Ketamine and Ketamine Metabolite Pharmacology: Insights into Therapeutic Mechanisms

A. Metabolism

Ketamine undergoes extensive metabolism (Fig. 1), initially via nitrogen demethylation to norketamine, a reaction that is catalyzed primarily by the cytochrome P450 liver enzymes CYP2B6 and CYP3A4 (Kharasch and Labroo, 1992; Yanagihara et al., 2001; Hijazi and Boulieu, 2002; Portmann et al., 2010; Mossner et al., 2011; Desta et al., 2012; Rao et al., 2016). The demethylation of ketamine occurs in a stereoselective manner, as CYP3A4 demethylates the (S)-ketamine enantiomer more rapidly than the (R)-ketamine enantiomer, whereas CYP2B6 demethylates both enantiomers of ketamine with near equal efficiency (Portmann et al., 2010). The individual variability in the metabolism of ketamine (Hijazi and Boulieu, 2002; Cheng et al., 2007; Desta et al., 2012) has been attributed, in part, to differences in the expression of P450 enzymes (Shimada et al., 1994; Hijazi and Boulieu, 2002).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Major metabolic pathways. In the predominant metabolic pathway, racemic ketamine [(R,S)-KET] is initially metabolized to norketamine [(R,S)-norKET], by either CYP2B6 or CYP3A4. Subsequently, norketamine can be further metabolized to form DHNK or the HNKs. Hydroxylation of norketamine at the six position by CYP2A6 results in (2R,6S;2S,6S)-hydroxynorketamine [(2R,6R;2S,6S)-HNK]. Alternatively, CYP2B6 or CYP2A6 can hydroxylate norketamine at the four position, resulting in the 4-hydroxy isomers. In the third case, CYP2B6 can hydroxylate norketamine at the five position, resulting in (2R,5S;2S,5R)-HNK and (2R,5R;2S,5S)-HNK. (R,S)-DHNK can result either from direct dehydrogenation from norketamine via CYP2B6 or via dehydration from either diastereomer of the 5-hydroxynorketamines via a nonbiologically catalyzed process.

Following demethylation of ketamine to norketamine, norketamine is further metabolized to the hydroxynorketamines (HNKs) and dehydronorketamine (DHNK) (Fig. 1). Early studies noted that the HNKs are formed through the hydroxylation of the cyclohexyl ring of norketamine at various locations (Adams et al., 1981). Several of these HNK metabolites have been detected in humans following ketamine infusion, with (2R,6R;2S,6S)-HNK and (2S,6R;2R,6S)-HNK being the predominant circulating HNKs in plasma (Moaddel et al., 2010; Zarate et al., 2012a). Metabolism to (2R,6R;2S,6S)-HNK is primarily carried out by CYP2A6 and CYP2B6 (Moaddel et al., 2010; Desta et al., 2012). These enzymes are also responsible for the formation of the (2S,4S;2R,4R)- and (2S,5S;2R,5R)-HNKs. CYP3A4 and CYP3A5 are the principal enzymes identified to catalyze the conversion of norketamine to (2S,4R;2R,4S)-HNK, whereas CYP2B6 is predominantly responsible for the catalysis of the conversion of norketamine to (2S,5R;2R,5S)-HNK (Desta et al., 2012). The other secondary metabolite is DHNK (Chang and Glazko, 1972; Adams et al., 1981). DHNK is directly formed from norketamine primarily via the actions of the CYP2B6 enzyme, or from 5-HNK via a nonenzymatic dehydration event (Adams et al., 1981; Bolze and Boulieu, 1998; Turfus et al., 2009; Portmann et al., 2010; Desta et al., 2012).

In addition to the major metabolic pathways of ketamine, there are several other pathways that have also been studied (Fig. 2). One of these pathways is the direct hydroxylation of ketamine to 6-hydroxyketamine (HK) (Woolf and Adams, 1987; Desta et al., 2012). Metabolism of ketamine to (2R,6R;2S,6S)-HK is primarily catalyzed by CYP2A6, whereas (2S,6R;2R,6S)-HK production is catalyzed by the CYP3A4 and CYP3A5 enzymes (Desta et al., 2012). The formation of (2R,6R;2S,6S)-HK is associated with greater hydroxylation of (S)-ketamine relative to (R)-ketamine, suggesting this reaction is enantioselective (Desta et al., 2012). The (2R,6R;2S,6S)-HK metabolite is readily demethylated via CYP2B6 to the corresponding HNK (Desta et al., 2012). However, analogous demethylation from the (2S,6R;2R,6S)-HK metabolite is reported to occur very slowly, with a modest contribution from CYP3A5. In addition to 6-HKs, evidence for the production of the 4-HK metabolite has been reported (Adams et al., 1981; Moaddel et al., 2010; Desta et al., 2012; Zarate et al., 2012a). Whereas hydroxylation of the phenyl ring was initially ruled out as being part of the metabolism of ketamine, more recent studies have provided evidence for the formation of such hydroxyphenyl ketamine metabolites via the actions of the CYP2C9 [primarily for the (R)-ketamine enantiomer] and flavin-containing mono-oxygenase enzymes [primarily for the (S)-ketamine enantiomer; Desta et al., 2012]. Lastly, phenolic isomers of HNKs have also been observed, potentially resulting from the hydroxylation of norketamine (Turfus et al., 2009).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Minor metabolic pathways. Although the majority of ketamine is metabolized via the major metabolic pathways (Fig. 1), there are several minor metabolic pathways, which provide unique, albeit low abundance, ketamine metabolites. The aryl ring of ketamine can be directly hydroxylated by flavin-containing mono-oxygenase enzymes or CYP2C9 to provide hydroxyphenyl-ketamine (hydroxyphenyl-KET). 4-Hydroxyketamine has also been observed; however, the metabolic enzymes responsible for this are currently unknown. CYP3A5 can directly hydroxylate ketamine at the six position to provide (2R,6S;2S,6R)-HK. Demethylation of (2R,6S;2S,6R)-HK with CYP3A5 provides (2R,6S;2S,6R)-HNK. CYP2A6 can also directly hydroxylate ketamine to provide (2R,6R;2S,6S)-HK, which is then transformed to (2R,6R;2S,6S)-HNK. Finally, norketamine can be hydroxylated via an unknown enzyme directly on the aryl rich to provide hydroxyphenyl-norketamine (hydroxyphenyl-norKET).

A population pharmacokinetic model was constructed for ketamine and its metabolites in patients suffering from treatment-resistant bipolar depression in a study that identified norketamine, DHNK, and (2R,6R;2S,6S)-HNK as the major circulating metabolites in plasma following a single 40-minute i.v. infusion of ketamine (0.5 mg/kg) (Zhao et al., 2012). These were also the major metabolites identified in plasma of patients suffering from unipolar or bipolar depression (Zarate et al., 2012a) or CRPS (Moaddel et al., 2010) and treated with ketamine. Specifically, norketamine, DHNK, and (2R,6R;2S,6S)-HNK were detected in the plasma of patients suffering from treatment-resistant unipolar and bipolar depression as early as 40 minutes after the end of i.v. ketamine administration (0.5 mg/kg delivered during a single 40-minute infusion; Zarate et al., 2012a; Zhao et al., 2012). The average time for metabolites to reach peak plasma concentration was estimated to be approximately 1.33 hours for both (R)- and (S)-norketamine and 3.83 hours for (R)-DHNK, (S)-DHNK, and (2R,6R;2S,6S)-HNK (Zhao et al., 2012). In plasma samples from these patients, the ratios of (S)- to (R)-ketamine, (S)- to (R)-norketamine, and (S)- to (R)-DHNK were 0.84, 1.0, and 0.67, respectively, during a 40- to 230-minute postinfusion period (Zhao et al., 2012). Similar to these findings, i.v. administration of ketamine (2 mg/kg) in surgical patients resulted in a plasma ratio of (S)- to (R)-ketamine of 0.91 (Geisslinger et al., 1993). Additionally, CRPS patients receiving continuous i.v. ketamine infusion at a dose of 40 mg/h over a total period of 5 days had plasma ratios of (S)- to (R)-ketamine, (S)- to (R)-norketamine, and (S)- to (R)-DHNK of 0.77, 0.71, and 0.71, respectively (Moaddel et al., 2010).

Following a 40-minute i.v. ketamine infusion at 0.5 mg/kg in patients diagnosed with treatment-resistant major depressive disorder, peak plasma concentrations were 204.13 ± 101.46 ng/ml or 0.86 ± 0.43 μM for ketamine (at 40 minutes), 73.54 ± 31.86 ng/ml or 0.33 ± 0.14 μM for norketamine (at 80 minutes), 13.27 ± 6.92 ng/ml or 0.06 ± 0.03 μM for DHNK (at 110 minutes), and 23.19 ± 11.88 ng/ml or 0.097 ± 0.05 μM for (2R,6R;2S,6S)-HNK (at 230 minutes) (Zarate et al., 2012a); see Table 2. In patients with treatment-resistant bipolar depression, peak plasma concentrations were 177.23 ± 53.8 ng/ml or 0.75 ± 0.23 μM for ketamine (at 40 minutes), 69.96 ± 19.98 ng/ml or 0.31 ± 0.09 μM for norketamine (at 80 minutes), 50.5 ± 27.44 ng/ml or 0.23 ± 0.12 μM for DHNK (at 110 minutes), and 37.59 ± 14.23 or 0.16 ± 0.06 μM for (2R,6R;2S,6S)-HNK (Zarate et al., 2012a; Table 2). In a patient with CRPS receiving chronic ketamine treatment (infusion beginning at 10 mg/h, titrated to 40 mg/h, and lasting 5 consecutive days), significant plasma levels of several HNK metabolites were detected, with (2R,6R;2S,6S)- and (2R,6R;2S,6S)-HNKs being the major metabolites present in samples obtained on day 3 (Moaddel et al., 2010).

In a study conducted by Cohen et al. (1973), brain concentrations of ketamine metabolites were measured following tail vein administration of ketamine (20 mg/kg) in rats. These authors showed that both ketamine and norketamine rapidly accumulated in the brain with peak concentrations achieved within 1 minute of administration (Cohen et al., 1973). Subsequently, it was demonstrated that (2R,6R;2S,6S)-HNK also accumulates in brain tissue shortly after dosing (Leung and Baillie, 1986; Paul et al., 2014; Moaddel et al., 2015b). Intravenous tail vein injection (2-minute infusion) of 20 mg/kg (S)- or (R)-ketamine to rats resulted in higher brain levels of (2S,6S)-HNK relative to (2R,6R)-HNK, respectively, with maximal concentrations of 769 ± 133 ng/g or 3.21 ± 0.55 μmol/kg at 20 minutes for (2S,6S)-HNK and 274 ± 47 ng/g or 1.14 ± 0.20 μmol/kg at 10 minutes for (2R,6R)-HNK (Moaddel et al., 2015b; Table 2). It was hypothesized that the difference was due to a passive uptake process of these metabolites into the brain (Moaddel et al., 2015b). A ∼1:1 ratio for the plasma:brain levels of the corresponding (2R,6R;2S,6S)-HNK was observed, indicating that blood-brain barrier penetration or the central nervous system transport process was not mediated by an enantioselective process (Leung and Baillie, 1986; Moaddel et al., 2015b). Importantly, no in situ metabolism was observed when ketamine was incubated with rat brain microsomes (S9 fraction) (Moaddel et al., 2015b). Likewise, ketamine metabolites were below detectable levels in the brain of mice following in vivo intracerebroventricular administration of ketamine (P.Z., R.M., J.N.H., T.D.G., unpublished data), a finding that suggests that local ketamine metabolism does not occur in the brain.

In mice, norketamine, DHNK, and (2R,6R;2S,6S)-HNK metabolites were detected in plasma within 10 minutes of i.p. administration of 10 mg/kg ketamine (Can et al., 2016; Zanos et al., 2016). The maximum plasma concentrations were 561.89 ± 86.09 ng/ml or 2.36 ± 0.18 μM at 10 minutes for ketamine, 1098.89 ± 216.89 ng/ml or 4.91 ± 0.97 μM at 10 minutes for norketamine, 83.92 ± 53.63 ng/ml or 0.38 ± 0.24 μM at 30 minutes for DHNK, and 674.59 ± 278.23 ng/ml or 2.81 ± 1.16 μM at 10 minutes for (2R,6R;2S,6S)-HNK (Zanos et al., 2016), as summarized in Table 2. In the brain, ketamine (1162.34 ± 202.05 ng/g or 4.89 ± 0.85 μmol/kg tissue), norketamine (450.94 ± 199.7 ng/g or 2.02 ± 0.89 μmol/kg tissue), and (2R,6R;2S,6S)-HNK (498.35 ± 50.99 ng/g or 2.08 ± 0.21 μmol/kg tissue) were detected within 10 minutes of ketamine administration (Zanos et al., 2016). The maximum brain concentration of ketamine was 51.66% higher than the corresponding plasma concentration, whereas the brain tissue concentrations of norketamine and (2R,6R;2S,6S)-HNK were 58.96% and 26.13% lower than the corresponding maximum plasma concentrations, respectively (Zanos et al., 2016). Levels of DHNK in brain tissue were below the limits of quantification, consistent with the findings that DHNK partitions into red blood cells (Moaddel et al., 2016) and has poor penetration of the blood-brain barrier (Can et al., 2016).

B. Absorption

Ketamine is administered to humans via multiple routes, including i.v., i.m., oral, intranasal, epidural, and intrarectal (Malinovsky et al., 1996; Andrade, 2017b). The most typical route of administration is via i.v. infusion, which rapidly attains maximum plasma concentrations (e.g., Clements et al., 1982; Weber et al., 2004). Intramuscular administration, which is used in emergency cases of uncooperative patients, neonates, and children, has high bioavailability of 93%, with peak plasma concentrations achieved within 5–30 minutes of administration (e.g., Clements et al., 1982); however, a population pharmacokinetic analysis reported a much lower bioavailability following i.m. administration of ketamine in children (41%; Hornik et al., 2013). In contrast, oral bioavailability of ketamine is limited to 16%–29%, with peak concentration levels of the drug occurring within 20–120 minutes (Grant et al., 1981; Clements et al., 1982; Sekerci et al., 1996; Chong et al., 2009; Rolan et al., 2014; Karch and Drummer, 2015), due to extensive first-pass hepatic metabolism (e.g., Kharasch and Labroo, 1992; Yanagihara et al., 2003). Oral bioavailability of (S)-ketamine was calculated to be 8%–11% (Peltoniemi et al., 2012; Fanta et al., 2015), consistent with the greater first-pass metabolism of (S)-ketamine relative to (R,S)-ketamine. Intranasal and intrarectal ketamine bioavailability is 45%–50% and 25%–30%, respectively (Malinovsky et al., 1996; Yanagihara et al., 2003). Intranasal administration is considered an attractive alternative to the i.v. administration of ketamine because it is less invasive, results in rapid systemic absorption, and is not subject to first-pass hepatic metabolism (Malinovsky et al., 1996).

Following oral administration of (2S,6S)-HNK in rats (20 mg/kg), maximum plasma concentrations were reached at 0.4 ± 0.1 hour. Oral bioavailability of (2S,6S)-HNK was estimated to be 46.3% in rats (Moaddel et al., 2015b). In mice, the oral bioavailability of (2R,6R)-HNK is estimated to be approximately 50% at the dose of 50 mg/kg (P.Z., R.M., J.N.H., T.D.G., unpublished data). The oral bioavailability of other ketamine metabolites remains to be determined.

C. Distribution

Ketamine is rapidly distributed into highly perfused tissues, including the brain, and has a plasma protein binding between 10% and 50% (Wieber et al., 1975; Dayton et al., 1983; Sinner and Graf, 2008; Peltoniemi et al., 2012, 2016; Karch and Drummer, 2015), resulting in a large steady-state volume of distribution (V_d = 3–5 l/kg; Karch and Drummer, 2015). A single i.v. bolus administration of an anesthetic dose of racemic ketamine in humans (2 mg/kg) leads to equal plasma concentrations of (S)-ketamine and (R)-ketamine 1 minute postadministration (C_max = ∼1800 ng/ml or 7.6 µM—estimated from Geisslinger et al., 1993). However, i.v. (bolus) administration of 1 mg/kg (S)-ketamine resulted in a higher plasma concentration of the drug 1 minute postinfusion (C_max = ∼2600 ng/ml: 11 µM—estimated from Geisslinger et al., 1993). These results are particularly important when comparing the outcomes of (S)-ketamine with those of the racemic ketamine or (R)-ketamine, because lower doses of (S)-ketamine are required to produce similar or greater ketamine concentrations in the plasma (e.g., White et al., 1985; Mathisen et al., 1995). Notably, there is no interconversion between (S)- and (R)-ketamine, because administration of (S)-ketamine does not result in the formation of (R)-ketamine in vivo, and vice versa (Geisslinger et al., 1993; Ihmsen et al., 2001). Plasma of patients suffering from treatment-resistant bipolar depression, who were treated with a 40-minute i.v. infusion of 0.5 mg/kg (R,S)-ketamine, had a ratio of (S)- to (R)-ketamine of 0.84 (Zhao et al., 2012), with peak ketamine concentrations of 177.23 ± 53.8 ng/ml or 0.75 ± 0.23 μM (Zarate et al., 2012a).

In mice, administration of subanesthetic doses of either (S)- or (R)-ketamine (10 mg/kg, i.p.) resulted in similar brain levels of both drugs [area under the curve (AUC)_last = 483.1 hours.ng/ml or 2.03 hours.μmol/kg versus 591.9 hours.ng/ml or 2.48 hours.μmol/kg, respectively], with peak levels being C_max = 1743 ± 560.6 ng/g or 7.33 ± 2.36 μmol/kg for (S)-ketamine and 1886 ± 459.6 ng/ml or 7.93 ± 1.93 for (R)-ketamine at 10 minutes postinjection (Zanos et al., 2016). Similarly, there were no differences in (S)-ketamine (C_max = 2732 ± 535 ng/ml or 11.49 ± 2.25 μM) and (R)-ketamine (C_max = 3430 ± 400 ng/ml or 14.4 ± 1.68 μM) levels in the plasma of rats 10 minutes following an i.v. administration of 20 mg/kg of each of these enantiomers (Moaddel et al., 2015b).

Direct i.p. administration of (2R,6R)-HNK and (2S,6S)-HNK in mice results in a 1:1 ratio between circulating plasma and brain tissue concentrations (Table 2), with higher total levels (AUC_last) of (2S,6S)-HNK observed in plasma and brain tissue compared with (2R,6R)-HNK (brain: 7.55 versus 3.05 h.μmol/kg; plasma: 11.60 versus 3.22 h.μM; Zanos et al., 2016; Table 2). Following i.v. administration of (2S,6S)-HNK in rats (20 mg/kg), total drug exposure was calculated as AUC_last = 28,981 ± 6162 h.ng/ml or 120.91 ± 25.71 h.μM, with a volume of distribution V_d = 7.35 ± 0.74 l/kg (Moaddel et al., 2015b). Following oral administration of (2S,6S)-HNK to rats (20 mg/kg), total drug exposure was AUC_last = 10,120 ± 1313 h.ng/ml or 42.22 ± 5.48 h.μM (Moaddel et al., 2015b).

D. Elimination

Although plasma levels of ketamine are below detectable limits within 1 day following an i.v. antidepressant dose of ketamine (0.5 mg/kg administered over a 40-minute infusion), circulating levels of DHNK and (2R,6R;2S,6S)-HNK were observed for up to 3 days after ketamine infusion in patients diagnosed with bipolar depression (Zhao et al., 2012) or treatment-resistant major depression (Zarate et al., 2012a). Norketamine and ketamine were detectable for up to 14 and 11 days, respectively, in the urine of children who received anesthetic doses of ketamine, with reported concentrations of 0.1–1442 ng/ml (or 0.0004–0.031 μΜ) for norketamine and 2–1204 ng/ml (or 0.008–5.06 μM) for ketamine (Adamowicz and Kala, 2005).

In adult humans, ketamine has a high rate of clearance and a short elimination half-life (2–4 hours; Clements et al., 1982; White et al., 1985; Domino, 2010). White et al. (1985) also demonstrated a short elimination half-life (155–158 minutes) for both (S)-ketamine and (R)-ketamine. Elimination of ketamine is primarily performed by the kidneys, with low levels excreted as ketamine (2%), norketamine (2%), and DHNK (16%) (Haas and Harper, 1992; Lin and Lua, 2004; Adamowicz and Kala, 2005; Karch and Drummer, 2015; Dinis-Oliveira, 2017). The majority of the drug (∼80%) is excreted as the glucuronic acid-labile conjugates of HK and HNK (Dinis-Oliveira, 2017), which are eliminated in urine and bile (Chang and Glazko, 1974).

In adult humans, terminal plasma half-life and the clearance rates of ketamine do not significantly differ between i.v. (half-life = 186 minutes; total body clearance = 19.1 ml/min per kilogram) and intramuscular (half-life = 155 minute; total body clearance = 23.2 ml/min per kilogram) routes of administration (Clements et al., 1982). However, there is evidence that repeated administration of ketamine prolongs its elimination time. For example, Adamowicz and Kala (2005) reported that, among three instances of single i.v. infusions of ketamine during a 2-year period (doses ranged from 0.75 to 1.59 mg/kg), the elimination of ketamine was slowed from 2 days following the first infusion to 5 days after the second, and 11 days following the third. Elimination of norketamine remained constant (i.e., 5 days after each infusion; Adamowicz and Kala, 2005).

When compared with adults, ketamine is eliminated approximately twice as fast in children (Haas and Harper, 1992). This is in accordance with evidence supporting a longer duration of anesthesia in adults relative to children following i.m. administration of 6 mg/kg ketamine (Grant et al., 1981, 1983; Akin et al., 2005). Moreover, a negative correlation between age and ketamine dose per body weight required for anesthesia was reported in children (Lockhart and Nelson, 1974). These differences might be due to differences in the enzymatic metabolism of ketamine in children, as compared with adults (Edginton et al., 2006).

In humans, (S)-ketamine has a slightly longer elimination half-life than racemic ketamine [∼5 hours for (S)-ketamine versus 2–4 hours for racemic ketamine; Hagelberg et al., 2010; Peltoniemi et al., 2012], and its systemic clearance is faster when administered alone than when administered in the racemic mixture [26.3 ± 3.5 ml/kg per minute for (S)-ketamine versus 14.8 ± 1.7 ml/kg per minute when administered as the racemic ketamine; Ihmsen et al., 2001]. This may suggest an inhibition of (S)-ketamine’s clearance by the (R)-ketamine enantiomer when the racemic mixture is administered (Kharasch and Labroo, 1992). Such inhibition could contribute to the prolonged awakening time in patients receiving racemic ketamine relative to those receiving (S)-ketamine (White et al., 1985). We note that systemic clearance of (R)-ketamine following racemic ketamine administration is 13.8 ± 1.3 ml/kg per minute, which is similar to the (S)-ketamine enantiomer (14.8 ± 1.7 ml/kg per minute; Ihmsen et al., 2001).

Following i.v. administration of (2S,6S)-HNK in rats (20 mg/kg), the clearance rate was calculated to be 704 ± 139 ml/kg per hour, with an elimination half-life of 8.0 ± 4.0 hours. Oral administration of this metabolite resulted in an elimination half-life of 3.8 ± 0.6 hours (Moaddel et al., 2015b).

Overall, it is important to note that there are important species differences in regard to half-life values, AUCs, C_max, and clearance rates of ketamine and its metabolites (see Table 2; also see Zarate et al., 2012a; Zanos et al., 2017c). This should be taken into consideration when comparing the behavioral actions of specific dose regimens for ketamine and its metabolites in mice, rats, and humans. Nevertheless, the brain levels of ketamine and its metabolites following administration of ketamine in humans are not known, and, therefore, direct comparisons are not straightforward.

A. N-Methyl-D-Aspartate Receptors

Historically, the primary recognized receptor target of ketamine is the NMDAR, in which ketamine acts as a noncompetitive open-channel blocker (Lodge et al., 1982; Anis et al., 1983; MacDonald et al., 1987). NMDARs are glutamatergic ion channels made of different combinations of four subunits encoded by one of seven genes: GluN1, GluN2A–D, and GluN3A–B (Vyklicky et al., 2014). NMDARs are highly permeable to calcium ions, which can trigger the activation of a number of intracellular pathways in neurons and glial cells. At resting state, NMDAR channels are tonically blocked by magnesium (Mg²⁺). Efficient receptor activation requires the following: 1) membrane depolarization, which displaces the Mg²⁺ block, and 2) binding of both glutamate and the coactivator glycine and/or D-serine (Paoletti et al., 2013).

Ketamine was initially characterized as a NMDAR antagonist by David Lodge and colleagues (Lodge et al., 1982; Anis et al., 1983), a finding that was subsequently confirmed by other investigators (Harrison and Simmonds, 1985; Thomson et al., 1985). Ketamine binds to the allosteric phencyclidine (PCP) site that is located within the channel pore of the NMDAR, and thus it blocks the receptor noncompetitively (Kohrs and Durieux, 1998; Mion and Villevieille, 2013). Ketamine has a relatively high (∼86%) trapping capability (binding within the ion channel pore following closure of the channel) to block NMDARs, via binding to the same site as PCP (>98% trapping) and MK-801 (100% trapping; Huettner and Bean, 1988; Lerma et al., 1991; MacDonald et al., 1991; Jahr, 1992; Orser et al., 1997). The binding affinity of ketamine to the PCP binding site has been reported to be between 0.18 and 3.1 μM in the presence of Mg²⁺ (Table 3; Wong et al., 1986, 1988; MacDonald et al., 1987; Kornhuber et al., 1989; Reynolds and Miller, 1989; Sharif et al., 1991; Bresink et al., 1995; Lynch et al., 1995; Parsons et al., 1995; Kapur and Seeman, 2001, 2002; Sun and Wessinger, 2004; Seeman et al., 2005; Gilling et al., 2009; Moaddel et al., 2013; Bonifazi et al., 2015; Wallach et al., 2016; Kang et al., 2017; Morris et al., 2017).

NMDAR blockade is thought to underlie the dissociative anesthetic and amnesic effects of ketamine, as well as the antidepressant, analgesic, and altered psychotomimetic effects induced by the drug (White et al., 1980; Oye et al., 1992; Yeung et al., 2010; Li et al., 2010, Autry et al., 2011; Miller et al., 2014). Ketamine-induced cognitive deficits are also hypothesized to be due to NMDAR inhibition (Shaffer et al., 2014). (S)-ketamine has an approximately fourfold higher affinity/potency for the PCP site of the NMDAR compared with the (R)-isomer, and twice that of the racemic mixture [(S)-ketamine: K_i = 0.3–0.69 μM; (R)-ketamine: K_i = 1.4–2.57 μM; and (R,S)-ketamine: K_i = 0.18–3.1 μM, in the presence of extracellular Mg²⁺ (Ebert et al., 1997; Kohrs and Durieux, 1998; Moaddel et al., 2013; Zanos et al., 2016; Temme et al., 2018)]. The effects of (S)-ketamine and (R)-ketamine were also assessed on NMDA receptor-activated cation currents of whole-cell voltage-clamped cultured rat hippocampal neurons (Zeilhofer et al., 1992). These authors showed that both enantiomers exhibited voltage- and use-dependent blockades of NMDAR currents, with (S)-ketamine being about twice as potent compared with (R)-ketamine (IC₅₀ = 0.80 versus 1.53 μM, respectively; Zeilhofer et al., 1992). Moreover, (S)-ketamine has 2.5–3 times higher potency to inhibit NMDA-evoked currents in cat dorsal horn neurons compared with the (R)-ketamine enantiomer (Lodge et al., 1982). This higher affinity/potency of the (S)-ketamine isomer is hypothesized to explain why (S)-ketamine is a more potent anesthetic than (R,S)-ketamine (Yamakura and Shimoji, 1999). Consistent with these stereospecific differential potencies to inhibit the NMDAR by ketamine's isomers, the ED₅₀ value for induction of hypnosis (loss of righting reflex) was lower for (S)-ketamine and (R,S)-ketamine (3.5 and 5.6 mg/kg, respectively) compared with (R)-ketamine (10.3 mg/kg; Marietta et al., 1977). Similarly, Ryder et al. (1978) showed that (S)-ketamine is an ∼3 times more potent analgesic, 1.5 times more potent hypnotic (loss of righting reflex) and 1.8 times more potent locomotor stimulant agent compared with (R)-ketamine. In particular, the median effective analgesic (s.c.) doses were found to be 6.5, 3.7 and 11 mg/kg for (R,S)-ketamine, (S)-ketamine and (R)-ketamine, respectively (Ryder et al., 1978). The median hypnotic doses for (R,S)-ketamine, (S)-ketamine and (R)-ketamine were calculated to be 45, 38 and 56 mg/kg, respectively (Ryder et al., 1978). In addition, (S)-ketamine (25 mg/kg, s.c.) induced a more profound disruption in sensorimotor gating compared with the (R)-ketamine (25 mg/kg, s.c.) enantiomer in the rat pre-pulse inhibition paradigm, although (R)-ketamine also showed a subtle effect in this study compared with the control-treated rats (Littlewood et al., 2006). In agreement with this finding, Yang et al. (2015) showed disruption of sensorimotor gating and hyperlocomotion to only occur from administration of (S)-ketamine, but not (R)-ketamine in mice. Subanesthetic concentrations of ketamine (40-minute i.v. infusion; 0.5 mg/kg), which exert antidepressant actions in patients suffering from major depression (Zarate et al., 2012a), resulted in a maximum of 31% ± 18% NMDAR occupancy (Shaffer et al., 2014). This occupancy is similar to the NMDAR occupancy estimated (32% ± 6% maximum; Shaffer et al., 2014) following an antidepressant-relevant dose of ketamine in rats (10 mg/kg, i.p.; Yeung et al., 2010). Nevertheless, (R)-ketamine was reported to be a more potent and longer-lasting antidepressant compared with the (S)-ketamine enantiomer in several rodent models (Zhang et al., 2014; Yang et al., 2015; Zanos et al., 2016; Fukumoto et al., 2017), when using a 30-fold dose range (Zanos et al., 2016). There do not appear to be differences in brain exposure of the two enantiomers (Zanos et al., 2016; Fukumoto et al., 2017), thus challenging the NMDAR inhibition hypothesis as the sole mediator of the antidepressant actions of ketamine.

In membrane fractions of postmortem human brain homogenates, IC₅₀ values for [³H]MK-801 displacement by (S)- and (R)-ketamine were reported to be 1.6–1.9 and 7.2–10 μM, respectively, in the presence of extracellular Mg²⁺ (Oye et al., 1992). Similarly, in rat cortical tissue (S)-ketamine inhibited NMDA (10 μΜ)-evoked currents with an IC₅₀ of 0.9 ± 1.4 μM, whereas (R)-ketamine was a less potent inhibitor with an IC₅₀ of 3.0 ± 1.4 μM (Ebert et al., 1997). Whole-cell patch-clamp electrophysiological recordings obtained from human embryonic kidney (HEK)293T cells transfected with different NMDAR subunits revealed that, in the absence of extracellular Mg²⁺, ketamine inhibits the NMDARs containing GluN1/GluN2A (IC₅₀ = 0.33 ± 0.01 μM) and GluN1/GluN2B (IC₅₀ = 0.31 ± 0.02 μM) subunit compositions with a modestly higher potency than GluN1/GluN2C (IC₅₀ = 0.51 ± 0.01 μM) and GluN1/GluN2D (IC₅₀ = 0.83 ± 0.02 μM) subunits (Kotermanski and Johnson, 2009). In contrast, in the presence of physiologic levels of Mg²⁺ (1 mM), ketamine blocks NMDAR containing GluN1/GluN2C (IC₅₀ = 1.18 ± 0.0 μM) and GluN1/GluN2D (IC₅₀ = 2.95 ± 0.02 μM) subunits, with a higher potency than the GluN1/GluN2A (IC₅₀ = 5.35 ± 0.34 μM) and GluN1/GluN2B (IC₅₀ = 5.08 ± 0.02 μM) subunits (Kotermanski and Johnson, 2009). Nevertheless, Yamakura et al. (1993) failed to identify differences in ketamine-induced inhibition of the different NMDAR receptor subunits in Xenopus oocytes injected with subunit-specific mRNAs synthesized in vitro. These findings highlight a lack of clarity on any differential effects of ketamine on NMDAR subtypes composed of different subunits.

Studies have shown that (S)-ketamine inhibits NMDARs composed of GluN1/GluN2C (IC₅₀ = 1.11 μM) and GluN1/GluN2D (IC₅₀ = 1.50 μM) with higher potency than those composed of GluN1/GluN2A (IC₅₀ = 16.10 μM) in the presence of 2 mM Mg²⁺ (Dravid et al., 2007). (S)-ketamine’s potency to inhibit GluN1/GluN2B (IC₅₀ = 1.55 μM) is reported to be similar to its potency to inhibit GluN1/GluN2C- and GluN1/GluN2D-containing NMDARs in the presence of 2 mM Mg²⁺ (Dravid et al., 2007). These findings indicate that any preferential potency of ketamine is likely not the result of higher affinity of ketamine to bind to the GluN2C-NMDARs per se, but may be due to differential capacity for Mg²⁺ binding, or interactions between the drug and Mg²⁺ within the channel (Kotermanski and Johnson, 2009; Kotermanski et al., 2009). Thus, ketamine may differentially block specific NMDAR subtypes in the brain depending upon local Mg²⁺ concentrations. In support of this concept, in the absence of Mg²⁺, ketamine blocks GluN2B-containing NMDARs with a higher potency compared with the NMDARs containing other GluN2 subunits, as measured using recombinant NMDAR GluN2A–D subunits expressed in Xenopus oocytes (Dravid et al., 2007).

In the presence of extracellular Mg²⁺, ketamine’s N-demethylated metabolite, norketamine, also inhibits the NMDAR. (S)-norketamine has a reported K_i of 1.70–2.25 μM for NMDARs in the spinal cord and the cerebral cortex, whereas (R)-norketamine has an approximately eight times lower binding affinity (K_i = 13.0–26.46 μM; Ebert et al., 1997; Moaddel et al., 2013); also see Table 3. In accordance with these findings, (S)-norketamine (IC₅₀ = 3.0 ± 0.8 μM) more potently inhibited NMDA (10 μΜ)-evoked currents than (R)-norketamine (IC₅₀ = 39 ± 1.4 μM) in rat cerebral cortical neurons (Ebert et al., 1997). Therefore, because NMDAR inhibition was considered the primary mechanism of action of ketamine, the clinical effects of the drug were initially attributed to ketamine and norketamine (Leung and Baillie, 1986; Hirota and Lambert, 2011; Singh et al., 2014).

DHNK and HNK metabolites display weak or no ability to displace [³H]MK-801 binding to NMDARs. (R)-DHNK has lower affinity than (S)-DHNK (59.7–74.6 and 39.0–42.0 μM, respectively) for displacing [³H]MK-801 binding to the NMDAR (Moaddel et al., 2013; Morris et al., 2017). (2S,6S)-HNK has a K_i = 10.4–21.0 μM for displacing [³H]MK-801 binding, whereas (2R,6R)-HNK does not bind to the NMDAR-PCP site with appreciable affinity (K_i > 100 μM; Moaddel et al., 2013; Morris et al., 2017). In addition, at concentrations up to 10 μM, neither (2S,6S)-HNK nor (2R,6R)-HNK functionally inhibit NMDA-evoked currents in rat hippocampal interneurons (Zanos et al., 2016). Lack of functional NMDAR inhibition by (2R,6R)-HNK at 10 μM was also reported by Suzuki et al. (2017). At a higher concentration (50 μM), (2R,6R)-HNK moderately (∼40%) inhibited NMDAR-mediated miniature excitatory postsynaptic currents recorded from cultured hippocampal neurons in the absence of Mg²⁺. This finding supported the contention that, at concentrations higher than those relevant to antidepressant treatment and in the absence of Mg²⁺, (2R,6R)-HNK might functionally inhibit NMDARs (Suzuki et al., 2017; Zanos et al., 2017a). Notably, at the same concentration (50 μM) and under the same experimental conditions, ketamine induced >90% inhibition of NMDAR-mediated miniature excitatory postsynaptic currents recorded from hippocampal neurons (Suzuki et al., 2017). (2R,6S)-, (2S,6R)-, (2R,5R)-, (2S,5S)-, (2S,5S)-, (2R,5S)-, (2S,5R)-, (2R,4S)-, (2S,4R)-, (2R,4R)-, and (2S,4S)-HNKs do not have significant affinity to displace [³H]MK-801 binding (K_i > 100 μM; Morris et al., 2017).

There is also evidence that, by reducing extracellular levels of D-serine, ketamine’s enantiomers and its metabolites may indirectly decrease the activation of NMDARs (Singh et al., 2013). D-serine, an endogenous NMDAR coagonist that binds to the glycine_B site, is required for activation of the NMDAR complex (Paoletti et al., 2013) and is produced by enzymatic L-serine enantioconversion catalyzed by serine racemase (Wolosker et al., 2008). Incubation of PC-12 cells with increasing concentrations of (S)- and (R)-ketamine exerted differential effects on the intracellular and extracellular D-serine levels. Specifically, application of (S)-ketamine was associated with increased intracellular D-serine (EC₅₀ = 0.82 ± 0.29 μM) and decreased extracellular levels of D-serine (IC₅₀ = 0.82 ± 0.29 μM; Singh et al., 2015). In contrast, (R)-ketamine decreased both intracellular (IC₅₀ = 0.94 ± 0.16 μM) and extracellular levels of D-serine (IC₅₀ = 0.70 ± 0.10 μM; Singh et al., 2015; Table 3). Similar findings were observed using 1321N1 cells and primary hippocampal and cortical neuronal cells. Singh et al. (2015) also demonstrated that inhibition of the amino acid transporter, ASCT2, resulted in qualitatively similar effects to those induced by (S)-ketamine on D-serine levels. In addition, coincubation with sub-saturating concentrations of an ASCT2 inhibitor and (S)-ketamine resulted in an additive effect in both PC-12 cells and primary neuronal cells in regard to D-serine levels, indicating that the effects of (S)-ketamine might be due to an inhibition of the amino acid transporter systems.

The differential effects of ketamine’s enantiomers on D-serine levels might contribute to their differential behavioral effects. Indeed, whereas (S)-ketamine is a more potent anesthetic and analgesic drug (Marietta et al., 1977; White et al., 1985) than (R)-ketamine, (R)-ketamine is a more potent and longer-lasting antidepressant than (S)-ketamine in several animal tests (Zhang et al., 2014; Yang et al., 2015; Zanos et al., 2016; Fukumoto et al., 2017). In fact, D-serine plays a role in synaptic plasticity (Henneberger et al., 2010), and baseline plasma D-serine levels are negatively correlated with ketamine treatment response in patients suffering from major depression (Moaddel et al., 2015a), indicating a possible role of D-serine levels in the antidepressant responses of ketamine (also see Hashimoto, 2014). In vivo, sub-chronic (14-day) administration of ketamine to rats was shown to reduce serine racemase mRNA levels in the forebrain (Watanabe et al., 2010). However, a single administration of ketamine at the dose of 50 mg/kg resulted in an enhancement of serine racemase mRNA levels in the striatum, hippocampus, and cortex of rats (Takeyama et al., 2006), an effect that is predicted to induce an increase rather than a decrease in D-serine levels. Indeed, a single administration of (R)-ketamine (10 mg/kg, i.p.) slightly, but significantly increased cortical D-serine/L-serine ratio in mice (Ma et al., 2017). Therefore, further in vivo confirmation of the effects of ketamine and its enantiomers on D-serine levels is warranted.

DHNK has also been shown to modify D-serine levels. Singh et al. (2013) demonstrated that incubation of PC-12 and 1321N1 cells with 5–90 nM DHNK decreased the relative intracellular D-serine concentrations. Because DHNK is not produced in the brain and does not cross the blood-brain barrier in ketamine-treated rodents (Can et al., 2016; Moaddel et al., 2016), the behavioral relevance of this metabolite’s actions on D-serine levels is not clear (Zanos et al., 2016).

HNKs are also capable of reducing intracellular D-serine concentrations in PC-12 cells, with (2S,6S)-HNK being more potent than (2R,6R)-HNK (IC₅₀s are reported to be 0.18 ± 0.04 and 0.68 ± 0.09 nM, respectively; Singh et al., 2016c). It is possible that the HNK-induced reduction of intracellular D-serine levels may also result in a reduction of extracellular levels of this amino acid. However, modulation of extracellular levels of D-serine may not be an important determinant of the antidepressant effects of HNKs, because at least in mice, (2R,6R)-HNK exerts more potent antidepressant actions than (2S,6S)-HNK (Zanos et al., 2016) and DHNK (Sałat et al., 2015). In addition, electrophysiological studies failed to identify any inhibitory effects of antidepressant-relevant concentrations of these metabolites on NMDAR function (Zanos et al., 2016, 2017b; Suzuki et al., 2017), as decreased extracellular D-serine levels would predict. Moreover, acute D-serine administration induces ketamine-like antidepressant behavioral and biochemical responses in rats (Wei et al., 2017), further complicating the possible functional role of decreased extracellular D-serine levels following ketamine administration. Further verification, possibly using human brain-derived cells or measuring extracellular D-serine levels in vivo following administration of ketamine and/or its metabolites, would be informative in determining the functional relevance of these results.

B. Hyperpolarization-Activated Cyclic Nucleotide-Gated Channels

HCN channels are voltage-gated cation channels (HCN1–HCN4; Luthi and McCormick, 1998; Wahl-Schott and Biel, 2009). Activation of these channels by membrane hyperpolarization is facilitated by cyclic nucleotides, including cAMP. In the central nervous system, HCN channels play a major role in controlling neuronal excitability, synaptic activity, and rhythmic oscillations (Shah, 2014).

There is a report of subunit-specific inhibitory effects of ketamine (EC₅₀ = 8.2–15.6 μM) on HCN1–HCN2 heteromeric channels and hyperpolarization-activated pacemaker currents (I_h; Chen et al., 2009). This may be relevant to the anesthetic actions of ketamine, as ketamine-induced anesthesia was significantly suppressed in HCN knockout mice (Chen et al., 2009). Additionally, (S)-ketamine was found to be more potent at inhibiting these channels (EC₅₀ = 4.1–7.4 μM) compared with racemic ketamine (EC₅₀ = 8.2–15.6 μM; Chen et al., 2009), concordant with the greater anesthetic potency of (S)-ketamine. Indeed, it has been hypothesized that NMDAR inhibition is not the sole mechanism underlying the anesthetic properties of ketamine (Petrenko et al., 2014). Further studies are required to substantiate the exact role of HCN channel inhibition in this regard, and to replicate these findings.

In addition to a possible role in the anesthetic properties of ketamine, HCN1 channel inhibition may have a role in ketamine’s antidepressant actions because reduced HCN1 activity in the hippocampus has been associated with antidepressant effects in rodents (Lewis et al., 2011; Kim et al., 2012; Han et al., 2017). Of interest, mice lacking the HCN1 gene did not manifest ketamine-induced reductions in immobility time in the forced-swim test following chronic oral corticosterone treatment (Li et al., 2014). Furthermore, following ketamine administration, these mice did not show increased sucrose preference or decreased latency to feed in the novelty-suppressed feeding test (Zhang et al., 2016). Unfortunately, these results cannot be unambiguously interpreted as evidence that HCN1 mediates the antidepressant effects of ketamine because HCN1 deletion by itself induced baseline behavioral changes compatible with reduced depressive-like behavior (e.g., decreased immobility in the forced-swim test). In addition, Zhang et al. (2016) suggested that reduction of HCN1 function by ketamine was secondary to inhibition of presynaptic NMDARs. The authors did not test the hypothesis that direct inhibition of HCN1 by ketamine, as was suggested by Petrenko et al. (2014), accounts for its antidepressant effects. There are currently no published data on the activity of ketamine’s metabolites on the function of HCN1 channels or the involvement of these channels on the behavioral effects of these metabolites.

C. GABA Uptake and GABA Receptors

The primary inhibitory neurotransmitter, GABA, activates both the ionotropic GABA receptor subtypes A and C (GABA_A and GABA_C) and the metabotropic GABA receptor subtype B (GABA_B) in the brain (Jacob et al., 2008). Electrophysiological studies have revealed that high concentrations of ketamine potentiate GABAergic inhibitory postsynaptic currents in neurons of guinea pig olfactory cortical slices (300 μM; Scholfield, 1980) and of rat hippocampal slices (500 μM; Gage and Robertson, 1985). At high concentrations, ketamine potentiates GABA-activated GABA_A receptors ectopically expressed in Xenopus oocytes (365 μM; Lin et al., 1992) and HEK293 cells (>500 μM; EC₅₀ = 1.2 mM; Flood and Krasowski, 2000; but see Anis et al., 1983). There is also evidence for an inhibitory effect of ketamine on GABA uptake (K_i = 6.2 ± 1.1; IC₅₀ = 50 μM), as assessed by the [³H]GABA-binding assay in striatal synaptosomes of rats (Mantz et al., 1995), indicating that ketamine might cause an increase in extracellular GABA levels. Indeed, it was shown that intramuscular administration of ketamine to rats increases GABA content in the brain and in synaptosomal-enriched fractions of the brain (Wood and Hertz, 1980). However, the effect of ketamine on GABA uptake was not found at clinically relevant concentrations in mice (IC₅₀ > 1000 μM; Wood and Hertz, 1980).

The functional relevance of the actions of ketamine on GABA_A receptors is not clear because ketamine concentrations required to modify the activity of these receptors are much higher than those achieved following ketamine administration in clinical settings for anesthetic, analgesic, anti-inflammatory, and antidepressant effects. Nevertheless, there is preclinical evidence for both agonist and antagonist properties of ketamine at GABA_A receptors. For example, peripheral administration of subthreshold doses of ketamine (0.1 mg/kg, i.p.) combined with the GABA_A receptor agonist muscimol (0.1 mg/kg, i.p.) induced a synergistic antidepressant behavioral response in the acute (30 minutes postinjection) forced-swim test in mice (Rosa et al., 2016). In contrast, direct infusion of muscimol into the infralimbic prefrontal cortex abolished the sustained (24 hours postinjection) antidepressant behavioral effects of ketamine in rats (Fuchikami et al., 2015), suggesting that the in vivo interaction between ketamine and GABA_A receptors might be brain region specific. In addition, ketamine-associated delirium occurring at anesthetic doses (1–2 mg/kg, i.v. infusion) is effectively minimized via pretreatment with benzodiazepines (positive allosteric modulators of GABA_A receptors; Dundee and Lilburn, 1978; Perumal et al., 2015). In contrast, Irifune et al. (2000) showed that both muscimol and the benzodiazepine receptor agonist diazepam augment ketamine-induced anesthesia, whereas the GABA_A receptor antagonist bicuculline antagonizes ketamine-induced anesthesia in mice. Finally, at subanesthetic doses, ketamine does not bind to GABA_A receptors in the human brain, as assessed by positron emission tomography (PET) scan imaging (Salmi et al., 2005), and does not alter GABA_A receptor function at anesthetic-relevant concentrations (i.e., 10 μM) in HEK293 cells in vitro (Flood and Krasowski, 2000). These findings indicate that, at least at subanesthetic doses, ketamine itself might only indirectly affect GABA_A receptor activity to exert any relevant behavioral effects. Indeed, it is postulated that ketamine-induced dissociation/psychotomimetic effects are due to NMDAR blockade on GABAergic inhibitory interneurons, an action that is presumed to disinhibit excitatory neurotransmission via decreased GABA release, and to consequently reduce activation of the GABA_A receptors in glutamatergic synapses (Moghaddam et al., 1997; Farber et al., 1998; Homayoun and Moghaddam, 2007; Hare et al., 2017; Wohleb et al., 2017).

D. Cholinergic Receptors

Ketamine is reported to bind to both muscarinic and nicotinic acetylcholine receptors (mAChRs and nAChRs, respectively). To date, five subtypes of mAChRs have been identified (M1–M5). These are metabotropic receptors that signal primarily, although not exclusively, through G_αi/o (M1, M3, and M5) or G_αq (Eglen, 2005). Binding of ketamine to the mAChR subtypes M1, M2, and M3 has been described. In particular, assessment of [³H]quinuclidinyl benzilate displacement revealed that (S)-ketamine has ∼twofold higher affinity than (R)-ketamine for mAChRs (Hustveit et al., 1995). In a subsequent study, Hirota et al. (2002) demonstrated that, with K_i values of approximately 45, 294, and 246 μM, respectively, ketamine displaced [³H]N-methyl scopolamine binding to M1, M2, and M3 mAChRs ectopically expressed in Chinese hamster ovary (CHO) cells. However, the authors reported that ketamine had no significant effect on basal or methacholine-induced Ca²⁺ signals in M1-expressing CHO cells (Hirota et al., 2002). In contrast, Durieux (1995) reported that, at clinically relevant concentrations, ketamine inhibited M1 mAChR activation in Xenopus oocytes (IC₅₀ = 5.7 μM). The apparently discrepant results could be accounted for by the fact that receptors ectopically expressed in CHO cells and oocytes can have differential sensitivity (e.g., McIntyre et al., 2001). These findings suggest the need for additional studies to identify the exact effects of ketamine on mAChRs and the functional relevance of these effects.

In contrast to the metabotropic mAChRs, nAChRs are ionotropic receptors, which are nonselective cation channels activated by the neurotransmitter acetylcholine. These receptors are composed of five subunits. To date, 10 α (α1–α10) and four β (β1–β4) nAChR subunits have been cloned, and different combinations of these subunits give rise to a number of functional nAChR subtypes, which are expressed in neuronal and non-neuronal cells and have specific pharmacological, functional, and kinetic properties (Albuquerque et al., 2009).

Ketamine is reported to act as a noncompetitive, open-channel blocker of the α7, α4β2, α4β4, and α3β4 nAChR subtypes (Flood and Krasowski, 2000; Yamakura et al., 2000; Coates and Flood, 2001; Jozwiak et al., 2002; Pereira et al., 2002). Ketamine concentration dependently blocked acetylcholine (1 mM)-induced activation of α4β4 nAChRs ectopically expressed in Xenopus oocytes (IC₅₀ = 0.24 ± 0.03 μM), with complete inhibition being achieved with 10 μM ketamine (Flood and Krasowski, 2000). In addition, with IC₅₀ values of 20 and 50 μM, respectively, ketamine inhibited acetylcholine (1 mM)-induced activation of α7 and α4β2 nAChRs ectopically expressed in oocytes (Coates and Flood, 2001). Moreover, Moaddel et al. (2013) showed that ketamine inhibits nicotine-induced α3β4 nAChR activation (IC₅₀ = 3.1 μM). Because ketamine concentrations up to ∼10 μM are within the clinically relevant range, some nAChR subtypes may underlie the effects of ketamine in vivo.

Moaddel et al. (2013) also demonstrated that at 100 nM (R,S)-DHNK reduced the amplitude of acetylcholine-induced whole-cell currents in KXα7R1 cells that ectopically express α7 nAChRs by approximately 60%. At the same concentration, the metabolites (2S,6S)-HNK, (2R,6R)-HNK, and (R,S)-norketamine also reduced the amplitude of acetylcholine-induced α7 nAChR currents by approximately 54%, 51%, and 45%, respectively (Moaddel et al., 2013). In this study, the authors reported that (R,S)-DHNK was not acting as a channel blocker at α7 nAChRs, because its inhibitory effect was voltage independent. Moreover, (R,S)-DHNK did not bind at the agonist-binding site of the α7 nAChRs, because it did not displace α-bungarotoxin binding, suggesting that this metabolite might be acting as a negative allosteric modulator (Moaddel et al., 2013).

The α7 nAChR antagonist activity of ketamine metabolites may have implications for the antidepressant action of ketamine. It is noteworthy in this context that blockade of α7 nAChRs results in antidepressant effects in rodent models (Mineur and Picciotto, 2010; Philip et al., 2010). Thus, modulation of α7 nAChR activity could also be one of the underlying mechanisms involved in ketamine’s antidepressant actions, possibly through its metabolites. In support of this hypothesis, nAChR antagonists are aleady in clinical trials for the treatment of depression (Mineur and Picciotto, 2010; Philip et al., 2010).

Other nAChR subtypes are also sensitive to inhibition by therapeutically relevant concentrations of ketamine metabolites. Specifically, in KXα3β4R2 cells stably expressing rat α3β4 nAChRs, ketamine and (R,S)-norketamine were reported to inhibit the α3β4 nAChR with IC₅₀ values of approximately 3.1 and 9.1 μM, respectively, whereas DHNK, (2S,6S)-, or (2R,6R)-HNK did not have any significant effect on these receptors (IC₅₀ > 200 μM; Moaddel et al., 2013). Therefore, one cannot rule out the possibility that α3β4 nAChRs also contribute to the pharmacological effects of ketamine.

E. Monoaminergic Receptors and Transporters

Dopamine (DA) and serotonin [5-hydroxytryptamine (5-HT)] receptors are metabotropic receptors, with the exception of the 5-HT receptor subtype 3, which is ionotropic. Five different subtypes of DA receptors (D_1–5R) and seven subtypes of 5-HT receptors (5-HT_1–7R) have been characterized in the central nervous system (De Felice, 2017). Neurotransmitter transporters regulate DA and 5-HT uptake across the cellular/intracellular membranes and play a key role in the regulation of dopaminergic and serotoninergic neurotransmission.

Although there is some evidence that ketamine may act on DA receptors and transporters (see Table 3), there is also conflicting evidence indicating that ketamine does not directly alter dopaminergic signaling; also see Kokkinou et al. (2018). Ketamine was reported to have high affinity (0.06–1.0 μM) for D₂ receptors (D₂Rs; Kapur and Seeman, 2002; Seeman et al., 2005) and to act as a partial agonist at these receptors (EC₅₀ = 0.9 ± 0.4 μM). This partial agonist activity at D₂Rs was suggested to contribute to the psychotomimetic effects of ketamine (Kapur and Seeman, 2002; Seeman et al., 2005). This is further supported by the finding that a ketamine-induced decrease in D₂R binding is significantly correlated with schizophrenia-related symptoms in humans, as measured by PET scanning (Breier et al., 1998).

There are several published studies reporting ketamine-induced decreases in striatal D₂R binding (an indirect measure of DA release) in humans. Specifically, subanesthetic doses of ketamine (i.v. bolus of ketamine; 0.12 mg/kg), followed by an i.v. infusion of 0.65 mg/kg ketamine over 60 minutes (0.29–0.45 μM; Breier et al., 1998), 0.5 mg/kg ketamine over 20 minutes (Smith et al., 1998), or 0.014 mg/kg per minute (S)-ketamine for 90 minutes (Vollenweider et al., 2000) decreased striatal D₂R binding. However, subsequent studies have failed to replicate these findings using single-photon emission computed tomography/PET techniques. In particular, when ketamine was administered as an i.v. bolus (0.12 mg/kg) followed by a constant 70-minute i.v. infusion (0.65 mg/kg per hour), resulting in an average plasma concentration of 0.59 ± 0.22 μM in healthy individuals, there was no change in D₂R availability in striatal regions of the brain (Kegeles et al., 2002). Similarly, i.v. infusion of ketamine (66-minute infusion yielding a stable plasma ketamine concentration of 1.23 ± 0.12 μM) in healthy male volunteers did not alter striatal D₂R binding (Aalto et al., 2002). In accordance with the lack of direct effects of subanesthetic doses of ketamine on the dopamine D₂R, psychotomimetic effects of a 0.23 mg/kg bolus or 0.65 mg/kg per hour infusion of ketamine were not blocked by administration of a D₂R antagonist in humans (Krystal and D’Souza, 2001). In addition, ketamine was not found to modify electrically evoked accumbal DA release measured in real-time using fast-scan cyclic voltammetry in mice (Can et al., 2016). In contrast to the previously mentioned studies, it was also reported that ketamine lacks functional agonist and antagonist activity on all of the DA receptor subtypes at concentrations up to 10 μM (Can et al., 2016).

Inhibition of the DA transporter by ketamine in HEK293 cells has also been reported (K_i = 62.9 μM; Nishimura et al., 1998). However, this effect did not occur at concentrations up to 10 μM (Can et al., 2016). Because this effect was only observed at relatively high concentrations, its functional/clinical relevance remains to be determined.

Ketamine was also reported to bind to 5-HT₂ receptors with an affinity of 15 ± 5 μM (Kapur and Seeman, 2002). This finding might be relevant to the analgesic effects of the drug, because the 5-HT_2B/2C receptor antagonist methysergide inhibited the analgesic effects of ketamine in rats (Crisp et al., 1991), implicating serotonergic signaling in the mechanisms of ketamine analgesia. Moreover, ketamine administration induced an increase in extracellular serotonin (5-HT) levels in the prefrontal cortex and dorsal raphe nucleus of mice (Pham et al., 2017b).

In nonhuman primates, ketamine administration was shown to significantly increase accumbal and ventral pallidum 5-HT_1B receptor binding, an effect that was blocked by pretreatment with the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX; Yamanaka et al., 2014). This finding may suggest that this effect of ketamine is involved in its antidepressant actions because AMPA receptor activation is a convergent mechanism of ketamine’s antidepressant actions, and NBQX administration abolishes ketamine’s effects in several animal tests of antidepressant efficacy (Maeng et al., 2008; Autry et al., 2011; Koike et al., 2011; Walker et al., 2013; Fukumoto et al., 2014; Koike and Chaki, 2014; Zhou et al., 2014; Yang et al., 2015; Zanos et al., 2016; Zanos et al., 2018b,c). Antagonist actions of ketamine on 5-HT₃ receptors have been reported, but these occur at higher than clinically relevant concentrations (Ki > 90 μM; IC₅₀ > 100 μΜ; Appadu and Lambert, 1996; Yamakura et al., 2000; Ho and Flood, 2004).

Several studies suggest that increased 5-HT levels are necessary for the antidepressant-like effects of ketamine in rodents. For instance, pharmacological treatments that reduce 5-HT levels in the brain abolished ketamine’s antidepressant behavioral effects in the forced-swim test (du Jardin et al., 2016; Pham et al., 2017b) and in the novelty-suppressed feeding test (Fukumoto et al., 2014) in rodents. In addition, higher extracellular 5-HT levels were positively correlated with ketamine’s antidepressant activity in the forced-swim test in mice (Pham et al., 2017b). Whether these effects are due to a direct action of ketamine on 5-HT receptors is not clear and needs further investigation. Nevertheless, there are also conflicting data on the effects of ketamine on 5-HT receptors, because even at very high concentrations (1 mM) ketamine only slightly altered [³H]5-HT or [³H]spiroperidol binding to 5-HT₁ or 5-HT₂ receptors, respectively (Martin et al., 1982). It is thus possible that ketamine may interact with serotonin uptake as opposed to directly binding to serotonin receptors. Indeed, administration of antidepressant doses of ketamine (1.5 mg/kg; 40-minute infusion) to nonhuman primates reduced serotonin transporter (SERT) activity (Yamamoto et al., 2013), an effect that was hypothesized to reflect direct binding of ketamine to SERTs to regulate 5-HT reuptake. However, in vitro work indicated that ketamine inhibits SERTs at concentrations ranging from 75 (Azzaro and Smith, 1977) to 162 μM (Nishimura et al., 1998), which are not only above the antidepressant-relevant concentrations, but also well above the clinical anesthetic concentrations of ketamine. At antidepressant-relevant concentrations, ketamine does not have an agonist or antagonist effect on SERTs (Can et al., 2016). Therefore, in vivo evidence of ketamine-induced inhibition of serotonin reuptake (Martin et al., 1982) could be attributed to indirect interactions of ketamine with the serotonergic system, at least at subanesthetic concentrations.

Finally, although there is some evidence that ketamine acts as an uptake inhibitor at norepinephrine transporters (NETs; K_i = 66.8 μM; Nishimura et al., 1998), the inhibition constant indicates that ketamine would not modulate NET function at clinically relevant concentrations (up to 10 μΜ). There is also a noted lack of functional activity of clinically relevant concentrations of ketamine (up to 10 μΜ) on NETs (Can et al., 2016).

To date, there is only one published study assessing the effect of ketamine’s metabolites on monoaminergic receptors and transporters. At concentrations up to 10 μM there was no agonist or antagonist activity of (S)-norketamine, (R)-norketamine, (S)-dehydronorketamine, (R)-dehydronorketamine, (2S,6S)-HNK, or (2R,6R)-HNK on D₁, D₂, D₃, D₄, or D₅ receptors; DA transporter; NET; or SERT (Can et al., 2016). Although these findings suggest that direct effects of ketamine metabolites on DA receptors or monoaminergic transporters do not account for the antidepressant actions of ketamine, it cannot be ruled out that, at higher concentrations, ketamine’s metabolites may interact with and directly or indirectly modify the activity of these receptors and transporters.

F. Opioid Receptors

Opioid receptors are expressed throughout the central nervous system as well as in peripheral tissues (Trescot et al., 2008). These receptors are G protein–coupled receptors and are classified into three subtypes (μ-, δ-, and κ-opioid receptors) (Kieffer and Gaveriaux-Ruff, 2002). A primary function of opioid receptor activation is inhibiting the transmission of nociceptive stimuli, resulting in analgesia (Trescot et al., 2008; Stein, 2016).

Ketamine was reported to activate human recombinant μ-, κ-, and δ-opioid receptors expressed in CHO cells (K_i = 42.1, 28.1, and 272 μΜ, respectively; Hirota et al., 1999). (S)-ketamine was shown to have higher affinity compared with (R)-ketamine for binding to opioid receptors. In particular, (S)-ketamine’s affinities range from 11 to 29 μΜ for the μ-, 25–28 μΜ for the κ-, and 130–205 μΜ for the δ-opioid receptors (Hustveit et al., 1995; Hirota et al., 1999; Nemeth et al., 2010). In contrast, (R)-ketamine has affinities ranging from 28 to 84 μΜ for the μ-, 60–100 μΜ for the κ-, and 130–286 μΜ for the δ-opioid receptors (Hustveit et al., 1995; Hirota et al., 1999). Similar to these findings, (S)-ketamine (IC₅₀ = 23.0 ± 1.2 μM) showed higher potency to bind and displace the non-specific opioid ligand [³H]dihydromorphine compared with the racemic ketamine (IC₅₀ = 16.3 ± 7.4 μM) and (R)-ketamine (IC₅₀ = 45.5 ± 3.2 μM; Finck and Ngai, 1982).

The actions of ketamine at opioid receptors are hypothesized to be involved in its analgesic effects (Finck et al., 1988; Pacheco Dda et al., 2014), and these findings are consistent with the higher potency of (S)-ketamine in measures of antinociception compared with (R)-ketamine (White et al., 1980; Oye et al., 1992; Mathisen et al., 1995). However, the exact roles of opioid receptors in mediating these effects are unclear.

Although the analgesic actions of ketamine are blocked by intracerebroventricular administration (i.e., direct brain exposure) of μ- and δ-, but not κ-opioid receptor antagonists in mice (Pacheco Dda et al., 2014), global opioid receptor inhibition achieved by systemic administration of naloxone does not prevent ketamine-induced analgesia in humans (Mikkelsen et al., 1999). These seemingly opposing findings indicate that ketamine might induce analgesia via an indirect interaction with the opioid system, or may exert an opioid receptor subtype-specific action. In support of such a subtype-specific effect, in vivo evidence indicates that ketamine might be a κ-opioid receptor agonist and a μ-opioid receptor antagonist. Specifically, ketamine microinjection in the periaqueductal gray region, which contains μ- but not κ-opioid receptors, did not exert antinociceptive effects, but blocked the effects of the μ-opioid receptor agonist, morphine, in the same region (Smith et al., 1985). This could explain the failure of naloxone to block ketamine-induced analgesia and supports a subtype-specific role of opioid receptors in ketamine-induced analgesia. However, it should be noted that these findings are in opposition of the results of Pacheco Dda et al. (2014), as discussed above. Furthermore, it has been reported that subeffective doses of ketamine only partially antagonize the cataleptic actions of morphine in rats (Hance et al., 1989); however, a combination of subeffective doses of ketamine and morphine induced catalepsy in the same experimental settings (Hance et al., 1989). Altogether, these findings implicate the opioid system in ketamine’s analgesic effects, but further studies are needed to clarify the exact mechanisms. Interactions between ketamine and the opioid system may be more relevant in chronic pain, in which ketamine reduces opioid tolerance. In particular, via acting on the downstream extracellular signal-regulated kinase 1/2 pathway, ketamine (10 μΜ) abolished μ-opioid receptor-induced desensitization in vitro (Gupta et al., 2011).

Activation of κ-opioid receptors by ketamine (EC₅₀ = 29 μΜ) was reported to be involved in ketamine’s effects on attention and visual perception in rats assessed in the five-choice serial reaction time task (Nemeth et al., 2010). Pretreatment with a selective κ-opioid receptor antagonist blocked the ability of ketamine (20 mg/kg) to impair attention and visual perception (Nemeth et al., 2010). In contrast, in healthy volunteers, inhibition of opioid receptors by naloxone augmented the dissociative and cognitive deficit effects induced by a subthreshold administration of ketamine (1-minute 0.23 mg/kg bolus, followed by a 60-minute 0.58 mg/kg ketamine infusion; Krystal et al., 2006), possibly indicating an antagonist activity of low doses of ketamine on the opioid receptors in vivo. Nevertheless, further work is needed to establish the exact role of opioid receptor modulation in mediating the side effects of ketamine. There are currently no published studies assessing the effects of ketamine’s metabolites on opioid receptors. However, there is recent evidence indicating that (2R,6R;2S,6S)-HNK (at 10 and/or 30 mg/kg, s.c.) does not possess antinociceptive properties and does not alter opioid (morphine) tolerance in rats (Lilius et al., 2018a), indicating a possible lack of interactions between this metabolite and the opioid receptor system. Notably, norketamine, similar to ketamine, attenuates tolerance to morphine (Lilius et al., 2018b).

G. Sigma Receptors

Another potential site of action of ketamine is the sigma receptor. Sigma receptors are classified into two different subtypes, sigma I and II receptors (σ₁R and σ₂R, respectively; Bowen et al., 1989). Although a third subtype has also been suggested to exist (σ₃R), it has not been fully defined (Myers et al., 1994).

As illustrated in Table 3, ketamine binding has been described at both σ1R and σ2R (Smith et al., 1987; Hustveit et al., 1995; Robson et al., 2012). The first evidence of direct binding of ketamine to sigma receptors came from Klepstad et al. (1990), who demonstrated that (R)-ketamine has a potency of IC₅₀ = ∼15 μM to bind to sigma receptors, whereas (S)-ketamine has an IC₅₀ of ∼100 μM. Although these findings were qualitative and not absolutely quantitative due to the different brain regions used for the binding studies, Hustveit et al. (1995) confirmed these initial findings by showing that (R)-ketamine has an affinity of 19 μM at sigma receptors, whereas (S)-ketamine shows ∼15-fold lower binding affinity for these receptors (131 μM). Robson et al. (2012) reported that (R,S)-ketamine has a preferential affinity for the σ₂R (K_i = 26.3 μΜ) compared with the σ₁R (K_i = 139.6 μΜ). In support of such binding in vivo, studies in nonhuman primates showed competition binding between ketamine and [¹¹C]SA5845, a sigma receptor PET tracer (Kortekaas et al., 2008).

Sigma receptors are promising targets for antidepressant treatment (Fishback et al., 2010), and activation of these receptors induces antidepressant behavioral responses in animals (Matsuno et al., 1996; Skuza and Rogoz, 2002; Wang et al., 2007; Lucas et al., 2008) and humans (Pande et al., 1999). As a result it could be postulated that the action of ketamine on sigma receptors is involved in the mechanisms underlying its antidepressant responses, consistent with (R)-ketamine’s higher binding affinity for these receptors, and its more potent antidepressant effects compared with (S)-ketamine in rodent models (Zhang et al., 2014; Yang et al., 2015; Zanos et al., 2016; Fukumoto et al., 2017).

Although administration of σ₁R and σ₂R antagonists did not attenuate the antidepressant behavioral effects of ketamine in mice (Robson et al., 2012), administration of a σ₁R-selective antagonist blocked the potentiating effects of ketamine on nerve growth factor–induced neurite outgrowth, and thus, modulation of neuroplasticity-related pathways (Robson et al., 2012) that are involved in the antidepressant effects of the drug (see Kavalali and Monteggia (2012), Gerhard et al. (2016)). These findings may indicate that ketamine’s actions on sigma receptors could be involved in the neuroplasticity-related effects of the drug, and thus indirectly involved in its antidepressant actions. There is currently no evidence of activity of ketamine’s metabolites on sigma receptors.

H. Voltage-Gated Sodium Channels

Voltage-gated ion channels are among the first identified ion channels and are involved in the generation of action potentials (Hodgkin and Huxley, 1952). Local anesthetics typically induce concentration-dependent inhibition of sodium channel activity, via binding to sites within the channel pore (Becker and Reed, 2012). Whereas ketamine has been shown to act as a local anesthetic in veterinary practice and in human patients (Bion, 1984; Gomez de Segura et al., 1998; Hawksworth and Serpell, 1998; Kathirvel et al., 2000), there is currently conflicting evidence regarding the effects of ketamine on voltage-gated sodium channel activity.

In isolated guinea pig ventricular myocytes, ketamine at concentrations ranging from 30 to 300 μΜ induced a concentration-, but not use-dependent tonic inhibition of sodium channel currents (16%–36% inhibition; Hara et al., 1998b). In contrast, both tonic (IC₅₀ = 866.2 ± 34.7 μΜ) and phasic (IC₅₀ = 314.8 μΜ) blockade of sodium channels was induced by ketamine in rat dorsal root ganglion neurons (Zhou and Zhao, 2000). In addition, ketamine was reported to block voltage-gated sodium channels in Xenopus oocytes (tonic inhibition: IC₅₀ = 800 μΜ; phasic inhibition: IC₅₀ = 2.3 mM; Wagner et al., 2001). Moreover, although Benoit (1995) demonstrated no inhibitory effect of ketamine on nodal sodium- channel currents of myelinated nerve fibers, others have reported up to 71.1% blockade of sodium channel conductance by ketamine (ED₅₀ = 1.1 mM; Frenkel and Urban, 1992). A study assessing the effects of (S)- and (R)-ketamine on the activity of sodium channels showed that (S)-ketamine inhibits voltage-gated sodium channels with an apparent potency of 240 ± 60 and 59 ± 10 μΜ in neuronal and skeletal muscle isoforms, respectively. The potency of (R)-ketamine to block sodium channels was lower (IC₅₀ = 333 ± 93 and 181 ± 49 μΜ in neuronal and skeletal muscle isoforms, respectively; Haeseler et al., 2003) than that of (S)-ketamine. Similarly, Schnoebel et al. (2005) showed stereoselective tonic block of voltage-gated sodium currents with (S)-ketamine being more potent than (R)-ketamine (IC₅₀ = 128 and 269 μΜ, respectively). These findings predict that (S)-ketamine is a more effective local anesthetic than (R)-ketamine.

These discrepancies in the effects of ketamine on the activity of sodium channels might be due to differences in the cell populations and experimental procedures used. Overall, these findings support an inhibitory effect of ketamine on sodium channels, which is a characteristic of local anesthetics (Becker and Reed, 2012). These effects occur at concentrations well above circulating ketamine levels relevant for general anesthesia, but could possibly be relevant to local ketamine anesthesia.

I. L-Type Voltage-Dependent Calcium Channels

The L-type voltage-dependent calcium channel (VDCC) family consists of four different members referred to as Ca_v1.1–Ca_v1.4 (Catterall, 2011). Ca_v1.2 is the main LTCC expressed in the mammalian brain (Bhat et al., 2012). Antagonism of L-type VDCCs by ketamine has been reported over a wide range of concentrations. In porcine tracheal smooth muscle cells, for example, ketamine blocked VDCCs with an IC₅₀ of 1 mM (Yamakage et al., 1995). In rabbit portal vein smooth cells, 1 mM ketamine completely inhibited VDCC currents (Yamazaki et al., 1992). Finally, in bullfrog single atrial cells, ketamine inhibited VDCC currents with an IC₅₀ of 9.2 μΜ (Hatakeyama et al., 2001). This effect was not use-dependent, and ketamine did not act as an open channel blocker. Nonuse–dependent tonic inhibition of VDCCs was also reported in isolated guinea pig ventricular myocytes, in which 30–300 μΜ induced a 26%–53% inhibition (Hara et al., 1998b). The discrepancies in the reported concentrations at which ketamine inhibits VDCCs may be due to differences in species, cell type, or experimental preparations (e.g., differences in bath solutions).

Despite evidence of VDCC inhibition by ketamine, an in vitro study showed that AMPA receptor activation and subsequent increases in brain-derived neurotrophic factor release and mechanistic target of rapamycin complex 1 activation—actions that are believed to underlie ketamine’s antidepressant properties—require VDCC activation (Jourdi et al., 2009). Furthermore, ketamine’s antidepressant effects in mice are blocked by pretreatment with L-type calcium channel antagonists (Lepack et al., 2014). These findings are not in line with the inhibitory activity of ketamine at these channels, as previously described. However, it should be noted that these studies were performed in cells derived from peripheral tissues (e.g., heart, trachea) and that effects of ketamine on VDCCs in brain-derived cells at antidepressant-relevant concentrations have not been reported to our knowledge. In addition to the possible role of ketamine-induced inhibition of calcium channels in mediating the antidepressant actions of the drug, blockade of calcium channels has been also hypothesized to be involved in the psychoactive/psychotomimetic effects of ketamine (Lisek et al., 2016).