General Anesthetics Regulate Autophagy via Modulating the Inositol 1,4,5-Trisphosphate Receptor: Implications for Dual Effects of Cytoprotection and Cytotoxicity

General anesthetics are both neuroprotective and neurotoxic with unclear mechanisms. General anesthetics may control cell survival via their effects on autophagy by activation of type 1 inositol triphosphate receptor (InsP3R-1). DT40 or SH-SY5Y cells with only or over 99% expression of InsP3R-1 were treated with isoflurane or propofol. Cell viability was determined by MTT reduction or LDH release assays. Apoptosis was determined by measuring Caspase-3 or by TUNEL assay. Autophagy activity was determined by measuring LC3 II and P62. We evaluated mitochondrial integrity using MitoTracker Green and cytosolic ATP levels. Fura2-AM was used to measure the concentrations of cytosolic calcium ([Ca2+]c). Propofol significantly increased peak and integrated calcium response (P < 0.001) in cells with InsP3R-1 but not in cells with triple knockout of InsP3R. Both propofol and isoflurane increased autophagy induction (P < 0.05) in an mTOR- and InsP3R- activity dependent manner. Short exposure to propofol adequately activated InsP3-1 to provide sufficient autophagy for cytoprotection, while prolonged exposure to propofol induced cell apoptosis via impairment of autophagy flux through over activation of InsP3-1. Propofol damaged mitochondria and decreased cytosolic ATP. The effects of general anesthetics on apoptosis and autophagy are closely integrated; both are caused by differential activation of the type 1 InsP3R.

activate InsP 3 R 7 , it is reasonable to propose that GAs may control cell survival fate by affecting autophagy via differential activation of InsP 3 R.
Interestingly, newborn developing brains in various animal models and in some clinical studies have demonstrated a high sensitivity and vulnerability to anesthetic neurotoxicity 8 , with an unclear mechanism. One proposed mechanism is the disruption of intracellular Ca 2+ homeostasis via excessive activation of InsP 3 R or ryanodine receptors (RYR) 9 . Notably, the expression of InsP 3 R may be associated with the maturation of neurons 10 . Apoptosis may also be significantly different between immature neurons and mature neurons because of differing autophagy regulation 11 . It is possible that differential effects of GAs on autophagy via activation of InsP 3 R may account for the high vulnerability of the developing brain to GA-mediated neuronal injury or death.
In various cell culture models, we have demonstrated that low concentrations of isoflurane or propofol for short durations stimulate cytoprotective autophagy by an mTOR-dependent pathway via adequate activation of InsP 3 R-1. However, GAs at high concentrations for prolonged use can impair autophagy flux and cause cell death via over-activation of InsP 3 R-1. Our study discloses a novel molecular mechanism for GAs' dual effects of cytoprotection and cytotoxicity, which may provide a basis for better utilization of GAs.
Elevation of [Ca 2+ ] c via activation of InsP 3 R has been shown to release Beclin-1 from BCL-2 15 , which induces autophagy via a mTOR-dependent pathway by sequential activation of phosphorylated AMPK, inhibition of mTOR function and subsequent stimulation of autophagy 6 . As both isoflurane 7 and propofol ( Fig. 1) can activate InsP 3 Rs and raise [Ca 2+ ] c , we examined if those commonly used GAs (isoflurane vs. propofol) affect autophagy  2+ ] c ), which was indirectly reflected by measuring the 340/380 ratio using the Ca 2+ indicator dye Fura-2 AM in DT40 WT or TKO cells (a-c) or SH-SY5Y cells (d-f). Compared to DT40 chicken lymphocyte WT cells, TKO cells with triple knockout of InsP 3 R were resistant to pfl-mediated peak elevation in [Ca 2+ ] c (a-c). TKO (a-c) or BATPA-AM at 10µM (d-f) dramatically attenuated pfl-mediated peak elevation of [Ca 2+y ] c (b and e), as well as the average integrated Ca 2+ responses calculated as the area under the curve (AUC) (c and f), as an increase above its own baseline before addition of propofol at 200 µM. Statistical analysis was performed for cytosolic Ca 2+ level at the time of peak elevation and AUC in DT40 (b and c) or in the presence or absence of BAPTA-AM in SH-SY5Y cells (e and f). All values represent Mean ± SEM from the average of at least 30 cells measured from a minimum of four separate experiments (N ≥ 4). **** means P < 0.0001 (b,e,f), *** means P < 0.001 (c).
Adequate autophagy is protective against propofol-induced cytotoxicity. Because autophagy activity plays an essential role in determining cell survival fate 18 , we explored the effects of GAs on cell death via autophagy. Previous studies have suggested that adequate autophagy is cytoprotective against various types of detrimental stresses 19 . Although propofol caused dose-dependent cytotoxicity in SH-SY5Y cells, general inhibition of autophagy by 3-MA significantly increased this propofol cytotoxicity ( Excessive autophagy or impairment of autophagy flux promotes GA-mediated cytotoxicity. When SH-SY5Y cells were differentiated into neurons, they were much more sensitive to propofol-induced cytotoxicity, as measured by MTT release assay, even at concentrations as low as 5 µM, a clinically relevant plasma concentration, compared to non-differentiated SH-SY5Y cells (Fig. 4a, 96.2 ± 1.3 vs 86.9 ± 2.4, P < 0.05). Additionally, propofol-induced cytotoxicity is not only dose-dependent ( Fig. 4a) but also time-dependent ( Fig. 4b), which is consistent with the toxic effects of other inhalational GAs 13 . It seems that propofol-mediated autophagy and apoptosis are closely related to each other because the biomarkers for apoptosis (TUNEL) and for autophagy stimulation or flux impairment (elevation of LC3II) increased in parallel after propofol treatment of propofol in WT DT40 cells (Fig. 4c, 13.9 ± 1.0 vs 27 ± 2.2, P < 0.05 and 4d, 5.9 ± 0.6 vs 16.1 ± 1.1, P < 0.0001). Similarly, excessive stimulation of autophagy by rapamycin in SH-SY5Y differentiated neurons significantly potentiated propofol-induced cytotoxicity (Fig. 4e, P < 0.01 or 0.001) and apoptosis ( Fig. 4f, P < 0.05). The propofol-mediated increase of LC3-II could represent either the stimulation of autophagy or impairment of autophagy flux, especially the conversion of autophagosomes to autolysosomes.
We further determined that propofol at high concentrations (200 µM, Fig. 4g) elevated LC3-II with concomitant use of bafilomycin, an inhibitor of lysosome function and autolysosome formation 21 and an agent capable of differentiating the causes of LC3II elevation. Bafilomycin is a relatively selective agent that impairs autophagy flux by itself. If propofol increases LC3-II by stimulation of autophagy induction, addition of bafilomycin will increase LC3-II further more. Otherwise it will indicate that propofol impair autophagy flux. Propofol did not further increase LC 3 II even in the presence of bafilomycin (Fig. 4g, P < 0.01), suggesting that propofol at a high concentration (200µM) for a prolonged duration (24 h) increased LC3-II by impairment of autophagy flux rather than induction of autophagy activity, further worsening propofol-mediated apoptosis (Fig. 4c).
In contrast to its stimulatory effect on autophagy through inhibition of mTOR after short exposure ( Fig. 2h and i, 2.5 hr), propofol at the same concentration (200 µM) for a prolonged exposure (24 h) inhibited autophagy induction by increasing activation of mTOR via activation of InsP 3 R, as evidenced by the increased mTOR product p70-56 k only in DT40-R1 but not in TKO cells (Fig. 4h, P < 0.05). These results suggest that excessive propofol exposure may have negative feedback activation on mTOR function and can inhibit autophagy induction via over-activation of InsP 3 R-1.
p62 is an ubiquitin-and LC3-binding protein that is degraded by the autophagy-lysosome pathway and is accumulated during deficient autophagy or during impaired autophagy flux 22 . Similar to its dual effects on mTOR activity (Figs 2h and i vs. 4h), propofol at 200 µM for 2 or 24 h affected the p62 level differentially, Figure 2. Isoflurane and propofol stimulate autophagy in an InsP 3 R-and mTOR-dependent pathway. In DT40 chicken lymphocyte cells, autophagosome biomarker LC3II protein determined by Western Blot was significantly increased under the 200 µM propofol (Pfl) treatment for 2.5h in DT40 cells with expression of only type 1 InsP 3 R (R1), but not as much in DT40 with triple knock out of InsP 3 Rs (TKO) cells (a). (b,c) DT40 WT or TKO cells stably overexpressing the mRFP-GFP-LC3 constructs were treated with 2.4% isoflurane (Iso) for 24 h. Baseline autophagy activity (both autophagosome and autolysosome) increased in TKO compared to WT cells. 2.4% Iso for 24 h significantly increased condensed LC3-II in both autophagosomes (b) and autolysosomes (c) in WT but not in TKO cells. Data: Mean ± SEM, one-way ANOVA followed by Dunnett's post hoc test. *P < 0.05, **P < 0.01, ++ P < 0.01, # P < 0.05, ## P < 0.01 compared to wild type control (WT). In human SH-SY5Y neuroblastoma cells, propofol treatment for 2.5 hours at indicated concentrations dose-dependently increased the autophagosome biomarker LC3II (d), starting at a pharmacological concentration of 200 µM, and could be significantly augmented by a 30 minute pretreatment with the autophagy stimulator rapamycin at 1 µM (e), but inhibited by siRNA of type 1 InsP 3 R (InsP 3 R-1) (e), 30 min pretreatment of InsP 3 R inhibitor xestospongin C at 2 µM (f) and BAPTA-AM at 10 µM (g). Propofol dose-and InsP 3 R-1-dependently inhibited mTOR by demonstrating a proportional decrease of the mTOR enzyme product p-S6K in SH-SY5Y cells (h) and DT40 cells with expression of type 1 InsP 3 R but not in TKO cells (i). Data represent Mean ± SEM from a minimum of 2 repeats of 3 separate experiments (d-i, N ≥ 3), one-way ANOVA followed by Dunnett's post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001. # P < 0.05. decreasing at short exposure but increasing at prolonged exposure ( Fig. 4i 1.17 ± 0.15 vs 0.74 ± 0.19, P < 0.05 and 4j, 2.1 ± 0.21 vs 2.6 ± 0.13, P < 0.05). Therefore short exposure of propofol increased autophagy stimulation via a mTOR-dependent pathway ( Fig. 2h and i) and promoted autophagy flux, providing cytoprotection ( Fig. 3), while prolonged propofol exposure impaired autophagy flux ( Fig. 4g and h) and thus decreased the turnover of p62 via the autophagy-lysosome pathway (Fig. 4j), inducing cytotoxicity.
Anesthetics impaired autophagy by disruption of mitochondria and lysosome function. As mitochondria and lysosome function are integral to autophagy 23 and cell death 24 , we investigated if GA effects on mitochondria and lysosome function influence autophagy. In SH-SY5Y cells, propofol at prolonged treatment (24 h) obviously disrupted mitochondria structure in a single cell, as shown by mitochondria green staining (Fig. 5a). As for mitochondria function, propofol for prolonged (24 h) but not short (4 h) exposure significantly decreased ATP production, an indirect biomarker for impaired mitochondria function, in the cytosolic space (Fig. 5b, (100 ± 16.4 vs 46 ± 3.85, P < 0.05), consistent with the dual effects of propofol on cell damage (Fig. 3 vs. Figure 4) and autophagy (Fig. 4i,j). Excessive activation of InsP 3 R may deplete ER Ca 2+ and therefore impair protein maturation inside the ER, like that of V-ATPase, a hydrogen pump that transports H + from the cytosolic space into the lysosome lumen to maintain a low intraluminal pH 25 . We determined if excessive use of propofol impairs autophagy flux by disrupting lysosome function via increased pH inside the lysosome. Using lysosensor staining to monitor lysosome pH 26 , we demonstrated that propofol at prolonged exposure (24 h) but not short exposure (5 h) elevated lysosomal pH with robust reduced lysosensor signal (Fig. 5c). Eventually, the disruption of lysosome function will impair the turnover of autolysosomes and disrupt the normal function of autophagy. We can demonstrate this disruption by an increase of the lysosomal protein LAMP1, an important biomarker for autophagy flux from autophagysomes to autolysosomes 27 . Propofol at prolonged exposure significantly increased the LAMP1 protein level (Fig. 5d, 1.07 ± 0.08 vs 2.35 ± 0.23, P < 0.05), indicating decreased autolysosome turnover and impaired autophagy flux and providing a distinct mechanism for propofol-mediated elevation of LC3II in both tissue culture and animal studies (Fig. 4d,g,i,j,l).
Proposed mechanism of dual effects of cytoprotection and cytotoxicity by GAs by their differential effects on autophagy via activation of InsP 3 R. Moderate elevation of [Ca 2+ ] c by short exposure of general anesthetic via adequate activation of InsP 3 Rs may increase physiological autophagy activity via a mTOR-dependent pathway (Figs 2 and 6) and provide cytoprotection (Figs 3 and 6). Inhibition of this adequate autophagy by 3-MA or by impairing adequate elevation of [Ca 2+ ] c may promote GA-caused cell death (Figs 3  and 6). On the other hand, prolonged use of general anesthetics at high concentrations may cause over activation of InsP 3 R and abnormal elevation of [Ca 2+ ] c , which may result in mitochondrial overload of Ca 2+ , disruption of mitochondria structure, collapse of mitochondrial potential, and reduction of normal ATP production (Fig. 5). Over-activation of InsP 3 R and the subsequent abnormal Ca 2+ release from the ER may impair lysosome function (Fig. 5) in the following two ways: 1). Increased pH inside the lysosome's lumen; 2). Impairment of autophagosome  (24 h,j) trended to decrease P62 degradation measured by Western Blot, indicating dual effects of propofol for both increasing autophagy activity at short exposure and impairing autophagy flux from autophagosome to autolysosome at prolonged exposure. Data represents Means ± SEM from a minimum of 4 separate experiments (a-j). One-way (f,g,j) or two-way (a,d,e) ANOVA followed by Dunnett's post hoc test or student t test (c,d,i,j) *P < 0.05; **P < 0.01, ***P < 0.001 compared to corresponding control respectively.  . All these effects individually or coordinately impair autophagy flux and promote cell death or neurodegeneration (Figs 4, 5 and 6).

Discussion
Isoflurane alone at clinically relevant concentrations activates InsP 3 R and releases Ca 2+ from the ER, increasing cytosolic and mitochondrial Ca 2+ concentrations 7,13,14 . Indirect evidence suggests that sevoflurane and desflurane also activate InsP 3 R 14 . Intravenous anesthetic propofol, too, can activate InsP 3 R (Fig. 1). Likely, propofol activates the isoform InsP 3 R-1 because the neuroblastoma cells (SH-SY5Y) have primarily (>99%) InsP 3 R-1 (Fig. 1b) 29 . Considering InsP 3 R-1 is the primary subtype of InsP 3 R, an essential player in the regulation of cell survival, GAs may exert their effects on cell survival via activation of InsP 3 R-1 in neurons.
Anesthetics like halothane activate the other major ER Ca 2+ release pathway, the RyR channel complex 30 , which is thought to be the basis for the dangerous clinical occurrence of malignant hyperthermia 31 . Like InsP 3 Rs, RyRs are expressed throughout the nervous system and play important roles in both normal cell function and in neurodegenerative disease 32 . Our previous study indicated the involvement of RyR activation in isoflurane-induced apoptosis in neuronal tissue cultures 33 . Thus, both InsP 3 R and RyR contribute to regulating intracellular Ca 2+ homeostasis and may interact with each other through a common Ca 2+ -induced Ca 2+ release pathway in normal neuronal function and neurodegeneration.
Although general anesthetics have been shown to be both neuroprotective and neurotoxic in both in vitro and in vivo models depending on the degrees of exposure 9,34,35 , the underlining mechanisms are unclear. The proposed mechanisms for anesthetic neuroprotection include: adequate preconditioning, generation of reactive oxygen species (ROS), activation of G-protein coupled receptors, moderate elevation of [Ca 2+ ] c , etc. The proposed mechanisms for anesthetic neurotoxicity include: excessive and abnormal activation of NMDA and GABA A receptors, generation of free radicals, activation of P 75 growth factors, inflammation and excessive elevation of [Ca 2+ ] c , etc. 36 . Comparing these two processes, anesthetics often use similar mechanisms to be both neuroprotective and neurotoxic, with the degree of exposure and the level of activation on different pathways serving as the determining factor 9,35 . Like ischemic preconditioning 37 , adequate exposure of GAs may induce various endogenous neuroprotective pathways and provide neuroprotection, while excessive exposure to GAs may be overstep the limit of preconditioning and stress neurons enough to cause damage.
Regulation of intracellular Ca 2+ homeostasis and its role in cell survival is quite complex. The rise of [Ca 2+ ] c is required for both cytoprotection and cytotoxicity of GAs 9 . The difference between the two is the degree and duration of [Ca 2+ ] c elevation. Concerning the degree of elevation, previous research suggested that a moderate elevation of [Ca 2+ ] c (likely an increase from around 70 to 150 nM) is cytoprotective 38 and an excessive elevation of cytosolic Ca 2+ is cytotoxic, such as the glutamate excitotoxicity after cerebral ischemia (which is often multiple times above baseline) 39 . As for [Ca 2+ ] c elevation duration, short exposure may be cytoprotective and prolonged exposure may be cytotoxic 9 . While the principle governing GAs' dual effects on neuroprotection and neurotoxicity may be clear, the exact tipping point from neuroprotection at adequate GA exposure to neurotoxicity at excessive exposure may be varied among various types of cells, animal species and patients. Nevertheless, if this theory in preclinical studies is proven to be true in patients (partial support can already be found in some retrospective clinical studies where multiple rather than single anesthesia exposure was associated with learning or developing disorders 40 , it is critically important to delineate the safe GA doses and exposure durations in different patients for future guidelines of clinical practice. For now, it is reasonable to avoid excessive anesthesia exposure whenever possible, especially in patients vulnerable to anesthesia-mediated neuronal injury or death (e.g. the developing brains of children).
Activation of InsP 3 R may either inhibit autophagy via a mTOR-independent pathway 17 or activate autophagy via a mTOR-dependent pathway 6 . Although a few previous studies 41 have investigated or discussed GA effects on autophagy, the exact role of GA-mediated regulation of autophagy in cell survival or death and neurodegeneration in the developing brains was far from clear, especially the effects of GAs on intracellular Ca 2+ homeostasis. In this pioneer study, we first demonstrated that GAs can either adequately stimulate autophagy and support cell survival or pathologically impair autophagy flux and cause neuronal injury or death via differential activation of type 1 InsP 3 R (Fig. 6), providing a novel mechanism for the detailed mechanisms of GA-mediated dual effects of neuroprotection and neurotoxicity. Current experiments have demonstrated that short or adequate exposure of propofol or isoflurane promote physiological autophagy and inhibit cell death via moderate activation of type 1 InsP 3 R and elevation of cytosolic Ca 2+ concentration. So the adequate anesthetic exposure provide cytoprotection or neuroprotection. However, prolonged use of propofol or isoflurane may impair normal autophagy flux or function via and promote cell death via over activation of InsP 3 R and abnormal elevation of cytosolic Ca 2+ concentration. Therefore, the current study provide a molecular mechanism explaining the dual effects of neuroprotection and neurotoxicity by propofol or isoflurane, which may be representive of intravenous or inhalational general anesthetics respectively.
Previous research has suggested that various GAs may have differing potencies to cause cell injury and death 14,42 . It seems that propofol is generally less potent in causing cytotoxicity than inhalational anesthetics 43,44 . From this study, we gather that propofol needs clinically non-relevant high concentrations and durations to cause cytotoxicity or changes in autophagy activity. In contrast, isoflurane can usually cause similar changes at clinically relevant concentrations and long durations. If confirmed in animal studies and in humans, propofol at clinically used concentrations and durations may be generally less potent than isoflurane in causing neuronal injury or death.
This study is clinically important, urging the construction of strategies to minimize the use of GAs so that they do not adversely impair autophagy function and result in brain or other organ damage. Based on the tissue culture data from this study and others, the toxicity to propofol usually starting at concentrations much higher than the clinical relevant concentrations. We assume that clinically used the propofol likely provides a plasma concentration that is cytoprotective rather than toxic, although this need to be confirmed in the future clinical studies. Isoflurane seems to be more toxic than propofol as it can cause cytotoxicity at concentrations within clinical relevant range, although frequently need long duration. Clearly, more clinical studies are urgently needed to compare relative possible unwanted neurotoxic effects between propofol and commonly used inhalational anesthetics and make sure safe concentrations of general anesthetics can be used to improve safety, especially in the developing brains which may be more vulnerable to general anesthetics mediated neurodegeneration.
Previous studies have demonstrated that propofol or isoflurane can be neuroprotective against brain damage after cerebral ischemia by promoting the physiological autophagy [45][46][47] , which is consistent to the results from this study. This in vitro study, together with other studies in tissue cultures 48,49 , provides molecular mechanism of neuroprotection by general anesthetics via promoting autophagy via adequate activation of InsP 3 and ryanodine receptors. Isoflurane and propofol have also shown to be neurotoxic by themselves in the developing brains [50][51][52] but the molecular mechanism through their effects on autophagy flux is not well studied. Current in vitro study suggest importance of dose and duration of general anesthetics on fate of either neuroprotection or neurotoxicity via regulation of autophagy, which need further extensive studies in both animals and patients.
This study has the following limitations: 1) Most of the propofol concentrations used are pharmacological rather than clinically relevant and less valuable to be used guiding clinical practice directly. Further studies in animals and humans are needed. 2) We have only determined the major points of the mTOR-dependent autophagy pathway induced by GAs but not the detailed points (Fig. 6), but this pioneering study will inspire more research work in the area of GA-mediated regulation of autophagy and its relationship with GA-induced cell death. 3) We primarily provide a correlative connection between GA-mediated regulation of autophagy and its effects on cell survival. More selective inhibition of autophagy such as siRNA or knock out of important autophagy regulators (e.g. mTOR, Beclin-1) may be needed, rather than just 3-HK or rapamycin, to confirm the role of autophagy in GA-mediated cell death in the future studies.
In summary, our results indicate that an adequate exposure level of isoflurane or propofol stimulates physiological autophagy via an mTOR-dependent pathway and provides cytoprotection. But, an excessive use of GAs impairs autophagy flux and function and promotes neuronal injury or death. A reasonable strategy to use GAs safely is to minimize their excessive use, at either extreme concentration or duration. 1467) and DT40-KO cells stably over-expressing the rat type 1 InsP 3 R (DT40-R1) were cultured as we described previously 13 . Cells were maintained in suspension culture in RPMI 1640 media containing 10% FCS, 1% chicken serum (CS), penicillin (100 units/ml), streptomycin (100 μg/ml), and glutamine (2 mM) at 37 °C in a 95% air and 5% CO2 humidified incubator. Initial suspension densities of DT40-KO and DT40-R3 were between 1 and 3 × 10 6 cells per ml and cells were passed every 2-3 days. SH-SY5Y were cultured in DMEM/F12 media with 10%FBS, 100 units/ml penicillin, and 100 units/ml streptomycin as previously described 53 . To differentiate neurons, 0.5ml SH-SY5Y cells were seeded from a confluent T75 into a 6-well plate, and 10µM retinoic acid (RA) and B27 were added into the medium, which was changed every two days. Under those conditions, neurons could be obtained after one week 54 .

Reagents
Choice of General Anesthetics. Although isoflurane is not the most commonly used inhalational anesthetic, it is the most studied GA in the area of anesthetic-mediated neurotoxicity. Isoflurane may be more potent in causing cell damage compared to sevoflurane or desflurane 14 . Propofol is the most commonly used and most studied intravenous general anesthetic, with increasing clinical application in outpatient day surgery.
Immunofluorescence. Cells were rinsed briefly in PBS and fixed with 2-4% formaldehyde in PBS for 10 minutes at room temperature, and were permeabilized with 0.1% Triton X-100 in PBS for 5 minutes. Cells were then rinsed with PBS three times for 5 minutes each and blocked in 5% normal (goat) donkey serum in PBS for 60 minutes. The primary antibody was diluted 1:50 in PBS and was incubated overnight at 4 °C. The cells were rinsed in PBS for 5 minutes each with high salt PBS and then rinsed two more times for 2 min. They were then incubated in secondary antibody diluted with PBS for 1 to 2 hours at RT in the dark. Lastly, the coverslips were rinsed with PBS once and stained with Hoechst 33342 1:1000 in PBS for 2-5 minutes. After being washed with PBS 3X5 min, the cells were mounted with Gold antifade reagent, cured on a flat surface in the dark overnight and sealed with nail polish. Cytotoxicity assays. The MTT reduction or LDH release assays determine relatively early or late stages of cell damage, respectively 13 . The LDH assay was performed using a Promaga kit CytoTox 96 Non-Radioactive Cytotoxicity Assay (Product number G1780) per manufacturer's instructions. Cells were plated in a 24 well plate one day prior to the experiment. The first row was left blank and each treatment was done quadraplicately. After each treatment, 50 µl supernatant was collected from each well with equal amount of substrate. The collection was incubated for 30′ at room temperature in the dark and then read with a micro plate reader at 490 nm. For the MTT assay, cells were seeded in a 96 well plate. MTT was prepared using PBS to 5 mg/ml. After treatment, cells were changed into 100 µl fresh medium with 10 µl MTT per well and incubated for 4 h. The reactants were dissolved with 100 µl 10%SDS/0.01M HCl for 4 h or overnight. The results of both LDH release and MTT reduction assays were expressed as a percentage of control. TUNEL assay was performed per instructions of the manufacturer (Roche In Situ Cell Death Detection Kit, Fluorescein, product no 11684795910). Briefly fragmented DNA-3′-OH ends were labeled with fluorescein-dUTP in the presence of TdT enzyme. The coverslips were counterstained with DAPI.
Western blot analysis. The cells were seeded into a 6 well plate with 2ml of cells per well. Following treatment, cells were collected and spun down. They were washed with PBS and lysed in a buffer solution containing 50mM Tris/HCl, pH7.8, 150 mM NaCl, 1% Triton X-100, 100 uM PMSF, a 1X dilution of complete protease and a phosphotase inhibitor mixture (Roche) for 5 min and vortexed. The lysates were cleared by centrifugation at 12,000rpm for 10min. Protein samples were run on 8 or 15% polyacrylamide gels and transferred onto nitrocellulose membranes. Primary antibodies were incubated overnight at 4 °C. Each independent experiment was repeated three times. Protein amount was normalized with actin. The images were quantitated with ImageJ software (http://rsb.info.nih.gov/ij/).

Autophagy flux assay.
To determine the autophagic flux (transition from autophagosome to autolysosome to lysosome), we used an mRFP-GFP-LC3 construct 16 . DT-40 were transfected with the mRFP-GFP-LC3 plasmid (Addgene, Cambridge, MA) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. One day after transfection, cells were exposed to 2.4% isoflurane for 24 hrs. Following isoflurane exposure, cultures were fixed and mounted on cover glasses for microscopy. Colocalized LC3 puncta in both autophagosomes (GFP+mRFP+, yellow color) and autolysosomes (GFP-mRFP+, red color) were quantified across all conditions indicated in texts or figures. Colocalization efficiency was measured with ImageJ software and all data are given as LC3-puncta per cell.

ATP level measurement. ATP level was measured per the instructions of ATPLite Luminescence ATP
Detection Assay System (Perkin-Elmer) using the modified method described previously 55 . 50 μL of mammalian cell lysis solution was added to 100 μL of cell suspension per well of a microplate. The plate was shaken and then 50 μL substrate solution was added to the wells. The luminescence was measured with a BioTech Synergy H1 plate reader.
Mitochondria and lysosome staining. We used the methods described previously 56 . Cells were stained with 1µM MitoTracker Green (In Vitrogen) for 30′ and transferred to a phenol red free medium. TME staining was done by adding TMRE to cells in media to a final concentration of 50-400 nM and incubated for 20 minutes at 37 °C. Fluorescence was monitored and photographed using an Olympus IX70 inverted microscope and the IPLab version 3.7 Imaging Processing and Analysis software (Biovision Technologies, Exton, PA). Lysosomes were stained with 1µM LysoSensor Yellow/Blue DND-160 (Invitrogen) and imaged using a DAPI filter set.

Measurement of Cytosolic Calcium Concentration ([Ca 2+ ] c ).
[Ca 2+ ] c was determined by measuring the F340/F380 ratio using Fura-2 fluorescence (Molecular probe, Eugene, OR) with a photometer coupled to an Olympus IX70 inverted microscope and the IPLab v3.7 Imaging Processing and Analysis software. The protocol to determine the F340/F380 ratio was similar to what we have described previously [12][13][14] with some modifications. On the day of [Ca 2+ ] c measurement, the cells were loaded with 2.5 µM Fura-2AM (Molecular Probes, Eugene, OR) in Krebs-Ringer buffer for 30 min at room temperature. The cells were then placed in a sealed chamber (Warner Instruments, Hamden, CT) connected with multiple inflow infusion tubes and one outflow tube, which provided constant flow to the chamber. The cells were washed with Krebs-Ringer buffer through one inflow tube for the baseline measurement and then exposed to propofol via a separate inflow infusion tube driven by a syringe pump (Braintree Scientific, Braintree, MA). The fluorescence signals of Fura-2/AM were measured with excitation at 340nM and 380nM alternatively and emission at 510 nM for a period up to 18 min for each treatment. The F340/F380 ratio, which correlated to the level of cytosolic Ca 2+ concentration, was constantly determined after exposing cells to propofol. The results of the F340/F380 ratio were averaged from a minimum of 30 cells in at least three separate experiments. The peak cytosolic Ca 2+ and integrated calcium response, represented by area under curve (AUC), were determined and compared.
Statistical Analysis. All data are expressed as the mean ± standard error of the mean (SEM). Statistical analysis was performed using Graphpad Prism 6. P < 0.05 was considered to indicate a statistically significant result. Each experiment was repeated at least three times. The experimental units (n) and statistical analyses used are indicated in the figures and legends.
Data Availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.