Isoflurane produces antidepressant effects and induces TrkB signaling in rodents

A brief burst-suppressing isoflurane anesthesia has been shown to rapidly alleviate symptoms of depression in a subset of patients, but the neurobiological basis of these observations remains obscure. We show that a single isoflurane anesthesia produces antidepressant-like behavioural effects in the learned helplessness paradigm and regulates molecular events implicated in the mechanism of action of rapid-acting antidepressant ketamine: activation of brain-derived neurotrophic factor (BDNF) receptor TrkB, facilitation of mammalian target of rapamycin (mTOR) signaling pathway and inhibition of glycogen synthase kinase 3β (GSK3β). Moreover, isoflurane affected neuronal plasticity by facilitating long-term potentiation in the hippocampus. We also found that isoflurane increased activity of the parvalbumin interneurons, and facilitated GABAergic transmission in wild type mice but not in transgenic mice with reduced TrkB expression in parvalbumin interneurons. Our findings strengthen the role of TrkB signaling in the antidepressant responses and encourage further evaluation of isoflurane as a rapid-acting antidepressant devoid of the psychotomimetic effects and abuse potential of ketamine.

2 similar concentrations and protocol that was used with isoflurane. Body temperature was maintained with a heating pad throughout the treatment. Sham mice were kept in the induction chamber for 2 minutes without isoflurane. NBQX (2,3-Dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7sulfonamide, Tocris Bioscience, Bristol, UK) was injected (i.p., 10 mg/kg; dissolved in saline) 10 minutes before sham/isoflurane treatment. Mice were killed by cervical dislocation while still under anesthesia or following the described recovery periods (15 min or 24 h) after stunning with CO2.
Control mice were killed by cervical dislocation after stunning with CO2.

BDNF ELISA
For the ELISA analyses, NP-lysed brain samples (~100 μg) were acidified and processed as described 9 .
BDNF protein levels were quantified using a commercial kit according to instructions (Human BDNF Quantikine ELISA Kit, catalog # DBD00, R&D Systems Europe Ltd., Abingdon, UK) and normalized against total protein content.

qRT-PCR
Total Bdnf mRNA levels were measured from hippocampus samples using quantitative RT-PCR as described 10 . Briefly, extracted total RNA was treated with DNAse I mix (Fermentas GmbH, Helsinki, Finland) and then reverse transcribed using oligo(dT) primer and RevertAid First Strand cDNA synthesis kit (Thermo Scientific). The control reactions without reverse transcriptase were also performed. The amount of cDNA was quantified using Maxima SYBR green qPCR master mix (Thermo Scientific) by real-time PCR. Total Bdnf cDNA was amplified using the following primers: 5´-GAAGGCTGCAGGGGCATAGACAAA-3´ and 5´-TACACAGGAAGTGTCTATCCTTATG-3´.
For normalization GAPDH cDNA levels were analyzed with the following primers 5 -GGTGAAGGTCGGTGTGAACGG-3´ and 5 -CATGTAGTTGAGGTCAATGAAGGG-3´. C t values from each sample were obtained using the LightCycler 480 software (Roche Diagnostics Ltd.).

Spine analysis from fixed tissue
The density of dendritic spines was analyzed in fixed cortical sections from B6.Cg-Tg(Thy1- Images were obtained using a confocal microscope (Leica TCS SP5II HCS), and pyramidal neurons from the medial prefrontal cortex (mPFC) and somatosensory cortex (SSCx) were selected with the following criteria: intense fluorescence, soma located in layer V, and primary apical dendrite >200 μm long. We imaged the dendrites in three ~65 μm long segments (proximal/medial/distal). We distinguished different types of dendritic spines: (i) stubby (protrusion length <1.5 μm); (ii) mushroom (clearly visible head with a diameter >1.5 times the average length of the neck, and the total length of the protrusion <3 μm) and; (iii) filopodia/thin (the length of the protrusion >3 μm or non-headed 1.5-3 μm protrusion). Similarly, we analyzed extra-distal dendritic segments in the hippocampus and mPFC (see # symbol in Figure 5). The analysis was done by a person blind to the treatment.

In vivo two-photon microscopic imaging in awake mice
B6.Cg-Tg(Thy1-YFPH)2Jrs/J (Thy1-YFP, Jackson Laboratories, Bar Harbor, ME, USA) mice were used for the analysis of dendritic spine turnover in an awake in vivo imaging experiment. A cranial window with a 4 mm diameter circular glass coverslip was implanted onto the skull above the somatosensory cortex (SSCx) under a combination anesthesia of fentanyl (0.05 mg/kg), midazolam (5 mg/kg) and medetomidine (0.5 mg/kg) (i.p.). Three weeks post-surgery and one week prior to imaging the mice were trained for eight 2-hour sessions with their head fixed but otherwise freely moving in an airlifted flat-floored Mobile HomeCage (Neurotar Ltd., Finland). For imaging, animals were intravenously injected with Texas Red -tagged 70 kDa dextran tracer and placed under the Fluo View 1000MP multiphoton microscope (Olympus, Hamburg, Germany). Blood vessels were used to locate the selected dendritic segments throughout the imaging sessions. Stacks of the YFP-expressing dendrites displaying spines from layer 1 in the SSCx were acquired. Five different dendritic segments from 4 different areas were analyzed per animal to study the time-lapse spine turnover (n indicating the number of dendritic segments; see Ref. 11 ). The spine formation and elimination rates were calculated as the number of spines that have appeared or disappeared, respectively at a given time point compared to the previous time point relative to the total number of spines present at that or the previous time, respectively.

Immunohistochemistry and analysis of FOSB intensity in parvalbumin and somatostatin neurons
The tissue was processed as free-floating sections, washed in PBS, then incubated in 10% normal Image stacks from at least three sections per brain area per mouse were obtained using a Zeiss LSM 710 confocal microscope with a 20x objective. Single confocal planes (pinhole = 1AU) from the same depth of the tissue (2-3 μm below the upper surface) were processed using FIJI to analyze the expression fluorescence intensity of FosB in neurons. A macro function was written to automatically threshold the 1% of pixels with highest fluorescence intensity in parvalbumin or somatostatin channel.
We checked that these values effectively overlapped the profile of neurons expressing parvalbumin or somatostatin. Then, we used these profiles to mask the original image and analyzed the fluorescence intensity of FosB specifically within these subpopulations of interneurons. Finally, we subtracted the interneuron profiles from our original images and analyzed the remaining neurons with the 5% highest FosB fluorescence intensity, which therefore must consist predominantly of excitatory pyramidal neurons.

Electrophysiological recordings
At 24 hour after a 30 min isoflurane treatment, hippocampal slices (400 μm) were cut with a vibratome as described 12 . The slices were used 1-4 hours after cutting. After at least 1 hour recovery period at room temperature, the slices were transferred into an interface-type recording chamber (32°C) perfused

Forced swim test
For the forced swim test a mouse was placed into a glass cylinder (diameter 19 cm, height 24 cm) filled with water (21±1°C) to the height of 14 cm. The latency to the first immobility period (passive floating, when the animal was motionless or doing only slight movements with tail or one hind limb) and overall immobility was measured during the 6-minute testing period. The mice were tested in FST 15 minutes after the isoflurane anesthesia. The analysis was done from video recordings by a person blind to the treatment and to the genotype of the animals.

Open field test
The open field test was used to assess general locomotor activity. The test was performed for 30 min in an illuminated (300 lux) transparent acrylic cage (length 28.5 × height 8.5 × width 20 cm) (Med Associates). Interruptions of infrared photo beams were used to calculate the overall distance traveled (cm).

Animals
Male Wistar rats weighing 240-260 g at the beginning of each experiment were housed alone in a temperature controlled room (24±1°C) under standard laboratory conditions with free access to food and water and a 12 h light/12 h dark cycle (lights on at 7:00 a.m.). Procedures were conducted in conformity with the Brazilian Society of Neuroscience and Behavior guidelines for the care and use of laboratory animals, which are in compliance with international laws and politics.

Isoflurane treatment
A cylindrical chamber was used to the administration of isoflurane (30 cm high, 18,5 cm of diameter).
A suspended wad of cotton with 3 ml of isoflurane was used to induce anesthesia primarily. After 1 minute inside the chamber the rats were maintained under observation and anesthetized for total duration of 30 minutes (counting from the moment they entered in the chamber) using a cylindrical tube mask with a little wad of cotton. 0.1 ml of isoflurane was administered to the wad every five minutes, till the end of the 30 minutes.

Learned Helplessness
The Learned Helplessness test was performed as described 14  (24 h after the PT) the animals were anesthetized with isoflurane for 30 min in a chamber and the control animals were exposed (1 minute) to the same chamber without any trace of isoflurane.

Isoflurane administration
Isoflurane (Aerrane®, Baxter, France) in oxygen (2 l/min) was administered to the mice in their home cages to diminish the potential stress effect induced by a new environment. In the first 2 minutes, the mice were subjected to isoflurane 4%. The concentration of drug was decreased progressively to 3% for 2 min, then to 2% for 8 min and finally to 1% for 18 min. The control group was monitored in the same room for 30 min without receiving isoflurane. All cages were modified with a ceiling adapted for isoflurane administration.

Induction of neuropathic pain
Chronic neuropathic pain was induced by placing a cuff around the main branch of the right sciatic nerve 15,16 . Before surgery, mice were assigned to the experimental groups so that these groups did not initially differ for the mechanical nociceptive threshold or for the body-weight. Surgery was performed under ketamine (17 mg/ml)/xylazine (2.5 mg/ml) anesthesia (intraperitoneal, 4 ml/kg) (Centravet, Taden, France). The common branch of the right sciatic nerve was exposed and a 2 mm section of split PE-20 polyethylene tubing (Harvard Apparatus, Les Ulis, France) was placed around it for Cuff group (Neuropathic group). Sham-operated mice underwent the same procedure without cuff implantation (Sham group).

Nociceptive test
The mechanical threshold of hindpaw withdrawal was evaluated using von Frey filaments and the results were expressed in grams (Bioseb, Chaville, France) 16,17 . Mice were placed in clear Plexiglas® boxes (7 x 9 x 7 cm) on an elevated mesh screen. They were allowed to habituate for 15 min before testing. Calibrated von Frey filaments were applied to the plantar surface of each hindpaw until they just bend, in a series of ascending forces up to the mechanical threshold (0.4 to 10 grams). Each filament was tested five times per paw and the threshold was defined as 3 or more withdrawals observed out of the 5 trials 15,18 .

Novelty suppressed feeding (NSF) test:
We performed the NSF test at 12th hour after the isoflurane administration during the dark phase, under red light. The testing apparatus consisted of a 40 x 40 x 30 cm plastic box with the floor covered with 2 cm of sawdust. Twenty-four hours prior to test, food was removed from the home cage. At the time of testing, a single pellet of food was placed on a paper in the center of the box. Then, an animal was placed in a corner of the box and latency to eat the pellet was measured within a 5 minute period 15 . This test induces a conflict between the drive to eat the pellet and the fear of venturing in the center of the box 19 .

Statistical analysis
Results are represented as mean ± SEM (standard error of mean) unless otherwise stated. For statistical analysis two-sided tests including unpaired two-tailed Student's t-test, Mann Whitney U test (nonnormally distributed data), one-way analysis of variance (ANOVA), mixed model ANOVA, two-way ANOVA, Pearson's χ 2 test and Spearman's correlation test were used. Levene's test was used to define the equality of variances. If the variances differed significantly non-parametric test was used. Post hoc analysis was conducted with Tukey HSD or Dunnett's test. Statistically significant p value was set to ≤ 0.05. Outliers were considered as values differing more than 2x standard deviation from the mean of the group. To calculate the proper sample sizes we used power calculations (α=0.05 and power 0.80) and estimated the standard deviations and effect sizes based on our previous experience and literature.