Both animals and human beings are challenged with a variety of physical and psychological problems on a daily basis. It is theorized that distinct networks are dynamically recruited to fulfill various tasks. The default mode network (DMN) is associated with introspection. The salience network (SAL) processes salient information from external sources, and the central executive network (CEN) participates in working memory and attention. Studies have demonstrated an increase in the activity of the DMN and a decrease in SAL and the CEN in depressed patient. These alterations may increase the time patient spend in rumination or introspection, and impair their ability to deploy attentions to tasks that require responses to external stimuli (Greicius et al., 2007; Kaiser et al., 2015; Evans et al., 2018).

In healthy subjects, ketamine decreases functional connectivity of the DMN to the mPFC (Scheidegger et al., 2012), thus may disengage the mPFC from DMN in introspection status. In MDD patients, ketamine normalizes disconnectivity between the PFC and the rest of the brain (Abdallah et al., 2017; Kraus et al., 2020), and correlation analysis suggests that connectivity changes might be used as a prediction of subject’s response to the antidepressant effect of ketamine. In the PFC of rodents, ketamine increases gamma-band power, a putative measure of cortical disinhibition and an indicator of fast ionotropic excitatory receptor activation (Muthukumaraswamy et al., 2015).

These studies suggest that ketamine might exert its antidepressant-like effects by normalizing the PFC network activity and connectivity. Meanwhile, cortical electrophysiological measurement via non-invasive approaches may serve as a biomarker to evaluate the effectiveness of the antidepressant-like action of ketamine. However, more invasive methodologies on animal models are necessary to delineate the mechanism of ketamine’s antidepressant-like action at molecular level.

With the advancement of optogenetic and chemogenetic techniques, accumulating evidences indicate that several brain regions functionally connected with the PFC are highly involved with mood disorders including anxiety and depression. Ketamine may exert antidepressant-like effects via modulation of activity and plasticity in these neural circuitries. We will briefly discuss recent discoveries corroborating this hypothesis in the following paragraph.

In a learned helplessness paradigm model, a single dose of ketamine successfully rescued escape behavior in mice. Although fiber photometry and chemogenetic inhibition demonstrated that dopaminergic neuron activity in the ventral tegmental area (VTA) is necessary for the behavioral effects of ketamine, the authors suggested against a direct cell-autonomous regulation of VTA dopaminergic neurons by ketamine. Further experiments revealed that ketamine increased population activity in D1R-expressing pyramidal neurons in the mPFC (Marcus and Bruchas, 2021, Wu et al., 2021b). These neurons synapsed onto VTA dopaminergic neurons, thus indirectly altered the responses of VTA dopaminergic neurons in escape behavior. CCK administration into mPFC mimics the anxiogenic- and depressant-like effects of social stress in mice. Optogenetical stimulation of mPFC to basolateral amygdala (BLA) projection blocked the anxiogenic effect of CCK, whereas optogenetical stimulation of the mPFC to nucleus accumbens (NAc) reversed CCK-induced social avoidance and sucrose preference deficits (Vialou et al., 2014). Optogenetic stimulation of the projection from the infralimbic cortex (IL), a subregion of the mPFC, to lateral septum (LS) promoted anxiety-related behaviors, whereas activation of the projection from the IL to the central nucleus of the amygdala (CeA) exerted anxiolytic and effects (Chen et al., 2021). Optogenetic stimulation of the projection from the mPFC to the dorsal raphe (DRN) significantly increased kick frequency in forced swimming test in rat, indicating an antidepressant response. On the contrary, activation of the projection from the mPFC to the lateral habenula (LHb) decreased mobility in forced swimming test, indicating opposite roles of the mPFC-to-DRN and mPFC-to-LHb neural pathways in driving motivated behavioral response to a challenging environment (Warden et al., 2012). In forced swim test, pharmacological and optogenetic inactivation of the ventral hippocampus (vHipp) prevented the sustained antidepressant-like effect of ketamine. On the contrary, optogenetic and chemogenetic activation of the vHipp-mPFC pathway mimicked the antidepressant-like response to ketamine in a pathway specific manner (Carreno et al., 2016). Optogenetic activation of Drd1-expressing pyramidal cells in the mPFC mimicked rapid and sustained antidepressant-like effects of ketamine in behavioral responses. On the contrary, local infusion of D1R antagonist, optogenetic inhibition and chemogenetic inhibition of Drd1-expressing pyramidal neurons in the mPFC abolished the antidepressant-like actions of systemic ketamine. Stimulation of mPFC Drd1 terminals in the BLA recapitulates the antidepressant-like effects of somatic stimulation in the mPFC (Hare et al., 2019).

To summarize, these studies suggest that several brain regions, including VTA, NAc, BLA, CeA, vHipp, DRN, LHb, PAG, and LS, are directed targeted by PFC afferents and are implicated in depressive-like behaviors, and these regions may employ adaptive changes by antidepressant dose of ketamine. Although it is generally acknowledged that ketamine increases the network activity and induces glutamate surge (will be discussed later) in the mPFC, optogenetic and chemogenetic activation of the downstream targets of the PFC leads to different behavioral outcomes (anxiolytic vs. anxiogenic) depending on the particular circuit investigated. How a unanimous excitation in the PFC, generated by systemic ketamine administration, is translated into distinct activity profiles in different brain regions receiving reciprocal projection from the PFC? Since the total behavioral outcome of ketamine administration is antidepressive-like, does is mean the activation of anxiolytic regions can dominate and successfully antagonize the activation of anxiogenic regions under the influence of ketamine? Or other mechanisms are also involved in remodeling the activity of brain network by ketamine? Answering these questions at neural circuit level will deepen our understanding about the etiology of depression.

Ketamine was at first identified as an N-methyl-D-aspartate glutamate receptor (NMDAR) antagonist (Figure 2). Recently, a detailed structural analysis revealed the mechanisms by which ketamine inhibits NMDA receptors. Both hydrogen-bond (on GluN1 subunit) and steric-sensitive hydrophobic interactions (on GluN2 subunit) are essential to stabilize the binding of ketamine enantiomers. These principles apply to ketamine’s binding with both GluN1–GluN2A and GluN1–GluN2B NMDA receptors. Structural comparison among ionotropic glutamate receptors suggested that amino acid differences in key sites may interfere with the binding of ketamine, and may dictate the selectivity of ketamine for NMDA receptors over AMPA and kainate receptors. On the other hand, the hydroxyl group on cyclohexane in HNK may disrupt the hydrophobic interaction with GluN2 subunits, rendering HNK a much less potent antagonist for NMDAR blockade (Zhang et al., 2021a).

It has been repetitively verified that one of the mechanisms underlying the rapid antidepressant effects of ketamine is through blockade of NMDARs on GABAergic inhibitory interneurons in the PFC, which leads to a rapid increase glutamatergic release and activation of post synaptic AMPARs (Henley and Wilkinson, 2016; Krystal et al., 2019). A subanesthetic dose of ketamine significantly increases glutamate cycling (Chowdhury et al., 2017) and presynaptic glutamate release (Moghaddam et al., 1997) in the PFC. As a consequence of augmented glutamate neurotransmission, ketamine increases overall excitability of the PFC in healthy volunteers and animal models (Zanos and Gould, 2018; Zhang et al., 2021a). Additionally, ketamine enhances gamma band electroencephalography power, which is closely related with cortical disinhibition (Hong et al., 2010).

Perfusion of ketamine significantly diminishes the frequency of action potential in fast spiking interneurons in the mPFC, and pharmacological inhibition of the parvalbumin (PV) neurons in the mPFC abolishes the antidepressant-like effects of ketamine in behavioral animal models (Zhang et al., 2021a). Subanesthetic ketamine rapidly suppresses the activity of somatostatin (SST) interneurons in the PFC. This loss of dendritic inhibition by SST neurons augments evoked synaptic calcium transients in the apical dendritic spines of pyramidal neurons (Ali et al., 2020). In the mPFC, virus-mediated knockdown of the GluN2B subunit of NMDA receptor in GABAergic interneurons blocks ketamine-induced disinhibition of principle neurons, and occludes the antidepressant-like actions of ketamine in behavioral animal models (Gerhard et al., 2020).

The preferential action of ketamine at inhibitory interneurons is postulated to be due to higher frequency of interneuron firing compared with pyramidal neurons, which allows for depolarization-dependent relief of Mg²⁺ blockade and exposure of binding site for ketamine at NMDAR channel pore selectively (Zanos and Gould, 2018). Ketamine may selectively inhibit certain subunits of NMDARs predominantly expressed in interneurons. For instance, at physiological extracellular Mg²⁺ concentration, ketamine is reported to have higher affinity for NMDARs containing GluN2D subunit, which are abundantly expressed in forebrain interneurons (Monyer et al., 1994; Perszyk et al., 2016). This propensity may also contribute to the observed preference of inhibition on GABAergic transmission by ketamine.

Brain-derived neurotrophic factor (BDNF) belongs to a small family of secreted proteins such as nerve growth factor, neurotrophin 3 and neurotrophin 4. BDNF-mediated signaling pathway is acknowledged as a key regulator of neural circuit development and activity-dependent processes, which include neuronal differentiation and growth, synapse formation and plasticity, and higher cognitive functions (Duman and Monteggia, 2006; Park and Poo, 2013; Castren and Antila, 2017).

Systemic or local administration of BDNF exerts antidepressant-like effects (Shirayama, et al., 2002; Hoshaw et al., 2005; Schmidt and Duman, 2010), and over-expression of BDNF in the hippocampus renders animals resilient to chronic stress (Taliaz et al., 2011). It is proposed that essentially all antidepressant drugs promote activation of the high-affinity BDNF receptor tropomyosin receptor kinase B (TrkB) signaling, and this signaling is required for their behavioral effects (Duman and Monteggia, 2006; Autry and Monteggia, 2012; Castren, 2013; Castren and Antila, 2017). Earlier work demonstrated that, in the hippocampus, acute administration of systemic ketamine is associated with increased BDNF protein levels (Garcia, Comim et al., 2008), and a rapidly elevation of phosphorylated TrkB, an indicator of activated BDNF/TrkB signaling pathway (Autry et al., 2011). On the other hand, ketamine failed to exert antidepressant-like effects in transgenic mice with BDNF gene specifically knockdown in the forebrain (Autry et al., 2011), and infusion of BDNF neutralizing antibody into the mPFC abolished ketamine’s antidepressant-like responses in forced swim test (Lepack et al., 2014). Mice carrying human BDNF ^Val66Met single nucleotide polymorphism (SNP), which compromises BDNF processing and induces deficits in activity-dependent secretion of BDNF, failed to response to ketamine’s antidepressant-like action (Liu et al., 2012).

Recently, it is reported that ketamine binds to BDNF receptor TrkB in a stereoselective manner (Casarotto et al., 2021). By directly binding to a site formed by a dimer of TrkB transmembrane domains with a therapeutically relevant affinity, ketamine facilitates cell surface expression of TrkB and thus promotes BDNF/TrkB signaling mediated plasticity (Casarotto et al., 2021). Ketamine also inhibits the endocytosis of TrkB by disrupting the interaction of TrkB with the cargo-docking µ subunit of the AP-2 complex (Fred et al., 2019).

The effect of augmentation of BDNF/TrkB signaling by ketamine may not be limited in neuron cells. For instance, in C6 glioma cells, ketamine translocates Gα_s from lipid rafts to non-raft domains at clinically relevant concentrations (Wray et al., 2019). Ketamine-induced Gα_s redistribution leads to increased functional coupling of Gα_s and adenylyl cyclase (AC), and subsequent elevated intracellular cyclic adenosine monophosphate (cAMP) level. Increased cAMP level promotes phosphorylation of cAMP response element-binding protein (CREB), and subsequently upregulates BDNF expression in glia.

It has been hypothesized that spontaneous and evoked glutamatergic transmission are functionally segregated (Atasoy et al., 2008; Kavalali et al., 2011). At resting status, tonic neurotransmission of glutamate by spontaneous presynaptic vesicle release may activate NMDARs, and regulate synaptic plasticity and behavior.

Eukaryotic elongation factor 2 kinase (eEF2K), also known as calmodulin-dependent protein kinase III, belongs to the atypical alpha-kinase family. The activity of eEF2K is regulated by intracellular concentration of calcium and calmodulin. The primary downstream substrate of eEF2K is eEF2, and eEF2 pathway plays an important role in regulation of protein synthesis, synaptic plasticity and memory consolidation (Taha et al., 2013). Resting NMDAR activity causes sustained eEF2K activation, which phosphorylates eEF2, effectively halting the translation of BDNF. Acute NMDAR blockade by antidepressant dose of ketamine attenuates eEF2 phosphorylation, thus lifting the inhibition on BDNF translation (Autry et al., 2011). Antidepressant-like response to ketamine is abolished in eEF2 kinase knockout mice, and acute low dose of ketamine fails to augment BDNF protein level in hippocampus in these transgenic mice (Nosyreva et al., 2013).

A recent study also indicated that maintenance of basal level NMDAR function may be a key permissive factor required for the antidepressant-like effect of ketamine. Activation of NMDAR at rest (due to spontaneous/asynchronized vesicle release) and by glutamate surge (due to evoked/synchronized vesicle release generated by presynaptic action potential) may lead to distinct responses in postsynaptic cells. Reelin is a glycoprotein of the extracellular matrix, and is involved with cell migration, synaptogenesis and neural network activity modulation (Faini et al., 2021). Transgenic mouse models with deletion of Reelin or apolipoprotein E receptor 2 (ApoER2) abolished ketamine’s antidepressant-like action and compromised synaptic plasticity in the hippocampal CA1 region. Pharmacological manipulation indicated that Reelin-ApoER2-Src family kinases (SFK) pathway components may partially underlie the antidepressant-like action of ketamine (Kim et al., 2021b). NMDAR differentiate between.

With a ubiquitously expression pattern in eukaryotic cell types, mechanistic target of rapamycin (mTOR) is a large (259 kDa), highly conserved, serine/threonine kinase which belongs to the phosphoinositide 3-kinase-related kinase family (Sabatini et al., 1999). mTOR signaling is mediated through two large biochemical complexes defined by their protein composition, known as mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2), respectively (Lipton and Sahin, 2014). Studies have demonstrated that mTOR signaling is broadly involved in important cellular processes in both physiological and pathological conditions, which include translation, transcription, protein degradation, autophagy, cytoskeletal assembly and synaptic development and plasticity (Hoeffer and Klann, 2010; Lipton and Sahin, 2014).

In depressed patients, decreased volume of hippocampus and the prefrontal cortex (PFC) (subgenual and anterior cingulate cortex) are documented. Atrophy of principle neurons in these regions is reversible with successful treatment or discontinuation of stress exposure (MacQueen et al., 2008; MacQueen and Frodl, 2011; McEwen et al., 2015; McEwen et al., 2016). This phenomenon leads to the hypothesis that ketamine may alleviate depression by promoting synaptogenesis, thus antagonizing synaptic degeneration and atrophy of neurons induced by depression.

Ketamine administration increased the levels of the postsynaptic proteins PSD95 and GluR1, as well as the presynaptic protein synapsin I in the PFC of rats (Li et al., 2010). These effects lagged behind ketamine-induced activation of mTOR signaling and lasted for up to 72 h, suggesting that synaptogenesis and structural changes may mediate long-term antidepressant-like effect of ketamine. Blockade of mTOR signaling completely abolished ketamine-induced synaptogenesis and behavioral responses in animal models of depression, indicating that mTOR signaling may function as a mediator that transfers transient intracellular changes into sustained synapse strengthening by regulating local protein translation (Moda-Sava et al., 2019).

Recently, studies have provided some clues regarding the time course of neural structural remodeling triggered by ketamine. Within 2 h of administration, ketamine rapidly enhances the potential of activity-dependent synaptogenesis in the mPFC, and this effect precedes changes in spine density (Wu et al., 2021a). Within 12 h, ketamine selectively reverses stress-induced dendritic spine elimination in the mPFC, and the rescue of spinogenesis is necessary for the sustained antidepressant-like behavioral effects of ketamine (Moda-Sava et al., 2019). In another longitudinal study, two-photon microscopy imaging demonstrated that a single dose of ketamine increases dendritic spine density in the mPFC within 24 h after administration, and this effect lasts for up to 2 weeks (Phoumthipphavong et al., 2016). This is due to an elevated spine formation rate, whereas spine elimination rate remains unchanged.

Collectively, these evidences raise a possible scenario in which, initially after administration, ketamine enhances glutamate-evoked synaptic plasticity, which may lead to a transient and labile spinogenesis process. Subsequently, ketamine increases spine density by stabilizing newly generated spinogenesis and preventing stress-induced spine elimination simultaneously, thus exerting prolonged antidepressant-like effects.

α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid receptors (AMPARs) are one type of the principle ionotropic transmembrane receptors responsible for fast synaptic neurotransmission at glutamatergic excitatory terminals (Derkach et al., 2007). Studies revealed that pretreatment with AMPAR antagonist NBQX abolished the antidepressant-like action of ketamine in animal models of forced swim test, tail suspension test, learned helplessness, novelty-suppressed feeding test, and stress-induced sucrose preference (Koike et al., 2011; Walker et al., 2013; Zhou et al., 2014b; Fukumoto et al., 2014). On the other hand, AMPAR agonist Cx546 enhanced the antidepressant-like effects of ketamine in forced swim test (Koike and Chaki, 2014; Zanos and Gould, 2018). These studies collectively highlighted the involvement of AMPAR activation in ketamine’s antidepressant-like action.

Ketamine increases AMPARs containing GluA1 and GluA2 subunits on the membrane of the hippocampus within 3 h after administration (Nosyreva et al., 2013), increases the levels of the postsynaptic proteins PSD95 and GluA1 in the PFC of rats (Li et al., 2010), and increases AMPARs containing GluA1 and/or GluA2 subunits at synaptosome fractions in the mPFC and the hippocampus within 24 h post-injection (Li et al., 2010; Zanos et al., 2016). These newly recruited AMPARs in postsynaptic sites may be utilized for subsequent synaptic strengthening. Electrophysiological recordings demonstrated that low doses of ketamine induce an enhancement of AMPAR-mediated synaptic transmission in the mPFC and hippocampus (Bjorkholm et al., 2015; El Iskandrani et al., 2015). On the contrary, ketamine fails to generate antidepressant-like response in mice lacking the GluA2 gene, and ketamine-induced AMPAR-dependent synaptic potentiation of Schaffer collateral-CA1 synapses in hippocampal slices is abolished in GluA2 knockout animals (Nosyreva et al., 2013). Activation of mTOR signaling by ketamine is completely blocked by AMPAR antagonist NBQX.