Low-concentration sevourane inhalation in treating MK801-induced schizophrenia like disease in mice and a feasibility study of schizophrenia patients

GABAergic decits have been considered to associate with the pathophysiology of schizophrenia and hence GABA receptor subtype A (GABA A Rs) modulators may have therapeutic values for schizophrenia. Sevourane, a commonly used volatile anesthetic, enhances GABAergic neurotransmission through the GABA A R. The present study aims to investigate the therapeutic effectiveness of low-concentration sevourane in MK801-induced schizophrenia-like mice and amongst schizophrenia patients in a single arm trial. Three weeks after administration of MK801 (0.5 mg/kg, i.p. twice a day) for ve days, mice were exposed to 1% sevourane for 1 hr/day for 5 days. One week after treatment, they were subjected to behavioral tests, and then sacriced for immunohistochemical stain, western blot assay and electrophysiology recordings in the prefrontal cortex. Ten schizophrenia patients received 5-hr sevourane (0.5–1.2%) for 6 days, and were assessed with the PANSS and the BPRS-18 in the 1st and 2nd week after the treatments. MK801 induced hypolocomotion and social decits, downregulated the expression of NMDARs subunits, and postsynaptic density protein 95 (PSD95), reduced parvalbumin- and GAD67-positive neurons, and changed the amplitude and frequency of mEPSC and mIPSC and evenly increased the excitation/inhibition (E/I) ratio. All these changes induced by MK-801 were attenuated by sevourane administration. Schizophrenia symptoms assessed with the scales were signicantly improved in the 1st and 2nd week after treatments. Low-concentration sevourane inhalation effectively reversed MK801-induced schizophrenia-like disease in mice and alleviated schizophrenia patients’ symptoms. Our work suggested that sevourane may be a valuable therapeutic strategy for treating schizophrenia patients.


Introduction
Schizophrenia is a chronic and devastating psychiatric disorder. It is clinically characterized by both positive (i.e., hallucinations, delusions, abnormal motor behavior) and negative symptoms (i.e., avolition and sociality) as well as cognitive de cits, and affects up to 1% population world-wide 1 . It is ranked among the top 25 leading causes of disability globally and in icts substantial social, healthcare and economic burdens 2 . Antipsychotics are the standard medication for schizophrenia and, indeed, dopamine or/and serotonin modulating antipsychotics remain the primary approved treatment for schizophrenia 3 .
However, the majority of these antipsychotics are mainly effective for positive symptoms and leave most patients with negative symptoms and their cognitive dysfunction relatively untreated, and thus, patients have poor lifelong quality of lives 4,5 . In addition, due to their different a nity towards various synaptic receptors, these drugs are associated with nonadherence owing to their side effects including extrapyramidal (EP), cardiotoxicity and abnormal metabolic syndrome (hyperglycemia, hyperlipidemia, weight gain) make them not ideally for long term use 6,7 . The acute phase such as the rst episode, or an acute exacerbation of schizophrenia is a pivotal phase because optimal treatment for the episode might improve the long-term prognosis 8 . Early diagnosis and, prompt and noncoercive interventions to relieve the symptoms during acute episode are critical to prevent aggressive behavior and symptomatic escalation. Indeed, delayed treatments and poor treatment compliance are considered to be signi cant barriers for the effective management of psychiatric conditions and the establishment of therapeutic alliance between patients and healthcare providers. The consensus for schizophrenia treatment is that any ideal medication should have the properties of rapid onset of action, calmness without sedation, easy to administer, non-invasive, non-traumatic/non-coercive, with good safety pro le and favorable tolerability 9 .
γ-aminobutyric acid type A receptors (GABA A R) are a family of ligand-gated ion channels essential for the regulation of synaptic inhibition in the brain 10 . Aberrant GABA A R activities or dysregulation thereof are associated with various neurodevelopmental disorders, neurodegenerative diseases and psychiatric diseases, including autism, Alzheimer's, depression and schizophrenia 11,12 . With growing studies on γaminobutyric acid type A (GABA A ) receptors (GABA A R) accessory proteins and the molecules modulating receptor signaling in the central nervous system, targeting GABA A R and associated proteins might shed a light to novel therapeutic development for psychotic disorders. Sevo urane, a commonly used volatile anesthetic for surgical anesthesia, is a potentiator of γ-aminobutyric acid type A (GABA A ) receptors and can induce rapid and well-controlled sedation 13 . Because of these advantageous properties, sevo urane and other inhalational anesthetics are increasingly used as general sedatives for invasive diagnostic and therapeutic procedures, mechanical ventilation and management of agitation in intensive care unit. The concentration required for sedation is approximately one third (0.3 minimum alveolar concentration) of the concentration required for general anesthesia with less incidence of nausea and vomiting 14 . The present study aims to test our hypothesis that inhaling low concentration of sevo urane attenuates the abnormal behaviors and reverses the changes to GABA A R associated proteins in a mice model of MK801induced schizophrenia. We also evaluated the therapeutic bene t of sevo urane at low concentration in a preliminary group of schizophrenia patients.

MK801 caused a persistent body weight reduction in mice
Most studies have consistently described that MK801 administration results in decreased weight compared to controls in rodents, and this weight retardation can persist into adulthood. We therefore measured body weight on a daily basis during MK801 treatment and once a week thereafter till the P32.
Sevo urane treatment rescued Mk801-induced hypoactivity in the OFT OFT was performed 7 days after sevo urane exposure to investigate the spontaneous locomotor activity, with the example of trials in each group showed in Fig. 2A-C. One-way ANOVA showed a signi cant effect on total distance between three groups (Fig. 2D, F (2, 51) = 32.92, p < 0.001). Bonferroni's post hoc tests showed that administration of MK801 signi cantly decreased the total distance compared with that in the control group (Fig. 2D, p < 0.001). Sevo urane exposure (MK801 + Sev) signi cantly increased the total distance compared with that in the MK801 group (Fig. 2D, p < 0.001). There was no signi cantly difference between the control and the MK801 + Sev group (Fig. 2D, p = 0.090). Similarly, One-way ANOVA showed a signi cant effect on the frequency of crossing squares between groups (Fig. 2E, F (2, 51) = 22.35, p < 0.001). Bonferroni's post hoc tests showed that the MK801 group crossed fewer squares than the control group (Fig. 2E, p < 0.001), while the MK801 + Sev mice displayed signi cantly higher frequency of crossing squares compared with the MK801 group (Fig. 2E, p < 0.001).
Sevo urane treatment restored MK801-induced social interaction defects During the social approach phase, the test mice in three groups explored freely in each chamber, as the example of recording trials and density map shown in Fig. 3A-B (Ctrl: left; MK801: middle; MK801 + Sev: right, respectively). There was no difference in the time exploring the left chamber with an empty black wire cup inside (Fig. 3C), while the times exploring the center chamber (Fig. 3D) and the right chamber with a novel mouse inside the wire cup (Fig. 3E) were signi cantly different. Bonferroni's post hoc tests showed that MK801-treated mice spent more time lingering in the center chamber and less time in the right chamber than mice in the control and MK801 + Sev groups ( Fig. 3C-D). The time sni ng the novel animal revealed signi cant difference between three groups (Fig. 3F, F (2, 51) = 6.75, p = 0.003). The MK801-treated mice spent less time sni ng the novel animal than the control group (Fig. 3F, p = 0.003). The MK801 + Sev group mice spent more time sni ng the novel animal than the MK801 group mice

Sevo urane reversed MK801-induced aberrant NMDARs composition in the mPFC
The expression of NMDARs subunits including NR1, NR2A and NR2B within the prefrontal cortex were determined by western blot assay. There were no signi cant differences in the expression of NR1 in the three groups (Fig. 4A, B, Figure F (2, 15) = 0.52, p = 0.602). However, NR2A was differentially expressed between the three groups ( Fig. 4C, D, F (2, 15) = 14.21, p < 0.001). Bonferroni's post hoc test showed that the expression of NR2A was signi cantly decreased in the MK801 group compared to that of control group (Fig. 4D, p < 0.001), while the expression of NR2A in the MK801 + Sev group was higher than that of MK801 group (Fig. 4D, p = 0.021). Similarly, NR2B expression was different amongst three groups (Fig. 4E, F, F (2, 15) = 15.58, p < 0.001). Bonferroni's post hoc tests showed that the expression of NR2B was signi cantly decreased in the MK801 group compared to that of control group (Fig. 4D, p < 0.001), while the expression of NR2B in the MK801 + Sev group was signi cantly increased, as compared to that of MK801 group (Fig. 4F, p = 0.019). In addition, there was a signi cant difference of the expression of PSD95 among three groups (Fig. 4H, F (2, 15) = 9.44, p = 0.002). Bonferroni's post hoc tests showed that PSD95 expression was signi cantly decreased in the MK801 group compared to that of the control group

Sevo urane reversed MK801-induced alterations of electrophysiological pro les and E/I balance in the mPFC
To identify whether sevo urane treatment can reverse the MK801-induced alterations to electrophysiological characteristics of pyramidal neurons in the laminar II/III of the mPFC, whole-cell recordings were performed 2 weeks after sevo urane treatment at P49. The excitatory postsynaptic currents (mEPSCs) recorded in the mPFC pyramidal neurons of the control (Fig. 8A), MK801 (Fig. 8B) and MK801 + Sev (Fig. 8C) showed that there was a signi cant difference in the mean amplitude of mEPSCs in the three groups (Fig. 8G, p < 0.001). Dunn's multiple comparisons tests showed that the mean amplitude was signi cantly decreased in the MK801 group compared to that of control group (Fig. 8G, p < 0.001), while the mean amplitude in the MK801 + Sev group was signi cantly increased, as compared to that of MK801 group (Fig. 8G, p < 0.001). Unexpectedly, in the MK801 group, the frequency cumulative distribution plot showed a distinct leftward shift, with a signi cant increase in the mean frequency of miniature excitatory postsynaptic currents (mEPSCs) compared to the control (Fig. 8H, p < 0.001) analyzed with Dunn's multiple comparisons tests, while the mean frequency of mEPSCs was reversed in the MK801 + Sev group, as compared to the MK801 group (Fig. 8H, p = 0.005). In addition, analysis of the miniature inhibitory postsynaptic currents (mIPSCs) recorded in the mPFC pyramidal neurons of the control (

Clinical trial results
Nine out of ten enrolled patients (seven male and three female) completed all 6 trial treatments and because of postural hypotension after treatment, 1 patient was stopped to participate after received one treatment (a male patient-no 3 in Fig. 9). Data from all patients were included in the nal data analysis. The clinical and demographic characteristics for the patients and their baseline PANSS and BPRS scores are shown in appendix D: Table 2.

Primary e cacy
Eight patients achieved 30% decrease in PANSS total score from baseline to end of week 2 following sevo urane treatment (appendix E: Table 3). A total 6 patients out of 10 patients (full analysis set) achieved 30% decrease in PANSS total score from baseline to end of week 1 following sevo urane treatment. The BPRS-18 total score revealed a rapid decrease from the baseline to week 1 and week 2. The BPRS-18 total score signi cantly improved over time with a mean (± SD) change from baseline of -9.7 (± 10.60) at the end of week 1, and to -30.9 (± 19.51) at the end of week 2 (Fig. 9). One patient with disease history of 16 years received one electroconvulsive therapy given that the clinical symptoms didn't improve after three times of treatments.

Safety
No abnormal physiological parameters and no sevo urane-related side effects were found and patients tolerated the treatment well.

Discussion
The present study aims to investigate the potential therapeutic effects of low-concentration sevo urane inhalation on behavior de cits and perturbations to glutamatergic and GABAergic neurotransmission in a MK801-induced mice model that mimics negative features of schizophrenia. Our ndings were as follows: i) Mice treated with MK801 at 0.5 mg/kg (twice/day for 5 days) displayed hypolocomotion and impaired sociability; ii) Low-concentration of sevo urane reversed these behavioral de cits; iii) Lowconcentration of sevo urane attenuated neurochemical and electrophysiological alterations, and restored NMDA glutamatergic and GABAergic neurotransmission. In view of the pharmacological advantages of sevo urane including rapid onset, relative non-toxicity, noninvasive route of administration and lesser side effects than currently available antipsychotics, our ndings indicate that sevo urane may be a potential therapy for treating an acute episode and negative symptoms of psychotic disorders.
Previous studies suggest the glutamatergic and GABAergic neurons and their neurotransmitter systems play crucial roles in the pathophysiology of schizophrenia 12 ; but, currently, there are no relevant new medications available for clinical use. GABA A receptors are heteropentameric structures assembled from a large family of subunits including two α (α1-6), two β (β1-3), and either one γ (γ1-3) or one δ subunit 15 . The majority of GABA A receptors are comprised of α1, β2, and γ2 subunits located at GABAergic synapses via direct association with key scaffolding protein gephyrin which is essential for phasic GABA A R signaling 16 . Among the major neurotransmitter-gated ion channels or ionotrophic receptors, GABA A receptor is critical for the hypnosis action of general anesthetics. Volatile anesthetics (such as iso urane, sevo urane and des urane) enhance the response of the GABA A Rs to GABA, increase the duration of inhibitory response, and potentiate the duration of GABA-mediated synaptic inhibition 17 . In addition, there is 'direct activation' of GABA A R by opening the anion channel at higher concentrations in the absence of GABA 18 . We found that 1-hour 1% sevo urane inhalation for 5 days reversed the decreases in locomotion and sociability induced by MK801. OFT was adopted to access the exploratory behaviors and emotional disorders including anxiety-like and depression-like behaviors 19,20 , whilst an automated three-chambered social approach task was used for evaluating sociability 21 . Our results are in agreement with studies showing that administration of MK801 to neonatal animals induced reduction in spontaneous locomotor activity in long-term which may re ect psychomotor retardation 22,23,24 . For example, a consecutive 5-day administration of MK801 in neonatal Balb/c mice (P7) produced a negative model of schizophrenia including hypoactivity in OFT 22 .
Glutamate (Glu) and GABA, as the major excitatory and inhibitory neurotransmitter in the brain, respectively, are crucial in maintaining the normal brain function. Glutamate and GABA neurotransmitter systems play central roles in the pathophysiology of mood disorder. NMDAR antagonists, e.g., phencyclidine (PCP) and MK801, can induce symptoms of schizophrenia in normal human subjects 25,26,27 . NMDARs, which predominantly localize at the excitatory postsynaptic membrane and interact with the postsynaptic density 95 (PSD95), are ionotropic glutamate receptors comprised of two conserved NR1 subunits and two NR2 and/or NR3 subunits. The heteromeric NMDARs provide a wide variety of wellorganized composition of subunits to engage in speci c neuronal functions within different brain regions. Therefore, it is not surprising that disturbed composition, misplacement and abnormal tra cking of NMDAR subunits were found in pathological conditions, psychiatric diseases and developmental disorders 28 . NMDARs show rapid maturation with switching from NR2B-to NR2A-NMDARs during the early development in mice, and are prone to MK801-induced schizophrenia-like symptoms 29,30,31 . Indeed, in our study, sub-chronic administration of MK801 during rapid neurodevelopmental phase led to disturbed composition of NMDARs subunits, with downregulation of NR2A and NR2B subunits and the scaffolding protein PSD95, but NR1 subunit expression remained unchanged. A previous study found GluN1 and GluN2A downregulation in the prefrontal cortex of postmortem brain of patients with schizophrenia 32 . A few studies also reported NMDARs subunits alteration after MK801 systemic administration in rodents 33,34 . In addition, cell adhesion molecule neuregulin 1 (NRG1) and its neuronal tyrosine kinase receptor ErbB4 were widely expressed in the interneurons in schizophrenia subjects 35 . NRG1, NR2 subunits and ErbB4 share a common anchoring region on the postsynaptic sites and NRG1-ERBB4 signaling can affect NMDAR and its function 36 . Indeed, we found the expressions of NRG1, ErbB4, NR2A, NR2B and PSD95 were signi cantly decreased following MK801 administration and lowconcentration sevo urane reversed all these changes induced by MK-801.
Prefrontal cortex (PFC) plays a fundamental role in regulating multiple complex behaviors, including memory, cognitive, attention, social interaction and emotional regulation. The PV interneurons, the most common subtype of GABAergic neurons known for their fast-spiking phenotype, have a strong control over the excitability of pyramidal neurons by innervating the soma and proximal dendrites 37 . Therefore, PV interneurons are powerful regulators for maintaining optimal E/I balance within prefrontal local circuits for information processing. Previously, it has been demonstrated that disrupting E/I balance in the PFC via pharmacological, chemogenetic and optogenetic approaches induced a range of PFCdependent abnormalities associated with psychiatric diseases 38 . MK801 administration likely disrupts the E/I balance in pyramidal neurons in laminar II-III of the mPFC via the reduced GABAergic neurotransmission and hypofunctional NMDARs. Indeed, we found that both excitatory and inhibitory synaptic transmissions in the pyramidal neurons were impaired by MK801 administration, evidenced by the decreased amplitude of mEPSC and mIPSC, the increased frequency of mEPSC and as a result, the E/I ratio was increased. Low-concentration of sevo urane increased GABAergic neurotransmission and reversed these electrophysiological changes induced by MK801.
Our proof of concept study indicated that patients tolerated the treatment very well and symptoms were noticeably improved by the treatment. Most importantly, of these patients, ve patients with disease history over 10 years, and whose symptoms that failed to respond to routine anti-schizophrenia medications, were improved with this novel therapy. The most likely antipsychotic action of sevo urane maybe attributable to its ability to enhance GABAergic neurotransmission, the de cit of which underlies the pathophysiology of schizophrenia. However, other mechanisms may be also involved: 1) subanesthetic concentration (0.4 MAC) of sevo urane increases regional cerebral blood ow 39 , and lowconcentration of sevo urane promotes hippocampal neurogenesis 40,41 . 2) Bene cial interactions between GABAergic neurons and serotonergic neurons 42 . Dexmedetomidine, a highly selective and potent α 2 -adrenergic receptor (α 2 -AR) agonist, is widely used for induction. In this trial, it was used for initial sedative induction and then sevo urane can be delivered smoothly to participants. Whether the bene cial effects we found in our patients are likely to be due to the interactive effects of both sevo urane and dexmedetomidine remain unknown and warrant further study.
In summary, present study further substantiated the nding that MK801 administration to neonatal mice could induce a negative model of schizophrenia 43,44,45 . Speci cally 22, 23, 31 , a consecutive 5-day administration protocol of MK801 in neonatal mice (P7) induced negative symptoms of schizophrenia including hypoactivity and less sociability. These abnormalities could be explained by changes to NMDA receptor subtypes and GABAergic neurotransmitter, and the consequential excitatory/inhibitory neurotransmission imbalance in the pyramidal neurons in laminar II-III of the PFC. All these changes can be reversed by sevo urane administration. Furthermore, low-concentration of sevo urane inhalation effectively reversed MK801-induced negative symptoms of schizophrenia. The good tolerance and effectiveness of sevo urane in schizophrenia patients noted in the current study would encourage further large-scale clinical study to comprehensively evaluate the therapeutic values of sevo urane for schizophrenia patients.

1) Pre-clinical studies Subjects and Drugs
Subjects All experimental procedures were performed in accordance with the guidelines of the Animal Experiment Ethics Committees of …………………….. Timed-pregnant Balb/c mice were purchased from ……………………... They were housed in a temperature and humidity-controlled room (23 ± 1℃, 45-55%) of 12-hour light/dark cycle (lights on at 08:00 a.m.) with free access to water and food. The day of birth was designated as the postnatal day 0 (P0). On the P7, they (weighing 4-6 g) were randomly allocated into three groups: control group (Ctrl group), MK801 group, and MK801 + Sevo urane group (MK801 + Sev group). At the time of weaning (P21), mice were separated from their mothers and housed in group of four to six per cage.
Drugs administration MK801 (Ref. M107, St. Louis, MO, USA) was used to induce negative symptoms of schizophrenia as reported previously 43 . It was dissolved in 0.9% sterile saline. At the P7, pups received intraperitoneal (i.p.) injection of 0.5 mg/kg MK801 in the MK801group and MK801 + Sev group or an equal volume of saline (Ctrol group) twice a day (at 10:00 am and 15:00 pm) for ve consecutive days. Their body weights were measured every day during the treatment period and once/week since cessation of the treatments till the P31. During the P31-P35, the MK801 + Sev group mice were placed in a 25 cm x 20 cm x 15 cm plexiglass chamber, and 1% sevo urane was delivered via inlet of the chamber via a sevo urane vaporizer (Easy ll / Cagemount, R58S, ……………………..) at 1 L/min in 30% oxygen balanced with nitrogen for 1hr (10:00 am to 11 am) for ve consecutive days. The concentration of sevo urane was monitored with a gas monitor (BeneView T8, Mindray Bio-Medical Electronics Co.Ltd., ……………………..) via the outlet of the chamber. The MK801 and control groups received identical gases without sevo urane under the identical setting for 1hr for 5 days. Open eld test was performed at the P43-P45, and then three-chamber social test was conducted after two non-stimuli days at the P47. After behavioral tests (see below), they were sacri ced under terminated anesthesia and their brain samples were harvested for electrophysiological recordings, western blot and immunohistochemical analysis, respectively. The experimental timeline is presented in Appendix: Supplementary Fig. 1.

Behavioral test
In order to avoid possible behavioral testing biases, mice were tested in a random order and all trails were carried out in the day at the same time period (9:00-16:00). In addition, the apparatus was cleaned with 70% ethanol after each trail to eliminate any effects from olfactory perception.

Open eld test (OFT)
Open eld test was adopted for assessing spontaneous locomotor activity. Mice were gently placed in the center of an open-top apparatus with 25 equal squares (50 cm × 50 cm × 40 cm) and allowed to move freely for 15 min. Distances and traces of 15 min movements were recorded with a video camera and analyzed using EthoVision XT 8.0 (Noldus, Wageningen, Netherlands), the total distance travelled and the number of squares crossed during recording period were calculated for further analysis.

Three-chamber sociability test
The three-chamber test was conducted to assess social interaction with time spent in a side chamber with a novel mouse in a wire cup versus time spent in a side chamber with an empty identical wire cup 46 .
A rectangular plexiglass box (60 cm X 40 cm X 20 cm) without a top cover was divided into three chambers (20 cm X 40 cm X 20 cm) with two partitions. Between each chamber, there is a 5-cm opening hole which can be closed or opened with a lever operated door. The middle chamber was empty, while the side chamber contained an empty wire cup or an identical wire cup with a novel sex-and weight-matched mouse in. The test was divided into two phases. In brief, after a ten-minute habituation period (phase 1), the test mouse was placed in the middle chamber and allowed free access to visit each chamber. For the social approach phase (phase 2), the mouse was placed in the central compartment and left to explore with either an empty black wire cup (in left side chamber) or a similar wire cup with a novel mouse inside (the right side chamber). The apparatus and wire cups were thoroughly cleaned with 70% ethanol between phases and after each trial. Behaviors were videotaped by an automated tracking software (TopScan/ObjectScan, Cleversystems, Leesburg, VA, USA). The time spent in each compartment, and the time explored (sni ng) within a 2 cm vicinity of the cup with a novel mouse in was calculated to assess sociability.

Slice preparation
Slices of prefrontal cortex were prepared as previously described 47

Whole-cell recording
After incubation, slices were transferred to a recording chamber where perfused at 2-3 ml/min with ACSF, saturated with 95% O 2 and 5% CO 2 and maintained at 32 ± 2 °C. The pyramidal neurons of PFC layer II/III were viewed with an Olympus microscope equipped with infrared DIC optics. The patch recording pipettes (4-6 MΩ) were lled with a solution (in mM: 135 cesium methanesulfonate, 8 CsCl, 0.25 EGTA, 10 HEPES, 7 Na 2 -phosphocreatine, 0.34 Na 3 -GTP, 2.168 Mg-ATP, pH = 7.2-7.3, with CsOH) and access resistance of recorded cells was less than 30 MΩ. mEPSCs /mIPSCs were recorded at holding potential of -70 mV/0 mV in ACSF supplemented with 1 µM tetrodotoxin. Only cells with membrane potentials lesser than − 65 mV and series resistance below 25 MΩ were included for further analysis. Cells were excluded if input resistance changed 20% over the entire experiment. Whole cell recordings were conducted by using Multiclamp 700B ampli er (Axon Instruments). All data were collected with 2 kHZ Bessel lter at a 10 kHz sampling frequency (DigiDATA 1550A, Axon Insteuments), and analyzed by Clamp t 10.0 software (Axon Instruments) following low-pass ltering at 1000 Hz. The synaptic response belonging to the inhibitory and excitatory amplitude was used to determine the E/I ratio (mEPSC amplitude/mIPSC amplitude) of the recorded pyramidal neuron.
Western blot PFC samples (n = 6/group) were harvested for western blot analysis and dissociated in lysis buffer (containing protease inhibitors, 50 mM Tris-HCl, pH 7.6) on ice for 30 min and homogenized via ultrasoni cation (Ningbo scientz biotechnology CO. LTD, Ningbo, China). After centrifugation at 12,000 g for 10 min at 4 °C, supernatants were collected and protein concentrations were measured with a BCA assay kit (Beyotime Institute of Biotechnology, China). Equivalent amounts of protein samples mixed with gel loading buffer (50 mM Tris-HCl, 10% SDS, 10% glycerol, 10% 2-mercaptoethanol, 2 mg/ml bromophenol blue) were boiled for 5 min and loaded to SDS-PAGE gels. The separated proteins were electrophoretically transferred to polyvinylidene di uoride membranes. Then the membranes were incubated in blocking buffer (5% fat-free milk in Tween20) for 2hrs at room temperature and probed with temperature. The antigen-antibody complexes were detected by enhanced chemiluminescence system (Bio-rad) and visualized by a computer image system (GENE GNOME Chemiluminescence apparatus, Quantity one, Bio-Rad Laboratories). Image processing and semiquanti cation were performed with Image J software. Measurements were repeated independently at least 3 times for each experiment.
Density of each band was normalized to the internal controls (β-tubulin or GAPDH).

Statistical analyses
All data were presented as mean ± SEM, and statistical analyses were done with SPSS (SPSS Inc., Chicago, IL, USA). Treatment effects were statistically analyzed by one-way ANOVA followed by Bonferroni's post hoc tests for comparison when normality (and homogeneity of variance) assumptions are satis ed, otherwise Kruskal-Wallis test was applied to analyze the differences between groups followed by Dunn's multiple comparisons tests. Body weight was analyzed using a two-way ANOVA followed by Bonferroni's post hoc tests. A p value < 0.05 was considered to be of statistical signi cance.
2) An open-labelled single arm trial Patients' recruitment After the completion of pre-clinical experiments, we went to evaluate whether the therapy is effective in patients. An open-labelled single arm, proof-of-concept clinical trial was conducted (registered in ……………………..) from December 24, 2019, to January 20, 2020. After obtaining approval from Ethic committee of …………………….., and written informed consent from patient or family member, 10 schizophrenia patients (7 men and 3 women; age 18-65 years), who were taking antipsychotics and experiencing an acute exacerbation of psychosis and met all the inclusion criteria (appendix A: Table 1), were recruited (appendix B: Flowchart).

Clinical endpoints and outcome measures
The primary endpoint was the percentage of the early response at week 2, and the clinical response was de ned as a minimum 30% reduction in Positive and Negative Syndrome Scale (PANSS) total score. The percent change of PANSS total score was de ned as a change /(baseline − 30) x100% 48 . The secondary endpoints were the change of Brief Psychiatric Rating Scale (BPRS-18) from the baseline to week 1 and week 2, and the early response rate at week 1. Safety and tolerability assessments included treatmentemergent adverse events (TEAEs), clinical laboratory tests, electrocardiograms and physiologic measures.

Trial protocol
Recruited participants received routine antipsychotic medications (in line with standard clinical guidelines; one or two antipsychotics of risperidone, paliperidone, aripiprazole and olanzapine). After fasting for 8hrs and establishing intravenous line, dexmedetomidine, a sedative, was intravenously administrated via an infusion pump with a loading dose of 0.5 µg/kg for 15 min and then maintained at the rate of 0.1-0.3 µg/kg/h. Sevo urane was delivered at 2.5 L/min in 100% oxygen for 5hrs using a face mask under spontaneously respiration. Sevo urane concentration was started with 6.0% for rst 5-min duration to induce a quick sedation, and then gradually decreased to 0.5-1.5% and adjusted to maintain the deep sedation. Participants were closely monitored with electrocardiography, pulse oximetry, noninvasive blood pressure, bispectral index (BIS), end-tidal partial pressure of carbon dioxide and temperature throughout. This treatment was repeated for 6 times with intervals of 1-2 days for a total of two weeks (see the detailed protocol in the appendix C).

Statistical Analysis
Simon's two stage design was used in this proof of concept study. Based on ndings of an earlier study, the percentage of patients with at least 30% improvement in total PANSS at 2 weeks was about 30% 49 . In this study, a 70% or higher early response rate was set as the target of "good" and 30% was set as "poor".
The optimal two-stage design to test the null hypothesis that P < = 0.300 versus the alternative that P > = 0.700 has an expected sample size of 6.08. After testing the drug on 2 patients in the rst stage, the trial was planned to be terminated if 0 respond. If the trial goes on to the second stage, a total of 10 patients was planned to be studied. If the total number of patients responding is less than or equal to 5, the drug was rejected.
The e cacy and safety analysis were conducted in the full analysis set, which included all patients who received at least one treatment of sevo urane inhalation and had at least one evaluation post-baseline.
The early response of BPRS-18 total score at week 1 and week 2 was reported as mean with standard deviation.      showing high resolution image of anti-GAD67 immuno uorescence staining and DAPI taken for analysis.
(D) Statistic analysis showed a signi cant decrease in GAD67-positive interneurons in MK801 group compared to the control (Ctrl), and after sevo urane inhalation treatment, GAD67-positive interneurons were signi cantly increased compared to the MK801 group. One-way ANOVA followed by Bonferroni's post-hoc tests was used for analysis. Data are represented as mean ± SEM (n = 7); *p < 0.05; **p < 0.01, ***p < 0.001. among three groups. One-way ANOVA followed by Bonferroni's post-hoc tests was used for analysis.