Despite nearly five decades of clinical use, a mechanistic description of ketamine’s ability to reliably produce anesthesia for surgery has been elusive. Ketamine is classified as a “dissociative anesthetic” because it produces a markedly distinct anesthetized state compared to other anesthetic drugs. Furthermore, ketamine acts by increasing the high-frequency oscillations in the beta-gamma range, which is distinct from γ-aminobutyric acid–mediated (GABAergic) anesthetics. Ketamine was thought to exert its anesthetic effect by acting as a non-competitive antagonist of N-methyl-D-aspartate (NMDA) receptors.  However, the lack of anesthetic effects with MK-801, a non-competitive NMDA receptor antagonist structurally similar to ketamine, suggests that NMDAR antagonism may not be the primary mechanism for ketamine anesthesia. Ketamine also affects α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPAR) which is thought to contribute to its antidepressant effects but its interaction with AMPARs in the context of general anesthesia mechanisms remains unexplored.

Several studies have reported the role of AMPARs in the antidepressant effect of sub-anesthetic doses of ketamine (concentrations estimated to range 30 to 50 µM) in rodents.  However, the involvement of AMPARs in ketamine-induced unconsciousness at higher doses (100 µM) has yet to be fully addressed. In this study, we used whole cell recordings to explore the effect of ketamine on the postsynaptic AMPARs in pyramidal cells at layer 5 of the somatosensory cortex. First, we investigated how ketamine alters AMPAR kinetics by measuring the decay time constant (tau) of AMPAR spontaneous excitatory postsynaptic currents (sEPSCs) before and after ketamine administration. We utilized this approach to show that ketamine 100 µM, which is a clinically relevant concentration for surgical anesthesia (see below), significantly prolonged the AMPAR-mediated decay time constant by about 34% in sEPSCs at -70 mV. However, at ketamine 30 µM, the impact was less pronounced, showing an increase of roughly 11%, albeit still statistically significant (fig. 1). Altogether, these results raise the possibility that the observed increase in decay time constant, may have clinical or physiologic relevance for anesthesia.

Fig. 1.
The effect of ketamine on the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors decay tau. Panels A and B represent average of spontaneous excitatory postsynaptic currents decay tau in different concentrations of ketamine 30 µM and 100 µM, respectively. All experiments were conducted in presence of amino-5-phosphonovaleric acid 50 µM and picrotoxin 50 µM. A line connects measurements from the same cell. Each connected pair of points on the graph represents the mean decay tau before and after ketamine addition, for randomly selected sets of 50 excitatory postsynaptic currents from the control and ketamine groups within each cell. Bars represent averages of decay taus before and after ketamine. The mean decay tau for ketamine 30 µM = 11.85 ms, and for ketamine 100 µM = 33.91 ms. Asterisks indicate *P = 0.04 for ketamine 30 µM and *P = 0.02 for ketamine 100 µM; paired Student’s t test. Error bars represent means ± SE.

The effect of ketamine on the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors decay tau. Panels A and B represent average of spontaneous excitatory postsynaptic currents decay tau in different concentrations of ketamine 30 µM and 100 µM, respectively. All experiments were conducted in presence of amino-5-phosphonovaleric acid 50 µM and picrotoxin 50 µM. A line connects measurements from the same cell. Each connected pair of points on the graph represents the mean decay tau before and after ketamine addition, for randomly selected sets of 50 excitatory postsynaptic currents from the control and ketamine groups within each cell. Bars represent averages of decay taus before and after ketamine. The mean decay tau for ketamine 30 µM = 11.85 ms, and for ketamine 100 µM = 33.91 ms. Asterisks indicate *P = 0.04 for ketamine 30 µM and *P = 0.02 for ketamine 100 µM; paired Student’s t test. Error bars represent means ± SE.

We compared NMDA and AMPA evoked EPSCs (eEPSCs) within cells by calculating their current ratio. In these conditions, both NMDAR and AMPAR are exposed equally to synaptically released glutamate, independent of the electrode positioning or number of presynaptic afferents, and thus provide a more reliable comparison of the effect of ketamine on NMDA and AMPA currents. Ketamine significantly increased the AMPA/NMDA ratio, suggesting that ketamine alters glutamatergic neurotransmission, by disrupting the balance between AMPAR and NMDAR activity (fig. 2 ). When ketamine was administered at 30 µM, its influence on AMPA current amplitude was less pronounced. This outcome aligns with the effect of ketamine 30 µM on the decay tau of this receptor.

Fig. 2.
Ketamine potentiates AMPA to NMDA ratio. (A) Sample traces of evoked synaptic AMPA and NMDA receptor mediated currents in the same cell (picrotoxin 50 µM was added to the superfusate). The AMPA component was obtained by measuring the peak amplitude of excitatory postsynaptic currents at –70 mV; the NMDA component was recorded at +40 mV and determined by measuring the excitatory postsynaptic current amplitude 70 ms after the peak. Control traces are in blue and red traces after adding ketamine 100 µM. (B) The change in AMPA/NMDA ratio analysis compared to control (before ketamine was added) showing a remarkable increase following ketamine administration. The average amplitude for ketamine 30 µM increased by 114% from the control (mean ± SE: 114 ± 27.52; n = 8), and ketamine 100 µM demonstrated a 90% increase in amplitude compared to control (mean ± SE: 90 ± 21.6; n = 8).

Ketamine potentiates AMPA to NMDA ratio. (A) Sample traces of evoked synaptic AMPA and NMDA receptor mediated currents in the same cell (picrotoxin 50 µM was added to the superfusate). The AMPA component was obtained by measuring the peak amplitude of excitatory postsynaptic currents at –70 mV; the NMDA component was recorded at +40 mV and determined by measuring the excitatory postsynaptic current amplitude 70 ms after the peak. Control traces are in blue and red traces after adding ketamine 100 µM. (B) The change in AMPA/NMDA ratio analysis compared to control (before ketamine was added) showing a remarkable increase following ketamine administration. The average amplitude for ketamine 30 µM increased by 114% from the control (mean ± SE: 114 ± 27.52; n = 8), and ketamine 100 µM demonstrated a 90% increase in amplitude compared to control (mean ± SE: 90 ± 21.6; n = 8).

Because there was no consistent increase in AMPA eEPSCs amplitude at 30 uM (P = 0.53), this suggests that the increase in AMPA:NMDA at the lower ketamine concentration was driven by the observed reduction in NMDA eEPSCs amplitude. To further confirm the previous observation, we measured the components of eEPSCs related to AMPA and NMDA receptors. AMPAR currents were isolated by using APV and picrotoxin to block NMDA and GABAergic activity, respectively, while NMDA currents were obtained by blocking AMPAR activity with CNQX. Ketamine significantly increased AMPAR current amplitudes and reduced NMDAR current amplitudes (Supplemental Digital Content 1, https://links.lww.com/ALN/D608: ketamine increased the AMPAR current from 121.9 ± 25.1 pA to 322.5 ± 16.7, **P = 0.004, n = 4, paired t test).

Given that any change in the properties of presynaptic release of glutamate would be expected to similarly impact AMPA and NMDA currents, the observed divergent effects of ketamine application on these currents indicates that the effect of ketamine on AMPAR EPSCs is likely postsynaptic. These findings reveal that AMPAR sEPSC potentiation by ketamine occurs with anesthetic concentrations and therefore may play a role in ketamine-induced unconsciousness.

It may seem paradoxical that an anesthetic agent which acts as an antagonist of the glutamatergic NMDAR could generate excitation. To our knowledge, this study is the first to indicate that ketamine might render patients unconscious by increasing synaptic activity of the brain rather than by decreasing it. This view is supported by human imaging studies, which demonstrate, contrary to volatile gases and propofol, that there is no decrease in brain metabolism when sub-anesthetic or anesthetic doses of ketamine are administered, compared to awake, conscious states.

In summary, based on our findings, we conclude that ketamine anesthesia can be better understood as disrupting essential networks for consciousness rather than simply an inhibition of cortical activity, like the unconsciousness present during seizures. Our results demonstrate that a disruption in the balance of AMPAR and NMDAR by ketamine in the somatosensory cortex may contribute to its mechanism of inducing unconsciousness. The simultaneous blockade of NMDAR and potentiation of AMPAR postsynaptic currents can create an imbalance in synaptic strength that ultimately results in the lowering of the threshold for action potential firing and consequently increased synaptic “noise” leading to disruption of network activity that is vital for consciousness. 

A limitation in our study arises from the specific ketamine concentrations used in our ex vivo preparation, which fails to completely replicate the plasma binding and diffusion kinetics of an in vivo brain.11,12  Our estimated brain concentrations were determined based on existing literature for subanesthetic and anesthetic doses.13  In mice, the recommended intraperitoneal injection dose of ketamine for surgical anesthesia is 50 to 100 mg/kg.  This corresponds to a brain level of approximately 90 to 180 µM after 10 min of administration. Therefore, we focused on the effect of 100 µM ketamine which falls within the anesthetic range, but also a concentration of ketamine (30 µM) representative of a subanesthetic dose. However, one should also bear in mind that working with coronal brain slices poses challenges, such as uneven drug distribution, a variability in AMPAR and NMDAR expression, and truncation of afferents that could ultimately impact the recorded activity and introduce a response-dose mismatch.