Antidepressant effects of ketamine on depression-related phenotypes and dopamine dysfunction in rodent models of stress

Abstract

Depression, the most prevalent psychiatric disorder, is characterized by increased negative affect (i.e. depressed mood) and reduced positive affect (i.e. anhedonia). Stress is a risk factor for depression in humans, and animal models of chronic stress are typically used to study neurobehavioral alterations relevant to depression. Common behavioral outcomes in rodent models of chronic stress include anhedonia, social dysfunction and behavioral despair. For example, chronically stressed rodents exhibit reduced reward preference, as measured by a loss of preference for sucrose solutions and time spent interacting with a novel conspecific, while also exhibiting less time struggling against inescapable stressors (e.g. forced swim, tail suspension). In both humans and rodents, anhedonia is associated with dysfunction of the dopamine (DA) system. Unlike traditional antidepressants, which are limited by inadequate efficacy and delayed therapeutic response, acute ketamine administration rapidly alleviates depressive symptoms in humans and reverses stress-induced changes in animal models. These effects are partially mediated via actions on the DA system. This review summarizes the clinical effects of ketamine, the neurobiological underpinnings of depression with a focus on DA dysfunction, as well as antidepressant effects of ketamine on depression-related endophenotypes (i.e. anhedonia, despair) and ventral tegmental area (VTA) activity in rodent models of repeated stress. Moreover, we discuss evidence regarding sex differences in ketamine’s antidepressant effects, wherein females appear to be more sensitive to lower dose ketamine, as well as novel findings suggesting that ketamine has prophylactic effects with regard to protection against the neurobehavioral impact of future stressors.

Introduction

Major depressive disorder (MDD), also referred to as depression in this review, is the most prevalent psychiatric disorder and one of leading causes of disability worldwide (Collins et al., 2011; Fava and Kendler, 2000; Otte et al., 2016; WHO, 2017). In the United States, the lifetime prevalence of depression is estimated to be ~17–20% (Kessler et al., 2005). Although depression is characterized by changes in mood (i.e. increase in negative affect, decrease in positive affect) (Gross and Munoz, 1995; Hofmann et al., 2012; Nutt et al., 2007), it is a symptomatically heterogeneous disease spanning cognitive, emotional, motivational and physiological domains and, as such, it is particularly difficult to treat (Otte et al., 2016). First line pharmacotherapies for depression, such as serotonin reuptake inhibitors (SSRIs), target monoaminergic systems but are limited by low response rate (e.g. ~40– 60% of patients fail to achieve remission) and slow onset of therapeutic effects in the patients that do respond, in that weeks or even months of treatment are required (Akil et al., 2018; Gaynes et al., 2009; Katz et al., 2004). This delay is particularly concerning given the elevated suicide risk in depressed individuals (Henriksson et al., 1993), and underscores a need for fast-acting antidepressants.

In contrast with the delayed therapeutic effects of existing antidepressants, ketamine rapidly improves depressive symptoms (Berman et al., 2000) and suicidal ideation in depressed individuals (DiazGranados et al., 2010; Price et al., 2009). It is particularly effective in those exhibiting treatment resistant depression (TRD)(Covvey et al., 2012; Zarate et al., 2006), which refers to an inadequate response to at least two different antidepressants administered at adequate doses and duration (Akil et al., 2018). Indeed, clinical studies have shown that a single low-dose ketamine infusion produces rapid antidepressant effects that last up to 7 days (Berman et al., 2000; Zarate et al., 2006), and these findings have been replicated many times (Coyle and Laws, 2015). Importantly, similar antidepressant effects have been observed in animal models of chronic stress, which are used to model depression-like physiological and behavioral changes in rodents and include changes in dopaminergic pathways within the brain’s reward system (Belujon and Grace, 2017; Slattery and Cryan, 2017). The ability of ketamine to reverse the effects of stress exposure, and even produce prophylactic effects in certain stress models, provides reverse-translational support for initial clinical results (Hare et al., 2017).

In this review, we provide an overview of mechanisms underlying ketamine’s antidepressant actions in humans and animal models of chronic or repeated stress. We focus on animal models of stress due to the known link between stress exposure and depression in the clinical literature (Hammen, 2005; Kendler et al., 1999; Lloyd, 1980), but also because repeated stress has significant adverse consequences on the mesolimbic dopamine (DA) system (Pani et al., 2000), which originates in the ventral tegmental area (VTA) and projects to the prefrontal cortex (PFC) and nucleus accumbens (NAc), among other areas (Ikemoto, 2007). Indeed, stress exposure increases glucocorticoid signaling via elevated corticosterone (CORT) secretion (Holsboer, 2001) to precipitate neuronal degeneration and dysfunction in the hippocampus (Conrad, 2008; Sapolsky, 2000) and medial PFC (Arnsten, 2009; McEwen and Morrison, 2013). These effects have direct influence on DA function, as glucocorticoids modulate firing of DA neurons (Belujon and Grace, 2015), and stress-sensitive regions that suffer glucocorticoid-mediated atrophy (i.e. hippocampus, PFC) (McEwen, 2004) are key regulators of mesolimbic and mesocortical dopaminergic pathways (Belujon and Grace, 2015; Grace et al., 2007). Thus, we review findings indicating that 1) MDD and chronic stress models are associated with compromised DA function, and 2) manipulations of the DA system, particularly VTA DA neuron activity, contribute to the antidepressant actions of ketamine.

Ketamine as a fast-acting antidepressant

Ketamine is a non-competitive antagonist acting at the NMDA glutamate receptor. Although ketamine has been used extensively in anesthesia and pain management (Kurdi et al., 2014), it shows great promise for its antidepressant effects (DeWilde et al., 2015). During the past 20 years, numerous clinical trials have provided strong evidence that a single sub-anesthetic dose of ketamine rapidly and robustly alleviates depressive symptoms in MDD patients (Berman et al., 2000; Coyle and Laws, 2015; Kraus et al., 2017; Murrough et al., 2013a; Phillips et al., 2019; Zarate et al., 2006). Unlike classic antidepressants whose therapeutic effects take weeks to be observed, an acute intravenous (i.v.) injection of ketamine is sufficient to induce quick and long-lasting antidepressant effects. The first clinical study to assess the antidepressant effects of ketamine reported that a low-dose ketamine infusion (0.5mg/kg, i.v.) significantly reduced depressive symptoms in MDD patients within 3 days compared with those that received saline placebo (Berman et al., 2000). A subsequent trial demonstrated that a single low-dose ketamine infusion (0.5mg/kg, i.v.) produced rapid and robust antidepressant effects in treatment-resistant MDD patients, with onset of effects occurring within 2 hours and lasting up to 1 week in a subset of patients (Zarate et al., 2006). Ketamine’s fast-acting effects are ideal for providing relief for MDD patients at risk for suicide, as ketamine infusions quickly decrease suicidal ideation in MDD patients within 24 hours (DiazGranados et al., 2010; Price et al., 2009). Moreover, repeated injections have been shown to induce sustained antidepressant effects with mild side effects in treatment-resistant MDD patients (aan het Rot et al., 2010; Murrough et al., 2013b; Phillips et al., 2019; Zheng et al., 2019).

The rapid and robust antidepressant-like effects of ketamine have been replicated in preclinical studies using rodent models relevant to depression, which have enabled researchers to unravel ketamine’s mechanisms of action (Browne and Lucki, 2013; Scheuing et al., 2015). In a pivotal study, researchers demonstrated that low dose ketamine injection [10mg/kg, intraperitoneal (i.p.)] rapidly and transiently activates the mammalian target of rapamycin (mTOR) signaling pathway, leading to increased synaptic signaling proteins (e.g. PSD-95, GluR1, and synapsin) and increased number and function of new spine synapses in the medial prefrontal cortex (mPFC) of male rats (Li et al., 2010). Moreover, blockade of mTOR signaling completely prevented ketamine-induced synaptogenesis and behavioral responses in animal models and assays of depressive-like behavior [e.g. forced swim test (FST), learned helplessness (LH)], indicating that the antidepressant effects on synapses and behavior require mTOR signaling (Li et al., 2010). These findings suggested that the effects of ketamine were opposite to the synaptic deficits resulting from stress exposure and could contribute to the antidepressant actions of ketamine. Indeed, a later study found that rats exposed to 21 days of chronic unpredictable stress exhibited anhedonia (i.e. reduced sucrose preference) and increased anxiety-like behavior (i.e. latency to feed) in the novelty suppressed feeding test, which was associated with decreased expression of synaptic proteins (e.g. PSD-95, GluR1, and synapsin) and reduced spine number within the mPFC; all of which were reversed by ketamine treatment in an mTOR-dependent manner (Li et al., 2011). Furthermore, a causal role for spine formation within the mPFC has recently been demonstrated for the sustained, but not acute, antidepressant effects of ketamine following chronic stress (Moda-Sava et al., 2019). Using two-photon imaging, researchers demonstrated that acute ketamine treatment (10mg/kg, i.p.) reversed branch-specific elimination of dendritic spines via restoration of lost spines in the PFC in stressed animals.

Besides mTOR activation and sustained mPFC spine formation, another mechanism for ketamine’s antidepressant effects involves the production and release of brain derived neurotrophic factor (BDNF), which is essential for neuron development, survival and synaptic plasticity, and has been implicated in depression and antidepressant action (Autry and Monteggia, 2012; Duman and Monteggia, 2006). Specifically, ketamine-mediated NMDA receptor blockade deactivates eukaryotic elongation factor 2 (eEF2) kinase (also known as CaMKIII), resulting in reduced eEF2 phosphorylation, de-suppression of translation of BDNF and enhanced BDNF expression within the hippocampus (Autry et al., 2011; Monteggia et al., 2013). Importantly, increased BDNF production and release is required for ketamine’s antidepressant action, as demonstrated by the lack of antidepressant-like response to ketamine in BDNF knockout mice (Autry et al., 2011). Additionally, a subsequent study has proposed that ketamine can also increase BDNF signaling via post-synaptic AMPA receptor stimulation, resulting in cell depolarization and activation of L-type voltage-dependent calcium channels, thereby allowing for calcium influx and resultant activity-dependent exocytosis of BDNF (Lepack et al., 2014).

Another mechanism involves a protein called glycogen synthase kinase-3 (GSK3), a master switch serine-threonine kinase implicated in MDD (Beurel et al., 2015; Li and Jope, 2010), the inhibition of which is necessary for ketamine’s therapeutic effects (Beurel et al., 2011). Administration of low dose ketamine (10mg/kg; i.p.) quickly inhibits GSK3 within the hippocampus and PFC of mice, and inhibition of GSK3 was shown to be necessary for the rapid antidepressant-like effect of ketamine in mice following learned helplessness (Beurel et al., 2011). Mice expressing constitutively active GSK3, meaning GSK3 knock-in mice in which GSK3 could not be inhibited, were resistant to the antidepressant like effect of ketamine. Furthermore, administering an acute high dose of a GSK3 inhibitor (i.e. lithium) produced antidepressant effects comparable to ketamine (Beurel et al., 2011). These data demonstrated that the rapid antidepressant effect of ketamine in helpless rats requires inhibition of GSK3 and can be mimicked by lithium administration.

In sum, the cellular mechanisms underlying the rapid antidepressant-like effects of ketamine administration include: i) activation of mTOR in the mPFC ii) increased BDNF production and release, iii) inhibition of GSK3, among others (Hare et al., 2017; Scheuing et al., 2015). However, systems-level mechanisms appear more complex. Over the past 5 years, the antidepressant effects of ketamine have been linked to adaptations within the brain’s reward system (i.e. mesolimbic DA system) in humans, non-human primates and rodents, and these dopaminergic effects are thought to contribute to its antidepressant effects (Kokkinou et al., 2018). Indeed, acute ketamine administration leads to DA release in the brain and increases DA levels in the striatum, nucleus accumbens and the prefrontal cortex and produces approximately 62–180% increases in DA neuron activity in chronically stressed rodents (Kokkinou et al., 2018). Thus, in this review we focus on stress and ketamine effects on the mesolimbic DA system, particularly the ventral tegmental area (VTA), in animal models of depression-related symptomatology.

Dopamine deficits in in depression: Evidence from clinical and preclinical studies

Humans.

Deficits in reward and motivation are common symptoms present in several psychiatric disorders, including depression (Admon and Pizzagalli, 2015; Der-Avakian et al., 2016; Lambert et al., 2018). Anhedonia, diminished pleasure and/or decreased reactivity to normally rewarding (i.e. pleasurable) stimuli, is one of the core symptoms in depression (Klein, 1974; Lewinsohn and Graf, 1973; Treadway and Zald, 2011). Importantly, anhedonia encompasses deficits in other complex reward and motivational processes and may include impairments in anticipatory pleasure, reward valuation, reward learning, as well as motivation and effort expenditure (i.e. willingness to work for rewards) (Der-Avakian et al., 2016; Treadway et al., 2012; Treadway and Zald, 2011). In humans, depression has been associated with abnormally reduced positive affect, reward processing, approach behavior and effort-based motivation (Admon and Pizzagalli, 2015; Pizzagalli et al., 2008; Treadway et al., 2012). For example, in the effort expenditure for rewards task (EEfRT), which involves offering subjects a series of trials where they may choose to expend more or less effort (e.g. number of button presses) for the opportunity to win varying amounts of monetary rewards, MDD patients were less willing to expend effort for rewards than controls (i.e. reduced effort expenditure), and were also less able to effectively use information about magnitude and probability of reward to guide their choice behavior (i.e. impaired reward-based decision-making) (Yang et al., 2014).

Although anhedonia is comprised of distinct processes involving different neural circuits, most involve dysfunction within the brain’s reward system (Berridge and Kringelbach, 2015; Der-Avakian and Markou, 2012). Indeed, altered reward function is theorized to underlie the loss of pleasure and amotivational syndrome experienced by many MDD patients (Alloy et al., 2016; Martin-Soelch, 2009), and human functional imaging and postmortem tissue studies have identified abnormalities in several brain regions, including nuclei within the brain’s reward pathway (Drevets, 2001; Manji et al., 2001; Price and Drevets, 2012; Whitton et al., 2015), which have been recapitulated in rodents (Belujon and Grace, 2017; Fox and Lobo, 2019; Russo and Nestler, 2013). Thus, both anhedonia and depression are linked to dysfunctions in the brain’s reward system and, in particular, the mesolimbic DA system (Belujon and Grace, 2017; Der-Avakian and Markou, 2012; Heshmati and Russo, 2015; Kapur and Mann, 1992; Naranjo et al., 2001; Yadid and Friedman, 2008). Initial data supporting a role for central DA dysfunction in depression came from studies of DA turnover, in which it was observed that MDD patients have decreased cerebrospinal fluid levels of homovanillic acid (HVA)- the primary metabolite of DA- thereby suggesting lowered basal dopaminergic tone in depression (Berger et al., 1980; Lambert et al., 2000; Willner, 1983). Moreover, in depressed patients with anhedonia, PET imaging studies have shown significantly lower DA transporter (DAT) binding compared with healthy subjects, suggesting a downregulation secondary to lower DA concentrations (Meyer et al., 2001; Sarchiapone et al., 2006). Additionally, pharmacological interventions that block or deplete DA can induce or deepen depressive symptoms in currently depressed or remitted individuals (Hasler et al., 2008; Ruhe et al., 2007) and drugs that enhance signaling can have antidepressant effects in depressed patients (Argyropoulos and Nutt, 2013; Treadway and Zald, 2011), further implicating DA dysfunction in MDD. Furthermore, functional neuroimaging studies have documented reduced ventral striatal activation in MDD patients to reward anticipation cues (Forbes et al., 2009; Smoski et al., 2009), reward receipt (McCabe et al., 2009; Pizzagalli et al., 2009), and reward prediction errors (i.e., the difference between experienced versus predicted rewards) (Kumar et al., 2008; Steele et al., 2007; Ubl et al., 2015). Taken together, these data indicate blunted reward-related signaling in MDD patients. Given that reward-related DA signaling is necessary for the attribution of incentive salience to motivational stimuli (Berridge and Robinson, 1998), downregulation of the reward system is thought to drive decreased motivation and goal-related cognitions, increased withdrawal, as well as emotions such as sadness or anhedonia (Alloy et al., 2016; Argyropoulos and Nutt, 2013), which is consistent with disruption of motivation to seek out pleasurable experiences described in MDD patients (Sherdell et al., 2012). This is significant because anhedonia is thought to contribute to the persistence of treatment-resistant depression, since the presence of anhedonic symptoms (i.e. blunted reward learning) is generally a predictor of poor antidepressant treatment response (Spijker et al., 2001; Vrieze et al., 2013), but is also a particularly difficult symptom to treat, as many first line pharmacotherapies (e.g. SSRIs) do not adequately address motivational and reward processing deficits in depression (Treadway and Zald, 2011). In fact, a study conducted in adults with remitted MDD found that they exhibited blunted reward responsiveness, indexed by an inability to modulate behavior as a function of reinforcement history compared with controls, suggesting that alterations of reward function may persist after remission and represent a trait-related abnormality in MDD (Pechtel et al., 2013).

Rodents.

Alterations in reward and mesolimbic DA system function have been a common finding in animal models for the study of depression (Douma and de Kloet, 2019; Fox and Lobo, 2019; Kaufling, 2019). Most of these models (3 of which will be discussed in the section below) involve some form of chronic stress, since repeated stress is one of the most important factors for precipitating depression in humans (Kendler et al., 1999; Lloyd, 1980). Rodents exposed to chronic stress typically exhibit anhedonia and other behavioral abnormalities relevant to depression, including behavioral despair and/or passive coping behaviors (Scheggi et al., 2018; Slattery and Cryan, 2017). In animals, common measures of anhedonia include the sucrose consumption and sucrose preference tests (Willner et al., 1987). Briefly, these tests involve measuring the consumption of a palatable sucrose solution during or after exposure to an anhedonia-producing event (i.e. stress) and the preference for a sucrose solution when given a choice between this highly palatable solution and water, respectively. Exposure to various forms of stress decreases sucrose preference in rodents (Katz, 1982; Monleon et al., 1995; Slattery and Cryan, 2017), although the reliability of the sucrose preference test as a measure of anhedonia in animals is subject to controversy (Forbes et al., 1996; Matthews et al., 1995; Weiss, 1997). Nonetheless, decreased consumption or no preference for the sucrose solution over water is thought to reflect anhedonia (Scheggi et al., 2018). Other measures of anhedonia in rodents include elevations in reward thresholds using intracranial self-stimulation (ICSS) and reduced time interacting with a novel conspecific, both of which are thought to reflect hypofunction of mesocorticolimbic reward systems (Carlezon and Chartoff, 2007; Scheggi et al., 2018; Zacharko and Anisman, 1991). Stressed animals also exhibit less time struggling against inescapable stressors such as the FST or tail suspension test (Scheggi et al., 2018; Slattery and Cryan, 2017), which is thought to reflect behavioral despair, and more recently, passive coping (Molendijk and de Kloet, 2019).

In a landmark study, Tye and colleagues revealed a causal role for VTA DA neurons in both the induction of depressive-like symptoms in non-stressed mice and in the alleviation of these symptoms in CMS-exposed mice (Tye et al., 2013). Using optogenetic tools, the authors demonstrated that acute photoinhibition of VTA DA neurons induced a depressive-like state in mice with respect to reduced escape behavior in the tail suspension test and decreased sucrose preference. These depressive-like behaviors were also observed in CMS-exposed mice. Phasic photoactivation of VTA DA neurons in CMS-exposed mice rescued stress-induced behavioral deficits and normalized escape-related behavior and sucrose preference to levels comparable to nonstressed (i.e. control) mice (Tye et al., 2013). Moreover, photoactivation of ChR2-expressing VTA DA neurons increased active escape-related behaviors (i.e. kicks) during the FST in CMS-exposed TH:Cre rats and this behavior was not time-locked to light pulses on the order of seconds, suggesting that it may be modulated by dopaminergic tone rather than driven by individual DA transients. Taken together, these data show bidirectional effects of VTA DA neuron activity on the rapid induction and alleviation of depression-related behaviors in which selective inhibition of VTA DA neurons produces depression-related behavior in measures of both motivation and anhedonia, and acute phasic activation of VTA DA neurons rescues CMS-induced depression-like phenotypes.

Importantly, the effects of stress on the mesolimbic DA system are complex, as different stressors can cause opposite responses from VTA DA neurons depending on pre-exposure, stressor type, timing and severity (Douma and de Kloet, 2019; Fox and Lobo, 2019; Holly and Miczek, 2016; Kaufling, 2019; Valenti et al., 2012). Moreover, the role of DA differs in the induction versus the expression of depression-related behaviors, and the duration of the stressor affects the magnitude and duration of the negative affective state associated with its withdrawal (Grace, 2016; Koob and Le Moal, 2008). For example, acute restraint stress potently activates the DA system (i.e. increases VTA DA neuron population activity) as measured by in vivo extracellular recordings conducted shortly after stress exposure (Valenti et al., 2011). In contrast, 24-hours after restraint stress there is a shift in the opposite direction (i.e. a reduction in DA neuron population activity), as measured using the same in vivo technique (Chang and Grace, 2013). Thus, acute activation of the DA system by stress induced VTA DA neuron activation is followed by an attenuation VTA DA neuron activity, as measured by reduced numbers of active DA cells per track. Specifically, our group has also shown that if the stressor is presented over a longer period of time (e.g. chronic cold stress, 4-week CMS paradigm), the consequent depressive-like state and VTA DA neuron attenuation is also maintained for an extended period (i.e. up to 1-week) after removal of the stressor (Chang and Grace, 2014; Moore et al., 2001; Moreines et al., 2017; Rincon-Cortes and Grace, 2017; Valenti et al., 2012). Thus, we propose that the initial increase in DA observed during and immediately after stress drives downstream plasticity, ultimately resulting in depressive-like states (Grace, 2016). This framework is consistent with findings from other studies in rodent models of chronic stress showing that: i) increasing DA neuron activity during a subthreshold chronic social defeat stress (CSDS) paradigm was necessary for the induction of the depressive-like phenotype that followed (Chaudhury et al., 2013), ii) phasic activation of VTA DA neurons drove recovery from the depressive-like phenotype resulting from CMS exposure (Tye et al., 2013), and iii) activation of DA neurons long after CSDS exposure has antidepressant effects (Friedman et al., 2014), suggesting that VTA hypoactivity may be present in this model as well. Consistent with these data, a study assessing c-Fos expression of VTA DA neurons and DA turnover in the mPFC of wild-type mice showed activation of the mesocortical DA pathway in response to social defeat, and attenuation within this pathway following repeated social defeat in wild-type mice (Tanaka et al., 2012). Within the context of these findings, we propose that the initial stress-induced activation of VTA DA neurons leads to the compensatory long-duration downregulation of DA neuron population activity after stressor withdrawal, which is supported by in vivo electrophysiological findings from our lab showing DA hypofunction in two different animal models relevant to depression (Belujon and Grace, 2014; Chang and Grace, 2014). Finally, we would like to acknowledge that several studies have reported an increase in VTA DA neuron firing rates in CSDS-exposed mice, which lasts up to 2 weeks in susceptible mice and appear to be specific to CSDS exposure (Cao et al., 2010; Krishnan et al., 2007). However, we will not be focusing on this literature because the primary measure of VTA DA neuron activity (i.e. population activity) used by our group in CMS and LH rats has never been assessed in mice exposed to CSDS and the effects of ketamine on VTA DA neuron activity following CSDS have not been examined.

Effects of ketamine on depression-related behavior and mesolimbic DA activity in animal models of repeated/chronic stress

Animal models of psychiatric disorders are procedures applied to laboratory animals which produce behavioral changes that are thought to be homologous to aspects of psychiatric disorders and can therefore be used as experimental tools to further the understanding of human psychopathology (Willner, 2015). Ideally, an animal model will have construct, face and predictive validity. This means that an animal model should: a) recreate some process, environmental risk factor or agent that causes disease in humans, which replicates neural and behavioral features of the illness (i.e. construct validity); b) recapitulate important anatomical, biochemical or behavioral features of a human disease (i.e. face validity); and c) respond to treatment in a way that predicts the effects of those treatments in humans (i.e. predictive or pharmacological validity)(Nestler and Hyman, 2010; Willner, 2015). For example, since stress exposure is a risk factor for depression in humans (Kendler et al., 1999; Lloyd, 1980), rodent models of repeated or chronic stress exposure are commonly used to produce behavioral and physiological abnormalities in rodents, including cellular, molecular and morphological changes within stress-sensitive brain regions that resemble those observed in clinical depression (Hare et al., 2017; Slattery and Cryan, 2017). Importantly, many of these stress-associated neurobehavioral changes can be rescued via treatment with traditional antidepressants, but also with fast-acting acting antidepressants such as ketamine. Below, we discuss stress-induced alterations in depression-related phenotypes with a focus on anhedonia and behavioral despair, which reflect alterations in positive and negative affect, respectively, in rodent models of repeated and/or chronic stress as well as their modulation by ketamine administration.

Chronic Mild Stress (CMS).

CMS is one of the most widely employed animal models used to study stress-induced neurobehavioral adaptations relevant to depression and consists of exposing rodents to a series of mild, repeated and unpredictable stressors (e.g. isolation, water and food deprivation, cage and light cycle disturbances, etc.) over a prolonged period of time (4–6 weeks)(Hill et al., 2012; Willner, 2017). This procedure has been shown to induce anhedonia, or a decreased response to various types of rewards in rodents, suggesting alterations in DA function (Scheggi et al., 2018; Willner et al., 1992). For example, mice and rats exposed to CMS exhibit reduced sucrose consumption and preference, attenuation of conditioned place preference, decreased sexual activity, and increased thresholds for ICSS in the VTA (D’Aquila et al., 1994; Gronli et al., 2005; Li et al., 2011; Monleon et al., 1995; Moreau et al., 1992; Papp et al., 1991; Tye et al., 2013; Willner et al., 1987). CMS also induces depression-related phenotypes (i.e. increased immobility, reduced latency to immobility) in rodent measures of behavioral despair, such as the FST and the TST (Chang and Grace, 2014; Moreines et al., 2017; Rincon-Cortes and Grace, 2017; Tye et al., 2013). Thus, CMS precipitates an anhedonic-like state associated with decreased reward responsiveness and increased behavioral despair, which may also be interpreted as passive coping. Importantly, in vivo electrophysiological studies conducted in CMS-exposed rats have consistently reported a decrease in VTA DA neuron activity that coexists with anhedonic and depressive-like states (Chang and Grace, 2014; Moreines et al., 2017; Rincon-Cortes and Grace, 2017) and is driven by increased activity within the basolateral amygdala (BLA)-ventral pallidal (VP) pathway (Chang and Grace, 2014). Moreover, this attenuation is restricted to the medial VTA DA neurons which project preferentially to the reward-related ventromedial striatum (Moreines et al., 2017). The dopamine deficit observed in CMS-exposed rats appears to be largely conserved in mice: CMS-exposed mice exhibit a decrease in the proportion of active DA neurons, but also in firing rate (Liu et al., 2018; Tye et al., 2013; Zhong et al., 2018).

Acute ketamine (10–15mg/kg; i.p.) treatment reverses the CMS-induced decreases in sucrose consumption in mice and rats (Garcia et al., 2009; Li et al., 2011). Moreover, systemic ketamine administration (10mg/kg) also reverses depressive-like behavior following CMS, as CMS rats of both sexes treated with ketamine exhibited reduced FST immobility and increased latencies to immobility (Rincon-Cortes and Grace, 2017). Ketamine also restored VTA population activity in CMS animals: CMS rats treated with ketamine exhibited greater numbers of active VTA DA cells compared with vehicle-treated CMS rats (Rincon-Cortes and Grace, 2017). Furthermore, the effects of ketamine on VTA population activity persisted between 2–7 days post-injection, indicating that ketamine induces a sustained increase in VTA population activity in male and female rats following CMS. Collectively, these data suggest that ketamine effects on CMS-induced depression-related phenotypes (i.e. anhedonia, despair) are likely mediated via long-lasting actions on the DA system.

Learned Helplessness (LH).

In this paradigm, rats are exposed to an inescapable stressor (i.e. shock) on day 1 in one chamber of a two-chamber shuttle box with the escape route blocked, and tested for active avoidance (i.e. escaping the shock by crossing to the other chamber) on the next consecutive days with the escape route open (Seligman and Beagley, 1975; Wagner et al., 1977). Failures to escape and the latency to escape are used to discriminate between nonhelpless and helpless rats: helpless rats typically exhibit a deficit in escape behavior, as indexed by increased number of failures to escape and longer latencies to escape compared with nonhelpless rats (Koike et al., 2011). Helpless rats also exhibit hyposensitivity to rewarding events, such as a reduced preference for sucrose (Sanchis-Segura et al., 2005), suggesting alterations in DA system function. Indeed, DA activity is different in helpless versus nonhelpless rats. A recent electrophysiological study conducted in Wistar-Kyoto rats found that rats exposed to the learned helplessness paradigm fell into two groups despite identical treatment: those showing helpless behavior and those showing normal escape. The nonhelpless rats had numbers of spontaneously active DA neurons (i.e. population activity) that were comparable to no-shock controls (Belujon and Grace, 2014). However, rats that exhibited helpless behavior showed a significant reduction in the number of active DA neurons within the VTA (Belujon and Grace, 2014). Thus, in helpless animals, a decrease in the number of spontaneously active DA neurons in learned helplessness is consistent with a decrease in VTA activity driving stress-induced depressive-like behavior following CMS exposure (Grace, 2016; Tye et al., 2013).

Injection (5–10mg/kg) of repeated or acute ketamine restored escape behavior in helpless rats, in that the rats had fewer escape failures and exhibited faster latencies to escape compared with vehicle-treated helpless rats (Belujon and Grace, 2014; Koike et al., 2011). This is in accordance with a prior study conducted in mice, which found that a single injection of ketamine (2.5mg/kg, i.p.) significantly reduced both the number of escape failures and the latency to escape in mice that had developed helplessness (Maeng et al., 2008). Furthermore, both repeated and acute ketamine treatment reversed the decrease in VTA population activity observed in helpless rats, and this effect was found to be mediated by restored plasticity (i.e. long-term potentiation) within the hippocampus-accumbens pathway (Belujon and Grace, 2014).

Chronic Social Defeat Stress (CSDS).

This paradigm focuses on social stressors and is based on resident-intruder paradigms in rodents, in which an adult animal (i.e. the intruder) is introduced into the home cage of a more aggressive dominant male of the same species (i.e. the resident), which leads to an antagonistic encounter involving fights and hostile interactions that results in the intruder adopting a submissive posture (Bartolomucci, 2009; Berton et al., 2006; Miczek, 1979). This physically stressful interaction is brief and generally followed by physical separation of two animals by a transparent plastic plate pierced with holes that is placed in the middle of the resident cage. This allows adding an emotional or psychogenic stress to the brief physical stress since the intruder will be able to see and smell the resident. In case of chronic models, the intruder is subjected to assault of a new resident every day or intermittently according to the protocol. Notably, this model results in the creation of two types of subjects: susceptible vs resilient: 40–50% of mice subjected to this paradigm will not develop depressive-like states (Krishnan et al., 2007). Thus, this model recapitulates the resilience phenomenon also observed in humans. CSDS profoundly alters the motivation for social interactions (Berton et al., 2006; Toth and Neumann, 2013) and brain reward function (Der-Avakian et al., 2014) in rodents. In contrast to undefeated (i.e. control) or resilient mice, mice susceptible to CSDS exhibit social avoidance and spend less time in close proximity to the unfamiliar, novel mouse (Bagot et al., 2017; Berton et al., 2006; Browne et al., 2018; Krishnan et al., 2007). This effect (i.e. social avoidance) lasts up to 4 weeks after the last defeat in susceptible mice (Berton et al., 2006; Krishnan et al., 2007) and is accompanied by alterations in the firing rates of VTA DA neurons, as measured using ex vivo and in vivo preparations, that are negatively correlated with social avoidance behavior (Cao et al., 2010; Krishnan et al., 2007). Specifically, mice exposed to CSDS exhibit a significant increase in VTA DA neuron firing rates the day after the last defeat, and this increase in firing rate persists up to 2 weeks in susceptible mice (Krishnan et al., 2007). In addition, rats and mice exposed to CSDS show other depression-related phenotypes including anhedonia, as indexed by reduced sucrose preference (Donahue et al., 2014; Dong et al., 2017; Krishnan et al., 2007) and elevated ICSS thresholds (Der-Avakian et al., 2014; Donahue et al., 2014), as well as negative affect, as indexed increased FST and TST immobility (Der-Avakian et al., 2014; Dong et al., 2017; Rygula et al., 2005).

Acute ketamine treatment reverses the socially-avoidant phenotype in CSDS mice and increases time spent in social interaction, but this effect has typically been observed at higher doses ranging from 10mg/kg to 20mg/kg i.p. in both male and female mice (Bagot et al., 2017; Donahue et al., 2014; Newman et al., 2019) and a recent study using 10mg/kg i.p. found no effect of ketamine on social avoidance in male mice (Browne et al., 2018). Despite its effect on social avoidance, acute ketamine treatment (20mg/kg, i.p.) failed to block the anhedonic effects of CSDS on ICSS (Donahue et al., 2014). However, a lower dose of ketamine (10mg/kg, i.p.) was sufficient to reverse anhedonia in the sucrose preference test, as susceptible CSDS mice treated with ketamine exhibited increases in sucrose preference (Dong et al., 2017; Wang et al., 2019). Moreover, this dose of ketamine also reversed behavioral despair in susceptible CSDS mice, as ketamine-treated susceptible mice exhibited reduced immobility in both the FST and the TST compared to vehicle-treated mice (Dong et al., 2017; Wang et al., 2019). The effects of ketamine administration on CSDS-induced electrophysiological alterations within VTA DA neurons remain to be determined. However, preclinical studies using molecular and genetic tools have revealed a vast array of changes in multiple regions within the mesolimbic DA system that are interconnected with the VTA, such as the PFC, the amygdala, and the nucleus accumbens (Bagot et al., 2017).

Circuitry Underlying Ketamine Actions on VTA Population Activity in the LH and CMS Rodent Models

As reviewed above, downregulation of the DA system has been described in the LH and CMS models, and this downregulation is reversed by ketamine (Belujon and Grace, 2014; Rincon-Cortes and Grace, 2017). Furthermore, a mechanism underlying ketamine’s antidepressant effects following learned helplessness has been identified (Belujon and Grace, 2014). In helpless rats, there is alteration of synaptic plasticity within the hippocampal-accumbens circuit responsible for activation of DA neuron population activity: the ventral subiculum (vSub)- NAc pathway. Briefly, activation of the vSub induces an increase in the number of spontaneously active DA neurons (i.e. population activity) without impacting average firing rate or bursting activity (Lodge and Grace, 2006), and this occurs via a polysynaptic pathway through the NAc and ventral pallidum (VP)(Floresco et al., 2001) (see Fig. 1 for overview of afferent regulation of VTA population activity).

Figure 1. Afferent regulation of VTA DA neuron activity by two opposing circuits.

The DA system is under regulation by an activating circuit consisting of the ventral subiculum of the hippocampus (vSub)-nucleus accumbens (NAc)-ventral pallidum (VP) as well as an inhibitory circuit consisting of the basolateral amygdala (BLA)-VP pathway. At baseline, about half of all VTA DA neurons are firing spontaneously. The spontaneous tonic firing state is regulated by the vSub through excitatory projections to the NAc, which in turn, inhibit the VP and release silent DA neurons from inhibition resulting in spike activity. Thus, activation of the vSub induces an increase in the number of spontaneously active DA neurons, without affecting burst firing. This is significant since only DA neurons that are firing spontaneously can respond to rapid phasic activation with bursts of action potentials, and this is driven by activation of the pedunculopontine tegmental nucleus (PPTg) glutamtergic afferents acting on NMDA receptors. This activation can be offset by the BLA, which can excite the VP and inhibit VTA DA neuron population activity. Normally these systems are in balance, keeping the DA neurons in an active state but capable of responding rapidly to an activating stimulus but kept in check by the inhibitory circuit.

Tetanic stimulation of the vSub-NAc pathway induced the expected long-term potentiation (LTP) in control and non-helpless rats; however, in the helpless rats the same stimulation induced long-term depression (LTD) (Belujon and Grace, 2014). This finding has been replicated using chronic multimodal stress in mice, in which stressed mice exhibit anhedonia and impaired LTP induction at hippocampal-NAc synapses (LeGates et al., 2018). Collectively, these data suggest that altered excitability (i.e. downregulation) within the vSub-NAc-VP (i.e. activating) pathway in helpless rats contributes to the decrease in VTA DA neuron population activity and anhedonia. This is consistent with Koob’s opponent process model (Koob and Le Moal, 2008), in which we posit initial stress activates the vSub-NAc-VP-VTA pathway to activate the DA system, but this causes a delayed but prolonged compensatory activation of the BLA-VP-VTA inhibitory circuit, which attenuates DA neuron population activity in the VTA. Thus, once the vSub-NAc pathway is taken offline by the chronic stress, the BLA-VP inhibitory pathway predominates and drives VTA DA neuron attenuation (Fig. 2). Importantly, ketamine administration restored stimulus-induced LTP (i.e. excitatory drive) in the vSub-NAc-VP pathway, which was accompanied by normalized escape behavior and restored DA neuron population activity in helpless rats (Belujon and Grace, 2014).

Figure 2. Circuit mechanisms underlying reduced VTA population activity following chronic or repeated stress.

Regulation of tonic DA activity is under 2 distinct and opposing circuits: an activating circuit (vSub-NAc-VP), which releases DA neurons from VP GABAergic inhibition and one inhibitory circuit (BLA-VP), which increases VP GABAergic inhibition of DA neurons (see Figure 1). In animal models of chronic stress, there is a downregulation in VTA population activity, as indexed by reduced numbers of spontaneously active DA neurons, which limits the amplitude of the phasic DA response that is driven by the PPTg. Activation of the BLA by chronic or repeated stress drives an attenuation in VTA DA neuron population activity via enhanced excitatory drive onto the VP, which increases inhibition of DA neuron firing. Chronic stress also attenuates excitation within the vSub-NAc-VP pathway, which increases the number of DA neurons that are under inhibitory control by the VP. Hence, chronic stress increases drive of the BLA-VP (i.e. inhibiting) circuit while also decreasing excitatory drive within the vSub-NAc-VP (i.e. activating) circuit, both of which converge to inhibit VTA DA neuron firing.

One question that remains is whether the site of action of ketamine is in the PFC or the hippocampus. In normal (i.e. naïve) rats, antidepressant-like effects of ketamine have been demonstrated in the FST and this effect is dependent on the infralimbic PFC (Fuchikami et al., 2015) and a ventral-hippocampal-mPFC pathway (Carreno et al., 2016). However, this was not tested in animal models for depression, in which our data show primary effects in the vSubaccumbens pathway in helpless rats. One potential explanation is that the impact in the vSub-NAc-VP pathway can only be observed when the rats are in a pathological-like state (i.e. chronic stress) in which the vSub-NAc-VP pathway has been down-regulated (Belujon and Grace, 2014). On the other hand, our group has previously showed that the mPFC is required for the vSub to exert maximal excitatory drive on the NAc and plays an essential facilitatory role in regulating vSub-NAc-VP information flow (Belujon and Grace, 2008). These data suggest that interference with the mPFC could disrupt the vSub-NAc-VP circuit in normal animals; however, in the depressive-like state, normalization of the vSub-NAc downregulation can account for the impact of ketamine.

Sex differences in the rapid and sustained antidepressant-like effects of ketamine

Women are diagnosed with depression at twice the rates of men (Kessler, 2003; Kornstein et al., 2000), and this has been partially attributed to pronounced sex differences in both the anatomy and function of the brain (Altemus et al., 2014; Grigoriadis and Robinson, 2007). As discussed above, stress exposure is a critical risk factor for mood disorders that are more prevalent in females (i.e. anxiety, depression), and sex differences in behavioral and physiological responses to stress are thought to contribute to this enhanced vulnerability (Bangasser and Valentino, 2014; Goel and Bale, 2009; Hodes and Epperson, 2019; Rincon-Cortes et al., 2019). Moreover, there are important sex differences in the pharmacokinetics and pharmacodynamics of antidepressants as well as in antidepressant treatment response (Damoiseaux et al., 2014; Keers and Aitchison, 2010; Sramek et al., 2016), and this extends to rodent models of depression and antidepressant action (Dalla et al., 2010; Williams and Trainor, 2018). However, until recently, most preclinical studies examining the neurobiological mechanisms of mood disorders and antidepressants have focused only on males (Beery and Zucker, 2011; Karp and Reavey, 2018). Growing awareness of sex differences in neurobiology and behavior, along with recent research policies at the National Institutes of Health (NIH) (Clayton and Collins, 2014; Zakiniaeiz et al., 2016), have increased focus on implementing sex as a biological variable and assessing how it can modulate behavioral responses to pharmaceuticals with antidepressant properties. This shift could ultimately contribute to better treatment for mood disorders. Clinical evidence supports the notion that responsiveness to traditional antidepressant drugs (e.g. SSRIs, tricyclic antidepressants) is sex-dependent and modulated by ovarian hormones (Damoiseaux et al., 2014; Keers and Aitchison, 2010; Sloan and Kornstein, 2003), and here we review preclinical evidence that these sex differences extend to ketamine’s antidepressant actions.

The first preclinical study to consider sex differences in the antidepressant effects of ketamine was conducted in stress naïve rats and demonstrated that female rats are much more sensitive to ketamine compared to male rats. Specifically, a low dose (2.5 mg/kg, i.p.) of ketamine was sufficient to reduce FST immobility and latency to feed in a novel environment in female Sprague-Dawley rats, but not in males (Carrier and Kabbaj, 2013). The researchers also found that the antidepressant-like response induced by ketamine was abolished in ovariectomized (OVX) female rats, but effective in OVX female rats that received replacements of both estrogen and progesterone mimicking the 4-day estrous cycle in intact rats, suggesting that the gonadal hormones estrogen and progesterone mediate the high sensitivity to low dose ketamine in female rats (Carrier and Kabbaj, 2013). A subsequent study conducted in rats exposed to CMS highlighted sex differences in stress-induced alterations on behavior and DA system function, although male and female rats exhibited comparable sensitivity to the rapid and sustained antidepressant effects of a higher dose (10mg/kg, i.p.) of ketamine on FST immobility and DA activity, respectively. Acute ketamine administered 30-minutes prior to the FST effectively reduced immobility in stressed animals and increased VTA DA neuron activity for up to 1 week in both sexes compared with vehicle-treated animals (Rincon-Cortes and Grace, 2017).

The greater sensitivity of female rodents to low dose ketamine was confirmed in later studies conducted in stress-naïve C57BL/J male and female mice (Dossat et al., 2018; Franceschelli et al., 2015; Zanos et al., 2016). Pre-treatment with a low dose (3mg/kg, i.p.) of ketamine 30-minutes prior to the FST reduced immobility in female mice but not in male mice (Franceschelli et al., 2015), and this sensitivity was enhanced in proestrus (i.e. high estrogen) females, which exhibited an antidepressant-like response in the FST to even lower doses (1.5 mg/kg, i.p.) of ketamine (Dossat et al., 2018). Sex differences in FST performance were also observed 24-hours post injection: females receiving 5mg/kg or 10mg/kg of ketamine exhibited reduced FST immobility, whereas only males receiving 10mg/kg (i.e. the highest dose) exhibited reduced FST immobility (Franceschelli et al., 2015). Thus, stress-naïve female mice are more sensitive to both the rapid and the sustained antidepressant-like effects elicited by ketamine, and these effects were demonstrated to be due to a different pharmacokinetic profile of ketamine metabolites in male versus female mice (Zanos et al., 2016). When given the same dose of ketamine, levels of (2S,6S;2R,6R)-HNK, the metabolite proposed to underlie the antidepressant actions of ketamine, were approximately three-fold higher in the brains of female mice compared to males (Zanos et al., 2016). Importantly, the study led by Franceschelli also compared antidepressant actions of ketamine in CMS-exposed male and female mice and found no effect of ketamine on CMS-induced decreases in sucrose consumption and a sex-dependent effect in FST immobility, in which a 10mg/kg dose reduced FST immobility at days 1 and 7 in males, but only at day 1 in females (Franceschelli et al., 2015). Taken together, these data suggest that stress-naïve female mice are more sensitive to the rapid and sustained antidepressant effects of ketamine in the FST and that CMS-exposed females are more reactive to the earlier antidepressant effects of ketamine (i.e. at 30-min, 24h), although the antidepressant effects of ketamine on FST performance following CMS are longer lasting in males.

Sex differences have also been reported in the antidepressant effects of ketamine following isolation stress. Chronic social isolation (8–11 weeks) during adulthood alters behavior, spine density, and synaptic molecules in the mPFC in a sex-dependent manner (Sarkar and Kabbaj, 2016). Male rats exposed to social isolation exhibited reduced sucrose preference, greater FST immobility duration, a decline in mPFC spine density, and reduced levels of synaptic proteins (PSD-95, GluR1, synapsin). Similar alterations were observed in socially-isolated female rats, with the exception of reduced sucrose preference (i.e. no effect in females). In male rats, acute administration of ketamine (5mg/kg, i.p.) rescued the isolation-induced reduction in sucrose preference and increase in FST immobility time (Sarkar and Kabbaj, 2016). These effects were dose-dependent, as a 2.5 mg/kg dose of ketamine did not improve behavior in the sucrose preference and FST in males. In females, no effect of ketamine was found for sucrose preference, and both doses (2.5mg/kg, 5mg/kg) elicited a reduction in FST immobility time. Importantly, sex-dependent ketamine effects on mPFC spine density were observed: ketamine increased the density of thin spines, but not mushroom spines, in isolated male rats but had no effect on spine density in females (Sarkar and Kabbaj, 2016). Consistent with the behavioral and spine density experiments, a single injection of ketamine (5mg/kg) reversed the decline in levels of synapsin 1, PSD-95 and attenuated the decline in GluR1 in isolated male rats. However, in contrast to male rats, ketamine did not elevate the levels of these proteins in isolated female rats, which is consistent with the lack of effect of ketamine on sucrose preference and spine density in female rats (Sarkar and Kabbaj, 2016). Altogether, these data demonstrate sex differences in the effects of chronic isolation stress and ketamine administration on behavior, mPFC spine density and synaptic protein levels, thereby suggesting divergent underlying mechanisms for the efficacy of ketamine in the two sexes.

Ketamine as a prophylactic against stress-induced adaptations relevant to depression

In addition to reversing a wide variety of stress-induced deficits across various models of chronic stress, there has been growing interest in discovering manipulations and treatments that might increase resistance/resilience to the effect of future stressors (Snijders et al., 2018). This has led some groups to investigate whether ketamine can protect against the neurobehavioral impact of future stress exposure. Parise and colleagues were the first to show that repeated ketamine administration (20mg/kg; i.p.; twice a day × 15-days) resulted in an enduring stress-resistant (i.e. resilient) phenotype that was detectable up to 2 months after the last dose (Parise et al., 2013). The researchers found that rats treated with ketamine during adolescence (postnatal days 35–49) or adulthood (postnatal days 75–89) spent more time in the open arm of the EPM, indicating an anxiolytic effect, and had longer latencies to immobility and reduced immobility duration in the FST compared with age-matched, vehicle-treated controls. Moreover, repeated ketamine administration increased active coping behavior (i.e. swimming) in the FST (Parise et al., 2013). Similar behavioral effects were observed regardless of whether the ketamine was administered in adolescence or adulthood, suggesting that these effects were independent of age at the time of ketamine administration (i.e. no sensitive period for the prophylactic effects of ketamine). Moreover, these prophylactic effects could not be attributed to ketamine-induced changes in basal locomotor activity, because no differences were detected 2 months after drug exposure (Parise et al., 2013). Collectively, these data show that repeated ketamine exposure yields long-lasting (~2 months) resilience against anxiety and depression-like behavioral responses in adolescent and adult rats.

Importantly, the prophylactic effect of ketamine is also observed in mice and rats that have been exposed to chronic stress. In mice, a single injection (30mg/kg; i.p.) of ketamine administered 1-week prior to 2-weeks of CSDS prevented the stress-induced increase in FST immobility, as indexed by reduced immobility levels in defeated mice treated with ketamine compared with vehicle-treated mice that underwent CSDS, as well as social avoidance in defeated mice (Brachman et al., 2015). These effects were specific to depression-related phenotypes, as ketamine did not protect against CSDS-induced increases in anxiety-like behavior in the EPM. Moreover, a later study found that prophylactic ketamine (30mg/kg; i.p.) effects in defeated mice are mediated by increased expression of △FosB, a transcription factor that regulates synaptic plasticity in brain reward regions and mediates resilience to stress (Nestler et al., 1999; Vialou et al., 2010) in the ventral CA3 region of the hippocampus (Mastrodonato et al., 2018). Furthermore, this same dose of ketamine given 1-week before LH training also protected against helpless behavior: mice injected with ketamine exhibited reduced latencies to escape the shock and shorter session lengths compared with saline-treated mice (Brachman et al., 2015). In addition to the CSDS and LH models, ketamine’s protective effects were also demonstrated in the chronic CORT model, which consisted of 3-weeks of chronic CORT administration in C57BL/6NTac mice. Chronic CORT treatment increased FST immobility duration and latency to groom in the sucrose splash test. Here, ketamine prevented increases in FST immobility duration and grooming latency, although the effective dose was substantially higher (90mg/kg; i.p.)(Brachman et al., 2015). Taken together, these findings indicate that ketamine can induce persistent stress resilience against depressive-like behaviors in 3 distinct mouse models of chronic stress. In rats, a single dose (10mg/kg; i.p.) of ketamine given 2 hours, 1 week or 2 weeks before inescapable shock protects against the neurochemical (i.e. increased 5-HT release in BLA) and behavioral (i.e. reduced social investigation) changes that typically follow inescapable, uncontrollable tail shocks (Amat et al., 2016). These results support findings (Brachman et al., 2015) using a different species, different stressor paradigm and different behavioral endpoints, suggesting that the finding of prophylactic ketamine effects has generalizability and replicability. Furthermore, these data also indicate that ketamine not only reverses the effect of past stressors but can also buffer against the effects of future stressors.

Given that the antidepressant effects of ketamine are known to be sex-specific, future studies should determine whether the prophylactic effects of ketamine extend to female animals and whether they vary depending on stressor type. Since all studies aimed at exploring the prophylactic effects of ketamine have been conducted in males, this is an important question that remains to be determined. Another future direction involves exploring whether the prophylactic effects of ketamine translate to humans. To date, only one group has explored the effect of prophylactic ketamine administration in humans. Ma and colleagues found that 0.5mg/kg given 10 minutes after child birth reduced the prevalence of postpartum blues and postpartum depression on postpartum depression in Chinese women that had undergone cesarean section (Ma et al., 2019).

It is important to keep in mind that ketamine is a powerful psychotomimetic agent that can produce transient dissociative and psychotic-like effects (especially at high or prolonged doses), and is a drug of abuse, which limits its widespread use in humans (Gerhard and Duman, 2018; Zhu et al., 2016). Since the ultimate goal of drug development efforts is to identify rapid acting agents that reverse and or prevent stress-induced neuronal deficits caused by stress and depression, but with fewer side effects, this has sparked further investigation into the chemical makeup of ketamine in hopes of discovering a more specific and efficacious treatment (Gerhard and Duman, 2018). Notably, ketamine is metabolized into an array of metabolites, and recent evidence suggests that an active metabolite of ketamine (2R,6R)-HNK exerts antidepressant-like effects in mice, but without the side effect behaviors involving sensory dissociation, ataxia, and abuse liability (Zanos et al., 2016; Zanos et al., 2018), as well as prophylactic efficacy in rodents (Chen et. al, biorXiv, 2019). Importantly, HNK metabolites have also been detected in humans following ketamine infusion, with (2R,6R;2S6S)-HNK and (2S,6R;2R,6S)-HNK being the predominant circulating HNKs in plasma (Zanos et al., 2018; Zarate et al., 2012; Zhao et al., 2012). However, clinical trials are required to determine the efficacy and side effect profile of (2R,6R)-HNK as a potential prophylactic agent in humans, as findings in rodent models do not always translate to the clinical setting. Finally, there is also the possibility that new compounds with safer pharmacological profiles will be used as prophylactics, as a recent report suggests prophylactic efficacy of serotonin 4 receptor (5-HT₄R) agonists against stress in mice (Chen et al., 2019).

Conclusions

In summary, depression is characterized by alterations in both negative and positive affect (i.e. depressed mood, anhedonia) and is associated with dysfunction of the mesolimbic DA system. These changes are recapitulated in rodent models of repeated or chronic stress, which are widely used to produce depression-related behaviors (i.e. despair, anhedonia) and mesolimbic DA deficits. Traditional antidepressants act via blocking the reuptake of monoamines, particularly serotonin. However, although they block uptake sites immediately and rapidly increase serotonin levels, weeks to months of treatment are required to obtain a therapeutic response. This suggests that it is not the increase in serotonin per se that is the therapeutic effect, but instead reflects a brain adaptation to the increased serotonin. In other words, a change in brain circuit activity or plasticity. In contrast, there are treatments that act faster, such as electroconvulsive therapy (ECT) and ketamine. The precise mechanisms through which these treatments act is a matter of intense research. Indeed, acute ketamine administration is sufficient to improve depressive symptoms in humans and reverse stress-induced changes in behavior and DA function in stressed animals. These antidepressant effects are mediated via actions on afferent circuits that regulate VTA DA neuron activity. Thus, we posit that ketamine and ECT have the same general effect – an alteration of synaptic plasticity, but over a more rapid time course. Therefore, rather than treating a persistent deficit in neurotransmission, these actions most likely alter synaptic strength between brain regions involved in affective regulation. Our work suggests that, at least with respect to the DA system, this involves a delicate balance between pathways that drive the DA system (i.e., the vSub-NAc-VP-VTA pathway) and the pathway that attenuates DA system function (i.e., the BLA-VP-VTA pathway; Fig. 1). Normally these systems would be in balance. However, if a stressor overdrives the hippocampal pathway, it would cause an activation of the DA system. This would be offset by a slower but persistent compensatory down-regulation via the ilPFC-BLA-VP pathway (Grace, 2016). This is consistent with Koob’s opponent process model of addiction (Koob and Le Moal, 2001, 2008)– i.e., initial activation of the VTA DA system is followed by a persistent down-regulation of this system and negative affect. Indeed, our studies suggest that high dose amphetamine may produce the same consequence – acute DA activation during the reinforcement phase, followed by an amygdala-driven persistent down-regulation producing negative affect (Belujon et al., 2016; Rincon-Cortes et al., 2018). Moreover, in rodents, both stress and antidepressant effects of ketamine vary between sexes, with females showing greater stress-induced neurobehavioral deficits while also exhibiting antidepressant-like responses to lower doses of ketamine. Whether there are sex differences with regard to ketamine’s clinical efficacy in humans remains to be determined. Finally, novel preclinical evidence suggests that, in addition to antidepressant effects, ketamine also has prophylactic effects and can circumvent the neurobehavioral impact of future stressors. Therefore, ketamine may serve as a type of system stabilizer, in that its impact on normal systems is transient, but can restore and may even prevent abnormal plasticity in cases of persistent stress-induced pathological changes.