Dexmedetomidine Prevents Excessive γ-Aminobutyric Acid Type A Receptor Function after Anesthesia

THE availability of safe, modern general anesthetic drugs has ensured that most patients now survive surgery and are discharged from the hospital. However, many patients experience cognitive deficits such as delirium in the first few days after surgery.^1,2  Delirium is associated with a loss of independence, a failure to return to work, a substantial increase in healthcare costs and death.^1–3  Given that more than 312 million patients undergo anesthesia and surgery each year,⁴  it is essential to develop strategies to reduce postoperative delirium. Although nonpharmacologic interventions are somewhat effective, pharmacologic prevention and treatments are needed.¹ 

The pathophysiology underlying postoperative delirium is complex and multifactorial; however, anesthetic drugs are likely a contributing factor. We have previously shown that even a single, brief exposure to a commonly used general anesthetic drug can trigger a sustained increase in cell-surface expression and function of extrasynaptic γ-aminobutyric acid type A (GABA_A) receptors containing the α5 subunit (α5 GABA_A receptors) in hippocampal neurons.⁵  These changes persist long after the anesthetic drugs have been eliminated from the body. To date, no effective strategy has been identified to prevent excessive α5 GABA_A receptor cell-surface expression after general anesthesia.

The α2 adrenergic receptor agonist dexmedetomidine lowers the incidence and duration of delirium in patients who have undergone cardiac and noncardiac surgery.^6–8  Dexmedetomidine also reduces delirium in patients who have been sedated for days in intensive care units.⁹  Studies of laboratory animals have confirmed that dexmedetomidine can prevent cognitive deficits after anesthesia and surgery.^10–13  Here, we investigated the cellular and molecular mechanisms that contribute to the cognition-sparing properties of dexmedetomidine. We postulated that dexmedetomidine reduces the excessive function of α5 GABA_A receptors in neurons that persists after general anesthesia and thereby attenuates postanesthetic cognitive deficits.

Experiments were approved by the Animal Care Committee of the University of Toronto. All studies were performed in accordance with guidelines from the Canadian Council on Animal Care. The mice were located in a pathogen-free facility (25 ± 1°C) on a 14-h light/10-h dark cycle (lights on 6:00 am to 8:00 pm). Timed-pregnant CD1 mice were used to prepare the cell cultures (Charles River, USA). Male C57BL/6 mice aged 8 to 9 weeks (Charles River) were used for behavioral studies, which were performed between 10:00 am and 5:00 pm. Only male mice were selected for the behavioral studies because the estrous cycle influences the expression of extrasynaptic GABA_A receptors.¹⁴  Each mouse was assigned a random number that was generated by Microsoft (USA) Excel 2016. The assigned numbers were sorted from the smallest to the largest, and then the mice were allocated to different treatment groups. Researchers were blinded to the treatment groups for all behavioral experiments and the studies using ex vivo brain slices.

The mice were treated with dexmedetomidine (25 μg/kg, intraperitoneally), which was followed immediately by etomidate (20 mg/kg, intraperitoneally). Normal saline (0.9% w/v of NaCl, intraperitoneally) and propylene glycol (35% v/v in normal saline) were used as vehicle controls for dexmedetomidine and etomidate, respectively. To prevent hypoxia, each mouse was anesthetized in an air-tight acrylic chamber that was flushed with oxygen and medical air (70% air, 30% O₂) delivered at a flow of 1 l/min, as described previously.⁵  Temperature in the chamber was maintained at 35°C with a heating blanket to prevent hypothermia. A total of 44 mice were treated with dexmedetomidine and etomidate. Four of these mice died 30 to 50 min after treatment. None of the mice treated with etomidate alone (n = 42) or dexmedetomidine alone (n = 43) died.

An anti-GABA_A receptor antibody for α5 (PhosphoSolutions, USA), an anti-β-actin antibody (GenScript Biotech Corporation, USA), and an anti-Na⁺/K⁺ ATPase antibody (Developmental Studies Hybridoma Bank, USA) were used to measure cell-surface expression of α5 GABA_A receptors, as previously described.⁵ 

Primary cultures of hippocampal and cortical neurons were prepared from mice (embryonic days 16 to 18).^15,16  For the astrocyte cultures, cortical explants were isolated from mouse embryos, and cell cultures were grown to confluence in Dulbecco’s modified Eagle’s medium and 10% fetal bovine serum for 4 weeks.^17–19  The cells were then enzymatically dissociated with 0.05% trypsin–EDTA (Life Technologies, USA). The astrocyte-containing cell suspension was placed over hippocampal neurons that were cultured for 7 days. Cocultures were maintained for 7 to 10 days before recording from the neurons.

Human astrocytes were isolated from optic nerve heads obtained from deceased donors. Such astrocytes grow well in primary cell culture and closely resemble cortical astrocytes. Their functional properties have been characterized.^20,21  After the fifth passage, the human astrocytes were plated and grown to confluence in a monolayer.

Conventional whole-cell voltage-clamp recordings were performed from cultured hippocampal and cortical neurons.^5,16  At the time of recordings, the neurons become polarized, have extensive axonal and dendritic arbors, and form numerous, functional synapses. Thus, they resemble mature neurons in vivo.^22–24  To record the tonic GABA_A receptor-mediated current, GABA (0.5 µM) was added to the extracellular solution. This GABA concentration was selected because it is similar to the extracellular concentration of GABA that occurs in vivo.^25–27  The competitive GABA_A receptor antagonist bicuculline (20 µM) was used to measure the amplitude of the tonic current.^5,16,28  Previous studies showed that the tonic current in hippocampal neurons is generated primarily by α5 GABA_A receptors.^5,29  Current was recorded at a holding potential of −60 mV at room temperature.

Astrocytes were treated with dexmedetomidine (0.1 µM) or vehicle (1 × phosphate-buffered saline) for 1 h. The culture medium was then exchanged with fresh medium without fetal bovine serum. After 2 h, the medium was collected and then lysed in a cold lysis buffer that contained phosphatase and a protease inhibitor cocktail (Sigma–Aldrich, USA). After centrifugation at 10,000 relative centrifugal force for 10 min, the protein concentration was measured with a bicinchoninic acid kit (Thermo Scientific Pierce Protein Biology, USA). Proteins were separated by electrophoresis with sodium dodecyl sulfate–polyacrylamide gel electrophoresis and then transferred onto nitrocellulose membranes. Primary antibodies to detect phosphorylated extracellular signal–regulated kinase (1:1,000, Cell Signaling, USA), extracellular signal–regulated kinase (1:1,000, Cell Signaling), and the corresponding secondary antibody (1:4,000, Cell Signaling) were used, and the membrane was developed with the ECL kit (PerkinElmer, Canada). An antibody against β-actin (1:4,000, Santa Cruz Biotechnology, USA) was used as an internal control.

Mouse astrocytes were maintained in culture for 4 weeks. The cells were treated with dexmedetomidine (0.1 µM) or vehicle (1 × phosphate-buffered saline) for 1 h. The culture medium was exchanged with fresh medium that lacked fetal bovine serum and were incubated for 2 h. Astrocytes were then lysed with a TRIzol reagent (Thermo Fisher Scientific, USA) to extract the total RNA. The first-strand cDNA was generated from total RNA with iScript reverse transcription supermix for quantitative real-time polymerase chain reaction (Bio-Rad Laboratories, Canada). Primer probe sets for mouse brain-derived neurotrophic factor were purchased from Bio-Rad Laboratories (Unique Assay ID: qMmuCED0050333, UniGene: Mm.1442). Polymerase chain reaction amplification was done using SsoAdvanced Universal SYBR Green Supermix reagent (Bio-Rad Laboratories). The level of messenger RNA (mRNA) was measured with a CFX96 real-time polymerase chain reaction detection system (Bio-Rad Laboratories) and normalized to β-actin RNA transcript levels. The comparative 2^−ΔΔCT method was used for relative quantification.

Mouse astrocytes were cultured in 96-well plates for 4 weeks and then treated with dexmedetomidine (0.01 to 10 µM) or vehicle control for 1 h. Six wells were used for each drug concentration. The culture medium was then changed to 50 µl of fresh medium that lacked fetal bovine serum for 2 h. The conditioned medium was collected and analyzed for brain-derived neurotrophic factor with an enzyme-linked immunosorbent assay kit (lot No. 381264701; BosterBio, USA). Briefly, 200 µl of medium was loaded in each well of a 96-well plate and then coated with mouse monoclonal antibody against brain-derived neurotrophic factor. A brain-derived neurotrophic factor-specific biotin-conjugated secondary antibody was then added, followed by the avidin–biotin–peroxidase complex enzyme and substrate incubation. The absorbance of each sample was measured at 450 nm using microplate reader (Synergy Mx, BioTek, USA). A standard curve was created by using the brain-derived neurotrophic factor standard, and then the concentration of brain-derived neurotrophic factor was measured.

The novel object recognition assay was performed as previously described.⁵  An opaque cylinder-shaped arena that measured 28.5 cm in diameter and 29 cm in height was housed in a dimly lit room. Each mouse was habituated to the chamber for 5 min 24 h before testing. The mouse was trained and tested 24 h after drug treatment. During the training phase, the mouse was allowed to explore two identical “sample” objects for 5 min. Interaction time with the objects was defined when the mouse was sniffing, licking, or touching the object with the nose. The time when the mouse was approaching, climbing, or sitting on the object was excluded.

Executive function was assessed using a puzzle box,³⁰  which consisted of an illuminated start box (58 × 28 × 27.5 cm³) and a dark, enclosed goal box (14 × 28 × 27.5 cm³). A doorway and an underpass that connected the start box with the goal box allowed the animals to move between the two compartments. The mice were tested during a 3-day period, and the level of difficulty increased sequentially over day 1 (trials 1 to 3), day 2 (trials 4 to 6), and day 3 (trials 7 to 8). To assess problem-solving abilities, the task was made increasingly more difficult by modifying the entrance to the goal box. On day 1, trial 1, the goal box was accessible through both an open doorway and an open underpass. On day 1, trial 2, the doorway was closed, and the goal box was accessible only through the underpass. Trial 3 was conducted 2 min after trial 2 under the same conditions to assess an ability to recall escape strategies. Trials 4 to 6 were conducted on day 2. Trial 4 was performed under the same conditions as trial 2. During trial 5, the task was rendered more difficult by closing the door and obstructing the underpass with corncob bedding. This condition required that the mouse burrow through the bedding to gain access to the goal box. Trial 6 was performed 2 min after trial 5 under the same conditions. Trials 7 to 8 were conducted on day 3. Trial 7 was performed under the same conditions as trial 5. During trial 8, the difficulty was increased by obstructing the underpass with a cardboard plug, which the mouse had to remove to gain access to the goal box. During the 2-min intertrial intervals, the mouse was taken out of the puzzle box and placed in a cage that was separate from their home cage. Performance was video-recorded, and the time required for each animal to enter the goal box, which was defined as having all four paws in the goal compartment, was measured. A maximum finish time of 5 min was assigned if the mouse failed to reach the goal box during the trial.

Tetrodotoxin and brain-derived neurotrophic factor were purchased from Alomone Labs (Israel). 6-Cyano-7- nitroquinoxaline-2,3-dione, (2R)-amino-5-phosphonovaleric, and bicuculline were obtained from Abcam (United Kingdom). GABA, etomidate, clonidine, yohimbine, and PD98059 were obtained from Sigma–Aldrich. Dexmedetomidine was obtained from Santa Cruz Biotechnology, sevoflurane was obtained from AbbVie Inc. (Canada), and ANA12 was from Cedarlane (Canada).

The data are presented as means ± SD, together with the 95% CI of the mean. An unpaired Student’s t test was used to compare two groups, where appropriate. For comparing three or more groups, the one-way ANOVA or two-way ANOVA followed by the Tukey’s multiple comparisons post hoc test was used. The normality of datasets was tested with the D’Agostino–Pearson omnibus test (n ≥ 8) or the Kolmogorov–Smirnov test (n < 8). To test for homogeneity of variances, the Brown–Forsythe test and an F test was used for the ANOVA and unpaired Student’s t test, respectively. When assumptions were not met, the nonparametric Mann–Whitney U test was performed instead of the unpaired Student’s t test. Because there is a high degree of variation for puzzle box data, the normality assumptions and homogeneity of variances for the log10-transformed data were confirmed with the D’Agostino–Pearson omnibus test and the Brown–Forsythe test, respectively. The Kaplan–Meier survival analyses of puzzle box data were compared with the log-rank pairwise comparison test, and right censoring was applied at the cut-off time of 5 min. A linear mixed model was employed for statistical analysis of cell-surface and total expression of the GABA_A receptor α5 subunit using IBM SPSS Statistics for Windows, version 23 (IBM Corp., USA). All other analyses were performed using GraphPad Prism 6.01 (GraphPad Software Inc., USA). A two-tailed hypothesis test was used, and statistical significance was set at P < 0.05. No statistical power calculation was conducted before the study, and the sample size was based on our previous experience with this experimental design. During peer review, the sample size for real-time polymerase chain reaction detection of brain-derived neurotrophic factor mRNA was increased, but no adjustments were made for interim analyses. No outliers were excluded from the statistical analyses.

We first asked whether dexmedetomidine reduced excessive cell-surface expression of α5 GABA_A receptors in mice that were treated with an anesthetic drug. Etomidate was selected for these proof-of-concept studies because it has no active metabolites, it is relatively selective for GABA_A receptors, it is rapidly eliminated in vivo, and it causes minimal depressant effects on cardiac function and respiration. Etomidate is also a structural analog of the “next generation” of injectable anesthetic drugs.^31,32 

The mice were treated with dexmedetomidine (25 µg/kg, intraperitoneally) followed by etomidate (20 mg/kg, intraperitoneally). Twenty-four hours later, the mice were euthanized, and ex vivo hippocampal slices were processed for biotinylation analyses of surface proteins. Etomidate triggered a sustained increase in the cell-surface expression of α5 GABA_A receptors (146.4 ± 51.6% of control, 95% CI 107.4 to 185.6%; n = 8 each; fig. 1A) as shown previously.⁵  Cotreatment with dexmedetomidine reduced this increase (118.4 ± 39.1% of control, 95% CI 89.1 to 147.6%; n = 8). The total expression of α5 subunit protein was not significantly changed by either etomidate or dexmedetomidine (fig. 1B).

To determine whether dexmedetomidine prevented the increase in α5 GABA_A receptor function that persists after the anesthetic has been eliminated, a tonic inhibitory current was recorded from hippocampal neurons that were cocultured with cortical astrocytes. The design of these studies was based on our previous work showing that astrocytes were required to generate the anesthetic-induced persistent increase in tonic current in neurons.⁵  Cocultures of neurons and astrocytes were treated with etomidate (1 µM) and/or dexmedetomidine (0.1 µM) for 1 h and then washed, and 24 h later, the tonic current was recorded. Etomidate alone increased the amplitude of the tonic current, whereas no increase was observed in cultures cotreated with etomidate and dexmedetomidine (control: 0.68 ± 0.19 pA/pF, 95% CI 0.48 to 0.88, n = 6; etomidate: 1.44 ± 0.33 pA/pF, 95% CI 1.20 to 1.67, n = 10; etomidate + dexmedetomidine: 1.01 ± 0.45 pA/pF, 95% CI 0.66 to 1.36, n = 9; fig. 1C).

We next asked whether the most commonly used inhaled general anesthetic, sevoflurane, also increased the tonic current in hippocampal neurons and, if so, whether dexmedetomidine prevented this increase. Treating the cocultures with a clinically relevant concentration of sevoflurane (266 µM) increased the amplitude of the tonic current, and this increase was prevented by cotreatment with dexmedetomidine (fig. 1C).

Cognitive deficits after anesthesia also involve the function of the cortex. Thus, we studied whether etomidate triggered a persistent increase in tonic current in cortical neurons. Cortical neurons that were cocultured with astrocytes were pretreated with etomidate, as described above. The amplitude of the tonic current measured 24 h after treatment was increased. This increase did not occur in cortical neurons that were cotreated with etomidate and dexmedetomidine (fig. 1D).

Dexmedetomidine prevented the anesthetic-induced increase in tonic current in neurons by acting either: (1) directly on the neurons or (2) indirectly on the astrocytes to release factors that regulate GABA_A receptors in neurons. To distinguish between these two potential mechanisms, astrocytes and hippocampal neurons were cultured separately. First, the astrocytes alone were treated with etomidate (1 µM), and the neurons were treated with dexmedetomidine (0.1 µM). After 1 h, the cells were washed with drug-free medium, and 2 h later, the conditioned medium from the astrocytes was transferred to the dexmedetomidine-treated neurons. Neurons were incubated for 24 h, and then the tonic current was recorded (fig. 2A). Under these conditions, dexmedetomidine did not prevent the etomidate-induced increase in current. Thus, dexmedetomidine treatment of astrocytes was necessary to prevent the anesthetic-induced increase in tonic current in neurons.

To determine whether dexmedetomidine treatment of astrocytes was sufficient to reduce the tonic current in neurons, astrocytes alone were treated with etomidate (1 µM) or dexmedetomidine (0.1 µM) plus etomidate for 1 h, washed, and incubated for 2 h. The conditioned medium was then transferred to hippocampal neurons, and 24 h later, the tonic current was recorded in the neurons (fig. 2B). Medium from etomidate-treated, but not vehicle-treated, astrocytes increased the amplitude of the tonic current. Importantly, medium from astrocytes treated with both dexmedetomidine and etomidate yielded no such increase. Thus, dexmedetomidine treatment of astrocytes was sufficient to prevent the etomidate-induced persistent increase in tonic current.

Astrocytes from humans and mice share many phenotypic properties; however, their transcriptome, structural, and functional features can differ.³³  Therefore, we studied human astrocytes obtained from optic nerve heads to determine whether they also released soluble factors that increased the tonic current in mouse hippocampal neurons and, if so, whether treatment of human astrocytes with dexmedetomidine could attenuate such an increase.

Cultured human astrocytes were treated with etomidate and dexmedetomidine for 1 h and then washed. The conditioned medium was transferred to the cultured neurons 2 h later, and the tonic current was recorded 24 h later (fig. 2B). The conditioned medium from the etomidate-treated human astrocytes increased the tonic current; cotreatment with dexmedetomidine prevented this increase (fig. 2C).

Dexmedetomidine is a high-affinity α2 adrenergic receptor agonist. To determine whether activation of α2 adrenergic receptor was necessary for the protective effects of dexmedetomidine, the antagonist yohimbine (1 µM) was coapplied with dexmedetomidine (0.1 µM). Yohimbine blocked the dexmedetomidine attenuation of etomidate-induced increase in tonic current (fig. 2D). In addition, the less potent α2 adrenergic receptor agonist clonidine (1 µM) mimicked the effect of dexmedetomidine, because it prevented the etomidate-induced increase in tonic current in hippocampal neurons (fig. 2E). Collectively, these results suggest that dexmedetomidine activates α2 adrenergic receptors in astrocytes to reverse the etomidate-induced persistent increase in tonic current in neurons.

Activation of α2 adrenergic receptors in astrocytes stimulates multiple G protein–coupled receptor signaling pathways, including those involving extracellular signal–regulated protein kinases 1 and 2.^34,35  We reasoned that the increase in phosphorylated extracellular signal–regulated kinase could contribute to dexmedetomidine reduction of the tonic current in neurons. To investigate whether dexmedetomidine increased levels of phosphorylated extracellular signal–regulated kinase in astrocytes, cultured murine astrocytes were pretreated with dexmedetomidine (0.1 µM, for 1 h) and then washed, and 2 h later, the cells were processed for Western blotting analysis. The results showed that dexmedetomidine increased phosphorylated extracellular signal–regulated kinase but not total extracellular signal–regulated kinase (fig. 3, A and B).

To determine whether the dexmedetomidine-induced increase in phosphorylated extracellular signal–regulated kinase was involved in the signaling pathway that prevented the etomidate-induced increase in tonic current in hippocampal neurons, cultured astrocytes were cotreated with dexmedetomidine and PD98059 (50 µM), a compound that inhibits the production of phosphorylated extracellular signal–regulated kinase. Dexmedetomidine failed to prevent the etomidate-induced increase in tonic current in neurons (fig. 3C), suggesting that dexmedetomidine acted via a phosphorylated extracellular signal–regulated kinase 1/2–dependent pathway.

The results show that dexmedetomidine stimulates astrocytes to release one or more soluble factors that act in a paracrine manner to prevent the etomidate-induced increase in tonic current in neurons. These factor(s) were heat-sensitive because heating the conditioned medium from dexmedetomidine-treated astrocytes to 99.9°C for 5 min prevented the supernatant from triggering the etomidate-induced increase in tonic current in neurons (fig. 3D).

Dexmedetomidine stimulation of phosphorylated extracellular signal–regulated kinase could lead to an increase in the expression of genes that encode for neurotrophic factors such as brain-derived neurotrophic factor, which is known to modulate GABA_A receptor activity.^34–36  Indeed, treating the astrocytes with dexmedetomidine (0.1 µM) increased the expression of brain-derived neurotrophic factor mRNA (fig. 3E). Dexmedetomidine also increased brain-derived neurotrophic factor protein levels (fig. 3F).

To test whether brain-derived neurotrophic factor was sufficient to mimic the effects of the unidentified soluble factor, exogenous brain-derived neurotrophic factor (200 pg/ml) was applied directly to the neurons. Brain-derived neurotrophic factor prevented the etomidate-induced increase in tonic current (fig. 3G). Because brain-derived neurotrophic factor is known to act on tropomyosin receptor kinase B (TrkB) receptors, we next tested whether treating the hippocampal neurons with an antagonist for TrkB receptors (ANA 12, 0.01 μg/ml) abolished dexmedetomidine reduction of the etomidate-enhanced tonic current (fig. 3H). ANA 12 prevented the effect of dexmedetomidine, suggesting that dexmedetomidine stimulates astrocytes to release brain-derived neurotrophic factor, which in turn acted on hippocampal neurons to reduce the etomidate-induced increase in tonic current.

Finally, we asked whether dexmedetomidine prevented the behavioral deficits in memory and problem-solving that are observed in vivo after the anesthetic drug has been eliminated from the body. The novel object recognition task was selected and performed first because this test has been shown to be sensitive to an increase in α5 GABA_A receptor activity.⁵ 

The mice were treated with dexmedetomidine (25 µg/kg, intraperitoneally) or vehicle then etomidate (20 mg/kg, intraperitoneally) and 24 h later were studied using the novel object recognition task.³⁰  Notably, dexmedetomidine markedly prolonged the duration of etomidate-induced anesthesia. Specifically, dexmedetomidine alone did not cause the loss of righting reflex, but cotreatment of dexmedetomidine plus etomidate prolonged the duration of the loss of righting reflex by 32% (etomidate: 26.8 ± 7.9 min vs. etomidate + dexmedetomidine: 35.3 ± 7.6 min, n = 22; unpaired Student’s t test, P = 0.0007). These results are consistent with the anesthetic-potentiating effects of dexmedetomidine observed in patients.³⁷ 

The novel object recognition task measures the ability of mice to distinguish a novel object from a familiar object (fig. 4A). During the training phase, the interaction time of the mice was similar for the two objects (data not shown). During the testing phase, the mice that were treated with etomidate failed to distinguish the novel object, as shown by the low discrimination ratio (i.e., the time spent exploring the novel object relative to time spent exploring both objects; fig. 4B). Dexmedetomidine markedly attenuated the etomidate-induced deficits in memory performance. The cognition-sparing effect of dexmedetomidine could not be attributed to differences in locomotion or overall exploratory drive (fig. 4C).

Next, we used the puzzle box assay to investigate: (1) whether a single dose of etomidate caused deficits in problem-solving and, if so, (2) whether dexmedetomidine prevented such deficits. The mouse was placed in a well lit environment that offered access to the preferred, darkened “goal” box. Admission to the goal box was made progressively more difficult over a series of trials that spanned over 3 days (fig. 4D).

Performance of simpler tasks, including the underpass and burrowing tasks, did not differ between treatment groups (fig. 4E). Etomidate caused a deficit for the more difficult plug task, which was performed on day 3 as evidenced by an increase in the latency to complete the task (control: 76.4 ± 65.5 s, 95% CI 47.4 to 105.4, n = 22; etomidate: 131.4 ± 93.8 s, 95% CI 89.8 to 173.0, n = 22; P = 0.08; but P = 0.016 for log-rank pairwise comparison). This deficit was prevented in mice cotreated with dexmedetomidine because the latency was significantly reduced to 69.1 ± 86.6 s (95% CI 30.7 to 107.5, n = 22). Etomidate also prolonged the latency to complete the second burrow task on day 2, which was performed 2 min after the first burrow task (fig. 4F). These results suggest that etomidate impaired problem-solving at least in part by reducing short-term memory and that cotreatment with dexmedetomidine prevented this impairment.

The goal of this study was to use a reverse translational approach to identify the cellular and molecular mechanisms that contribute to the cognition-sparing properties of dexmedetomidine. The results show that dexmedetomidine activates α2 adrenergic receptors in astrocytes and stimulates the release of brain-derived neurotrophic factor, which acts as a paracrine factor to prevent excessive cell-surface expression and function of α5 GABA_A receptors in neurons. Behavioral studies confirmed that cotreatment with dexmedetomidine attenuates the postanesthetic deficits in memory and problem-solving (fig. 5). Therefore, administration of dexmedetomidine is a plausible pharmacologic strategy to reduce excessive cell-surface expression of α5 GABA_A receptors in neurons after general anesthesia.

Several converging lines of evidence indicate that α2 adrenergic receptors in astrocytes, but not those in neurons, mediate the effects of dexmedetomidine. First, α2 adrenergic receptors are abundantly expressed in cortical astrocytes.³⁸  Second, our current studies of cultured astrocytes and neurons show that dexmedetomidine treatment of isolated astrocytes is both necessary and sufficient to prevent the etomidate-induced increase in tonic current in neurons. Finally, the α2 adrenergic receptor agonist clonidine mimics the effects of dexmedetomidine, whereas the antagonist yohimbine abolishes these effects.

Our results show that dexmedetomidine activation of α2 adrenergic receptors in astrocytes increases the level of phosphorylated extracellular signal–regulated kinase 1/2. These kinases can translocate from the cytosol to the nucleus where they activate transcription factors that promote gene expression.^39,40  Such changes could lead to an increase in brain-derived neurotrophic factor mRNA levels and subsequent release of brain-derived neurotrophic factor. These findings are also supported by previous studies showing an increase in brain-derived neurotrophic factor in association with dexmedetomidine.^35,41  Furthermore, exogenous brain-derived neurotrophic factor mimics the effects of dexmedetomidine in that it reduces the etomidate-induced increase in tonic current in neurons. In contrast, inhibition of TrkB receptors blocks the effects of dexmedetomidine. Collectively, these data support a model in which dexmedetomidine stimulates the release of brain-derived neurotrophic factor from astrocytes, and the brain-derived neurotrophic factor acts on hippocampal neurons to prevent a persistent increase in tonic current (fig. 5).

The results of this study are important for several reasons. First, dexmedetomidine alone did not cause the loss of the righting reflex, but it did prolong the duration of etomidate-induced anesthesia by 32%. Despite prolonging the hypnotic properties of etomidate, dexmedetomidine had cognition-restoring effects in mice. Second, the data provide a plausible model to account for the cognition-sparing properties of dexmedetomidine. The results identify both α2 adrenergic receptors in astrocytes and α5 GABA_A receptors in neurons as potential therapeutic targets for treatments of postoperative delirium. Dexmedetomidine has limited clinical utility because of potential adverse cardiorespiratory effects and should only be administered by a skilled care provider in a highly monitored environment. Other α2 adrenergic receptor agonists may mimic the ability of dexmedetomidine to prevent postanesthetic cognitive deficits. For example, clonidine can be administered orally for sedation⁴²  and may serve as an alternative in some patients. In addition, new drugs that target the expression and/or function of extrasynaptic α5 GABA_A receptors may offer additional prevention and treatment options. Finally, exogenous brain-derived neurotrophic factor prevents excessive GABA_A receptor function and may be useful for patients who fail to tolerate α2 adrenergic receptor agonists.⁶  Interestingly, treatment with brain-derived neurotrophic factor was shown to be safe in a study of patients with amyotrophic lateral sclerosis.⁴³ 

Our study has several limitations. First, dexmedetomidine could trigger the release of additional paracrine factors that have not been identified yet regulate extrasynaptic GABA_A receptor activity. Second, we did not compare the effects of dexmedetomidine on pro–brain-derived neurotrophic factor versus brain-derived neurotrophic factor levels. Astrocytes release pro–brain-derived neurotrophic factor, a 32-kDa precursor protein that often has effects opposite to those of brain-derived neurotrophic factor.⁴⁴  For example, pro–brain-derived neurotrophic factor promotes cell death through activation of a low-affinity nerve growth factor receptor (p75),⁴⁵  whereas the 14-kDa protein brain-derived neurotrophic factor, promotes cell survival via activation of TrkB receptors.⁴⁶  The equilibrium between pro–brain-derived neurotrophic factor and brain-derived neurotrophic factor may be a determinant in the protective effects of dexmedetomidine. In addition, the effects of other commonly used anesthetic drugs, such as propofol or sedative benzodiazepines, were not studied but are of interest for future studies.

In summary, this study supports a model in which dexmedetomidine prevents postanesthetic cognitive deficits by reducing cell-surface expression and function of α5 GABA_A receptors in neurons through the release of brain-derived neurotrophic factor from astrocytes. The implications of these results extend well beyond the perioperative context, given that increased α5 GABA_A receptor function has been associated with several neurologic disorders including stroke, Alzheimer disease, depression, and inflammation-induced brain injury.^16,47–49  Dexmedetomidine and related drugs that target similar signaling pathway may help to prevent and treat these disorders.

Supported by foundation grant No. FDN-154312 from the Canadian Institutes of Health Research, Ottawa, Ontario, Canada (to Dr. Orser); by a Kirk Weber Award in Anesthesia from Sunnybrook Health Sciences Center, Toronto, Ontario, Canada (to Ms. Kaneshwaran); by funds from the Department of Anesthesia at the University of Toronto, Toronto, Ontario, Canada, and from the Canadian Anesthesia Research Foundation, Toronto, Ontario, Canada (to Dr. Avramescu); by Canadian Institutes of Health Research, Ottawa, Ontario, Canada, operating grant Nos. 119298 (to Dr. Ramsey) and 123448 (to Dr. Sivak); and by a Natural Sciences and Engineering Research Council, Ottawa, Ontario, Canada, Discovery grant No. RGPIN-2016-05538 and funds from the Canada Research Chair in Sensory Plasticity and Reconsolidation and the University of Toronto Center for the Study of Pain, Toronto, Ontario, Canada (to Dr. Bonin). The work was also supported by the Perioperative Brain Health Centre, Toronto, Ontario, Canada (http://www.perioperativebrainhealth.com).

The authors declare no competing interests.