinase C (PKC) plays a key role in the regulation of the cell surface expression and trafficking of the SERT (Blakely et al. 1998). In a recent study on 5-HT neuronal firing by Mnie-Filali et al. (2009), the inhibitory effect of R-citalopram on escitalopram was abolished by pretreatment with the PKC inhibitor staurosporine. This suggests that, in addition to 5-HT output, surface expression and trafficking of the SERT may be affected by the allosteric interactions of R-citalopram and escitalopram.

Another example of SIPs is the neuronal nitric oxide synthase (nNOS), which is widely expressed in the brain with functional roles in learning, memory, and neurogenesis and has been associated with various CNS disorders such as depression, Parkinson’s disease, and Alzheimer’s disease (Zhou and Zhu 2009). In HEK293 cells, co-expression of nNOS with the SERT decreased both SERT cell surface density and 5-HT uptake (Chanrion et al. 2007), suggesting a tonic inhibition of SERT activity by nNOS under physiological condition in vivo. Conversely, 5-HT was able to activate nNOS, resulting in increased cGMP production. The effect of 5-HT was mediated through the uptake activity of the SERT, since the inhibitors citalopram and paroxetine prevented the nNOS-activating effect of 5-HT (Chanrion et al. 2007). Thus, escitalopram may be potentially able to modulate the reciprocal interaction between the SERT and nNOS, either through allosteric interaction (on the physical interaction between SERT and nNOS, i.e., independent of 5-HT uptake) or directly through 5-HT increase.

Potential mechanisms independent of the SERT

Zhang and Rasenick (2010) recently reported that escitalopram and fluoxetine were able to translocate the G protein alpha subunit Gs out of lipid rafts in C6 glioma cells. When residing outside the lipid raft compartment, Gs alpha proteins are more accessible for adenylyl cyclase stimulation, and hence, cells have an increased responsiveness to stimuli. This effect is likely mediated through a novel target since the SERT is not expressed in C6 cells. R-citalopram had no effect on the localization of Gs alpha, nor was it able to reduce the effectiveness of escitalopram, further suggesting that the SERT is not involved. Interestingly, escitalopram seemed to exert this effect with a fastest onset (1 day of treatment) than other SSRIs, which required at least 3–5 days of treatment (Rasenick, personal communication). These data are particularly intriguing, since in prefrontal cortices and cerebella of suicidal patients with confirmed unipolar depression, the level of Gs alpha sequestered in lipid rafts in prefrontal cortices and cerebella of suicidal patients with confirmed unipolar depression was twice that found in nondepressed controls (Donati et al. 2008). Thus, lipid raft dynamics with their influence on Gs alpha functionality may be a novel antidepressant mechanism.

Escitalopram at 1,000 nM did not have a significant affinity for the 144 targets screened, except for the sigma1 receptor (Sanchez et al. 2003). As mentioned earlier, escitalopram has serum or CSF concentrations well below the 1,000-nM range at clinical doses (Nikisch et al. 2004; Sidhu et al. 1997). Escitalopram is also predicted to be able to occupy approximately 80% of the SERT population in patients at corresponding steady-state plasma concentrations of 6–21 ng/ml (15–53 nM) (Kreilgaard et al. 2008). Thus, at clinically relevant concentrations of escitalopram, the targets mediating potential SERT-independent mechanisms remain elusive.

Conclusions

Escitalopram is an ASRI antidepressant. Data from numerous clinical trials and meta-analyses indicate that escitalopram is more effective and has a faster onset compared to other therapeutic agents, including citalopram and SSRIs, in treating major depression. Interactions at both the allosteric and the orthosteric binding sites in the SERT and demonstrated tolerability likely contribute to the superior clinical profile of escitalopram. Through the allosteric binding site, escitalopram can delay its own dissociation from the SERT, while R-citalopram can inhibit the association of escitalopram to the SERT at therapeutically relevant levels. The effectiveness of escitalopram, as well as the counter-productive effects of R-citalopram, has been demonstrated in a large number of preclinical and some clinical studies. The cellular, neurochemical, neuroadaptive, and neuroplastic changes induced by escitalopram after both acute and chronic administration have provided a mechanistic basis to account for the clinical advantages of escitalopram over other antidepressant therapies.