Sevoflurane Acts on Ubiquitination–Proteasome Pathway to Reduce Postsynaptic Density 95 Protein Levels in Young Mice

SEVERAL clinical studies reported that anesthesia and surgery were associated with cognitive impairment in children.^1–5  Recent prospective epidemiologic studies show that children do not develop cognitive impairment after single and short duration of anesthesia and surgery.^6,7  These findings are consistent with the outcomes reported by Wilder et al.¹  that children less than 4 years old who had had multiple (e.g., three times), but not single, anesthesia and surgery might develop a learning disability before age 15. Many preclinical investigations illustrated that anesthetics induce cognitive impairment and neurotoxicity in young rodents^8–12  and monkeys.^4,5,13–17 

In the mechanistic investigations, Tao et al.¹²  and Zhang et al.¹⁸  reported that anesthesia with 3% sevoflurane 2 h daily for 3 days, starting on postnatal (P) day 6, induced cognitive impairment and reduction in hippocampus levels of postsynaptic density 95 protein (PSD-95), an excitatory postsynaptic marker,^19–24  in young mice. However, the upstream mechanism by which sevoflurane decreases PSD-95 levels and the downstream consequences of such reduction in PSD-95 levels remain largely unknown.

PSD-95 regulates adhesion and function of receptors, e.g., affecting N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5methyl-4-isoxazolepropionic acid (AMPA) receptor clustering.^25–32  Loss of PSD-95 has been reported to release AMPA receptors from the postsynaptic membrane, allowing for the subsequent removal of PSD-95 from synaptic sites by endocytosis.^29,33,34  Given the role of PSD-95 during synaptic development and cognitive function, the studies to assess the effects of anesthesia on PSD-95 levels and the underlying mechanisms in young mice would illustrate new concepts regarding anesthesia neurotoxicity and neurobehavioral deficits in developing brain. Moreover, we could use anesthesia as a tool to further reveal the underlying mechanism of cognitive impairment in young mice.

PSD-95 can be regulated by the ubiquitination–proteasome pathway.^35–37  Specifically, PSD-95 is ubiquitinated by the E3 ligase mouse double mutant 2 homolog (MDM2), and the ubiquitinated PSD-95 can then be degraded by the ubiquitin–proteasome system.^36,37  Ubiquitination involves the sequential action of three enzymes: ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitin ligase (E3).³⁸  The key enzyme in the process is the E3, because E3 recognizes and interacts with specific protein substrate, in this case PSD-95. E3 catalyzes and then promotes the transfer of activated ubiquitin to the substrate. Finally, MG132 is an inhibitor of proteasome,³⁹  and Nutlin-3 is an inhibitor of MDM2.⁴⁰ 

Taken together, the objective of the current study is to determine the interaction of sevoflurane and the ubiquitination–proteasome pathway on PSD-95 metabolism. We tested a hypothesis that sevoflurane activates the ubiquitination–proteasome pathway to reduce PSD-95 levels, leading to cognitive impairment in young mice. The effects of sevoflurane on protein and mRNA levels of PSD-95, levels of ubiquitinated PSD-95, and interaction of PSD-95 and MDM2 in vitro (both neurons and synaptosomes) and in vivo were assessed. MG132 and Nutlin-3 were used to further determine the extent to which sevoflurane reduced PSD-95 levels and induced cognitive impairment by activating the ubiquitination–proteasome pathway.

The animal studies were conducted according to the guidelines and regulations of the National Institutes of Health (Bethesda, Maryland). Efforts were made to minimize the number of animals in the studies. The Standing Committee on the Use of Animals in Research and Teaching at Massachusetts General Hospital (Boston, Massachusetts) approved the studies (protocol number 2006N000219). Both female and male mice (C57BL/6J; The Jackson Laboratory, USA) were used in the studies. The mice were randomly assigned into the anesthesia group, control group, or intervention group, including control, sevoflurane plus dimethyl sulfoxide (DMSO), sevoflurane plus MG132, and sevoflurane plus Nutlin-3. The mice received sevoflurane from P6 to P8 and then were decapitated for hippocampus harvest at P8. The studies used a different group of mice in the behavioral studies. These mice received the anesthesia or control conditions from P6 to P8 or on P6, P8, and P10 and then had the Morris water maze (MWM) test from P31 to P37. The mice in the anesthesia group received sevoflurane (3%) plus 60% oxygen (balanced with nitrogen) 2 h daily for 3 consecutive days as performed in our previous studies^11,12  or for 3 nonconsecutive days (P6, P8, and P10). The 3% sevoflurane is a clinically relevant concentration, and the anesthesia with 3% sevoflurane 2 h daily for 3 days from P6 to P8 or on P6, P8, and P10 conceptually mimics the multiple exposures of anesthesia in patients. The control conditions were oxygen (60% oxygen and balanced with nitrogen) with an equal rate of flow in a chamber that was similar to the anesthesia chamber.^11,12  P6 or P8 mice are equal to postconception day 228 or 262 in humans.⁴¹  However, the brain growth spurt reaches the peak in humans around birth and at P7 in rodents. ^42,43  We therefore administered the sevoflurane anesthesia at P6, P7, and P8 or P6, P8, and P10 in the current studies. Note that the mice in the control conditions were also separated from the dams and given the control gas (60% oxygen) exposures. The induction flow rate of fresh gas was 2 l/min from the start up to 3 min (for the purpose of induction) and then 1 l/min with the rest of the anesthesia (for maintenance). The concentrations of sevoflurane and oxygen were continuously monitored by using a gas analyzer (Dash 4,000; GE Healthcare, USA) during the anesthesia. The anesthesia chamber temperature was monitored and controlled by a feedback-based system with the DC temperature control system (World Precision Instruments, USA), which controls and automatically adjusts the temperature to keep the rectal temperature of each mouse at 37°C (± 0.5°C) by placing a warming pad under this chamber. The previous studies from another lab¹⁰  and from our own lab⁴⁴  demonstrated that the anesthesia with 3% sevoflurane did not significantly change the levels of pH values, oxygen partial pressure, or carbon dioxide partial pressure as compared to the control group in the young mice. We therefore did not perform the blood gas analysis of the mice in the current studies. In the intervention studies, the mice were treated with MG132 (0.5 mg/kg, dissolved in DMSO at the concentration of 0.3 μg/μl; Sigma-Aldrich, USA)^45,46  or Nutlin-3 (10 mg/kg, dissolved in DMSO at the concentration of 0.3 μg/μl; Sigma-Aldrich)⁴⁷  through intraperitoneal administration 30 min before each sevoflurane anesthesia on P6, P7, and P8. The mice in the control group received 0.1 ml of DMSO solution (15 μl of DMSO dissolved in 1 ml of saline), which is the vehicle of MG132 and Nutlin-3.

The mice were allowed to totally recover from the anesthesia. Each of the mice was euthanized by decapitation at the end of the sevoflurane anesthesia on P8, and the hippocampus tissues were harvested. The collected hippocampus tissues were utilized in the analysis of Western blot. The harvested hippocampus tissues were homogenized on frost by using immunoprecipitation buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40) plus protease inhibitor cocktail from Sigma (catalog No. 11836170001). Finally, the lysates were collected, which were then centrifuged 10 min at the speed of 12,000 rpm.

Hippocampal neurons of mice were harvested as described in our previous studies with modification.^44,48  Specifically, the pregnant dams at gestation day 15 were euthanized with carbon dioxide. The embryos were delivered by Caesarian section, and the brains were dissected in a 100-mm dish of phosphate-buffered saline. Then we specifically placed the hippocampi into another 100-mm dish of phosphate-buffered saline. The neurons were dissociated by using trypsinization and trituration. The dissociated neurons were then resuspended in serum-free B27/neurobasal medium and placed into six-well plates.

The mice were euthanized by decapitation on P6. Hippocampal tissues from the mice were weighed and homogenized in a grinder using Syn-PER synaptic protein extraction reagent, purchased from Thermo Scientific (catalog No. 87793; Rockford, USA). Immediately before use, protease inhibitor cocktail from Sigma (catalog No. 11836170001) was added to the Syn-PER reagent. Mice hippocampus tissues were homogenized using Syn-PER synaptic protein extraction reagent at a volume ratio of 10 using a 7-ml Dounce tissue grinder with 15 up and down strokes. The homogenate was centrifuged at 2,000g for 10 min to remove cell debris. The resulting supernatant was centrifuged at 15,000g for 20 min. The supernatant formed the cytosolic fraction, and the synaptosome pellet was gently resuspended in Syn-PER synaptic protein extraction reagent. The dissociated synaptosomes were then resuspended in medium 199 HEPES modification (M7528; Sigma) medium and placed into six-well plates.

The neurons or synaptosomes were treated with 21% O₂, 5% CO₂, and 4.1% sevoflurane (2 minimum alveolar concentration) for 4 h as described by Dong et al.⁴⁹  We used an anesthesia machine to deliver 21% O₂, 5% CO₂, and 4.1% sevoflurane to a sealed plastic box in a 37°C incubator containing six-well plates seeded with the neurons. A gas analyzer (Dash 4000; GE Healthcare) was used to continuously monitor the concentrations of delivered carbon dioxide, oxygen, and sevoflurane as performed in our previous studies.⁴⁹  In the interaction studies, 50 μM MG132^50,51  or 10 μM Nutlin-3^52,53  was administrated to the neurons 1 h before the sevoflurane treatment. Although different in vitro experiments were performed at different times, the anesthesia and control experiments always occurred concurrently, aiming to reduce bias.

The amount of total proteins was quantified via the kit of bicinchoninic acid protein assay (Pierce, USA) as performed in other studies.⁴⁹ 

RNA was isolated from mouse hippocampus. Real-time one-Step reverse transcriptase polymerase chain reaction (PCR) was carried out using the QuantiTect SYBR Green real-time polymerase chain reaction kit (Qiagen, USA). PSD-95 messenger RNA levels were determined and standardized with glyceraldehyde-3-phosphate dehydrogenase as an internal control. Primers of mouse PSD-95 (ID No. PPM04258A) and mouse glyceraldehyde-3-phosphate dehydrogenase (ID No. QT01658692) were purchased from Qiagen. PCR was performed at 50°C for 30 min, 95°C for 15 min, followed by 50 cycles of 94°C for 15 s, 55°C for 30 s and 72°C for 30 s, and finally 95°C for 15 s and 55°C for 15 s.

The studies utilized PSD-95 antibody (1:1,000, molecular weight of 95 kDa; Cell Signaling, USA) to detect PSD-95 level. An β-actin antibody (1:5,000, molecular weight of 42 kDa; Sigma) was used to detect nontargeted protein β-actin. The quantification of Western-blot was accomplished as described in other studies.⁴⁹  In brief, we analyzed the signal intensity via the Quantity One image analysis program (Bio-Rad, USA). Two steps were used to quantify the Western blots. At the first step, β-actin was used to standardize protein amounts (e.g., calculating the ratio of PSD-95 in relation to β-actin amount) and limit the differences in the protein amount loaded. At the second step, we expressed the protein levels obtained from the treatment as a percentage in relation to control condition.

The coimmunoprecipitation of PSD-95 and ubiquitin or PSD-95 and MDM2 complexes was performed in neurons. The neurons were solubilized by lysis buffer: 0.5% Nonidet P-40, 1 M Tris, pH 8.0, 4 M NaCl, 0.5 M EDTA and stayed at 4°C overnight. After centrifugation at 12,000 × g for 30 min at 4°C, 800 μg of supernatant was incubated with 1 μg of antibody and 30 μl of prewashed protein A or G-Sepharose (50% slurry) overnight at 4°C. The unbound proteins were washed away, and bound proteins were eluted with SDS sample buffer, separated by SDS-PAGE on a 4 to 15% gradient gel, and transferred to nitrocellulose membrane. The other parts were the same as described under “Western Blot Analysis.”

MWM experiments were performed using the methods described in our previous studies.^11,12,18  Briefly, water was filled to a round and steel pool (diameter, 150 cm; height, 60 cm) until the water level reached 1.0 cm over the level of a platform (diametric distance, 10 cm). The pool was located in an isolated room and was surrounded by a black curtain. There were four graphic signals located in the black drape covering the MWM pool. The temperature of the water in the pool was kept at 20°C. The water was made opaque by using titanium dioxide. We tested the P31 mice in the MWM for 7 days (P31 to P37) with four trials daily. A video recording device was used to track the motions of the mice during swimming in the pool. Both escape latency (the time for mouse to reach the platform) in the MWM training from P31 to P37 and platform-crossing times (the counts the mouse moved across the original area of the removed platform) in MWM probe test on P37 were recorded to assess the mouse spatial learning and memory function. Mouse body temperature was maintained by using a heating device described in previous studies.^11,12,18  Specifically, we placed each mouse under a heat lamp for 5 min in a holding cage to dry the mouse before returning the mouse to its home cage.

The data obtained from biochemistry studies and escape latency of MWM are presented as means ± SD. The numbers of the platform-crossing time of MWM were not normally distributed. Hence, the data are presented as medians with interquartile range. The number of samples was 10 to 11 at every group in behavioral studies, 6 in each group in Western blot and PCR, and 3 in each group in coimmunoprecipitation studies. We estimated a difference in mean escape time from the maze of 30 s, with a SD of 20 s. Using two-sided Student’s t test for two sample mean comparison with 5% type 1 error, a sample size of 10 mice in each group will provide us with more than 80% statistical power to detect such a difference. Similarly, a sample size of six mice in each group will provide the same statistical power for Western blot and PCR. We did not quantify the coimmunoprecipitation studies, but the studies were repeated three times. Two-way analysis of variance (ANOVA) with repeated measurement was used to evaluate the difference of escape latency of the mice in the anesthesia group in relation to the mice in the control group in the MWM test. Post hoc analysis was used to compare the change in escape latency of the anesthesia mice as compared to the control mice for every day during the MWM test, cut-off α was adjusted using the Bonferroni method. We used Mann–Whitney U test to compare the difference in the platform-crossing times of the mice in the anesthesia group and the mice in the control group. Two-way ANOVA without repeated measurement was used to evaluate the interaction of group (control group vs. anesthesia group) and treatment (MG132 or Nutlin-3 vs. DMSO) on the levels of PSD-95. The data in the two-way ANOVA without repeated measurement were found to have normal distribution. A Kruskal–Wallis test was used to detect the platform-crossing times among three groups (MG132 or Nutlin-3 vs. DMSO). The studied employed two-tailed hypothesis, and statistically significant P values were less than 0.05. We used the software of Prism 6 (GraphPad, USA) to evaluate all of the study data.

Increasing evidence suggests that PSD-95 plays an important role during synaptic development and cognitive function.^54–56  We therefore first assessed the effects of sevoflurane on PSD-95 metabolism by determining both mRNA and protein levels in hippocampi of mice after anesthesia with 3% sevoflurane 2 h daily for 3 days on P6, P7, and P8. Immunoblotting of PSD-95 showed that the sevoflurane anesthesia reduced the visibility of the band representing PSD-95 as compared to the control conditions (fig. 1A). There were no significant differences in β-actin levels between the sevoflurane anesthesia and control condition. Quantification of the Western blot, based on the ratio of PSD-95 to β-actin level, revealed that the sevoflurane anesthesia decreased the protein levels of PSD-95: 48% versus 100%, P = 0.002 (fig. 1B). However, the sevoflurane anesthesia did not significantly change the mRNA levels of PSD-95 (fig. 1C, P = 0.950) in the hippocampi of the mice. These results suggest that sevoflurane may decrease the PSD-95 levels by enhancing its posttranslational degradation.

Given the findings that sevoflurane specifically affected the degradation of PSD-95, next we asked whether MG132, a proteasome inhibitor of the ubiquitination–proteasome pathway, would block the sevoflurane-induced reduction of PSD-95 levels in neurons, synaptosomes, and hippocampi of mice. In the neurons pretreated with DMSO, sevoflurane (fig. 2A, lane 2) decreased PSD-95 levels as compared to control conditions (fig. 2A, lane 1). However, in the neurons pretreated with MG132, sevoflurane (fig. 2A, lane 4) did not decrease PSD-95 levels as compared to control conditions (fig. 2A, lane 3). There were no significant differences in β-actin levels among these treatments. Quantification of the Western blot showed that sevoflurane (black bar in fig. 2B) reduced PSD-95 level as compared to the control conditions (white bar in fig. 2B), and MG132 blocked the sevoflurane-induced changes in PSD-95 levels (striped bar vs. gray bar in fig. 2B; F = 11.030, P = 0.003, two-way ANOVA).

Next, synaptosomes were employed in the studies. The successful extraction of synaptosome was demonstrated by higher levels of PSD-95 and synaptophysin in the synatosomes than those in homogenates and cytosol (Supplemental Digital Content, https://links.lww.com/ALN/B541). The quantitative Western blot showed that MG132 blocked the sevoflurane-induced changes in PSD-95 levels in neurons (fig. 2, A and B; F = 11.030, P = 0.003, two-way ANOVA), synaptosomes (fig. 2, C and D; F = 4.728, P = 0.042, two-way ANOVA), and hippocampi of mice (fig. 2, E and F; F = 11.030, P = 0.003, two-way-ANOVA). MG132 is a general inhibitor of proteasome; thus, these data suggest that sevoflurane may decrease PSD-95 levels through ubiquitination–proteasome pathway.

Given the findings that sevoflurane may decrease PSD-95 levels through ubiquitination–proteasome pathway and Nutlin-3 is an inhibitor of MDM2, the E3 ligase of PSD-95 ubiquitination, next we determined the effects of Nutlin-3 on the sevoflurane-induced reduction in PSD-95 levels in neurons, synaptosomes, and hippocampi of mice. The quantitative Western blot showed that Nutlin-3 blocked the sevoflurane-induced changes in neurons (fig. 3, A and B; F = 5.202, P = 0.034, two-way ANOVA), synaptosomes (fig. 3, C and D; F = 12.680, P = 0.002, two-way ANOVA), and hippocampi of the mice (fig. 3, E and F; F = 9.045, P = 0.007, two-way-ANOVA). Taken together, these data showed that both MG132 and Nutlin-3 blocked the sevoflurane-induced changes in PSD-95 levels and suggest that sevoflurane may decrease PSD-95 levels through promoting the ubiquitination of PSD-95.

Ubiquitin, a 76-amino acid protein, is covalently attached to substrate proteins. Once bound with ubiquitin, substrate proteins are often targeted for rapid degradation by the 26S proteasome.^38,57  We therefore assessed whether sevoflurane might affect the ubiquitination of PSD-95. Coimmunoprecipitation of ubiquitin and PSD-95 showed that sevoflurane (lane 2) reduced ubiquitinated PSD-95 levels as compared to control conditions (lane 1) in the neurons pretreated with DMSO (top panel in fig. 4). However, in the neurons pretreated with MG132, sevoflurane (lane 4) did not reduce the ubiquitinated PSD-95 levels as compared to control conditions (lane 3 in top panel in fig. 4). The studies were repeated three times. Consistent with this finding, MG132 blocked the sevoflurane-induced reduction in PSD-95 levels in the total cell extracts (bottom panel in fig. 4). These data further suggest that sevoflurane may activate the ubiquitination–proteasome pathway, leading to degradation of PSD-95.

E3 ligase MDM2 interacts with the ubiquitination of protein, e.g., PSD-95. Thus, we determined the interaction of PSD-95 and MDM2 in neurons. Coimmunoprecipitation of MDM2 and PSD-95 showed that sevoflurane increased the interaction of PSD-95 and MDM2 (top panel in fig. 5) in the neurons. The same sevoflurane treatment also decreased PSD-95 levels in total cell extract (bottom panel in fig. 5). Taken together, these data suggest that sevoflurane may reduce PSD-95 levels not only by activating proteasome but also by promoting the interaction of MDM2 and PSD-95 in the ubiquitination–proteasome pathway.

Finally, the interaction of sevoflurane with MG132 or Nutlin-3 on cognitive function in young mice was determined. In the mice pretreated with DMSO (the vehicle of MG132 and Nutlin-3), the anesthesia with 3% sevoflurane 2 h daily for 3 days induced cognitive impairment tested in MWM from P31 to P37, as evidenced by the significant interaction between treatment (control conditions and sevoflurane anesthesia) and time (P31 to P37) on escape latency of MWM training test (F = 2.202, P = 0.048, two-way ANOVA with repeated measurement; fig. 6A). A post hoc test (Bonferroni) showed that the mice exposed to sevoflurane had longer escape latency to locate the position of platform in MWM pool as compared to the control mice at P35 (P < 0.05), P36 (P < 0.01), and P37 (P < 0.05; fig. 6A). The mice that received the sevoflurane also demonstrated lower platform-crossing times as compared to the mice that received control conditions (fig. 6B; P = 0.036, Mann–Whitney test).

However, in the mice with the pretreatment of MG132, the sevoflurane anesthesia did not induce cognitive impairment tested in the MWM as evidenced in the fact that there was no significant interaction between treatment (control conditions and sevoflurane anesthesia) versus time (P31 to P37; F = 0.795, P = 0.576, two-way ANOVA with repeated measurement; fig. 6C). Consistent with this finding, the sevoflurane anesthesia did not significantly decrease the platform-crossing times as compared to control conditions in the mice pretreated with MG132 (fig. 6D; P = 0.912, Mann–Whitney test).

Similarly, in the mice with the pretreatment of Nutlin-3, the sevoflurane anesthesia did not induce cognitive impairment tested in MWM as evidenced by no significant interaction (F = 1.238, P = 0.292, two-way ANOVA with repeated measurement; fig. 6E) between treatment (control conditions and sevoflurane anesthesia) and time (P31 to P37) in the training of MWM test and by no difference in platform-crossing times of the probe test (P = 0.350, Mann–Whitney test; fig. 6F).

In addition, there was no significant interaction of treatment (DMSO or MG132 or Nutlin-3) and time (P31 to P37) on escape latency of MWM training test (F = 0.570, P = 0.864, two-way ANOVA; fig. 7A). There were no significant differences of platform-crossing times among these three groups. (fig. 7B; P = 0.247, Kruskal–Wallis test). Taken together, these data indicated that there were no significant differences of the effects of DMSO, MG132, and Nutlin-3 alone on cognitive function in young mice (fig. 7). Collectively, these findings suggest that both MG132 and Nutlin-3 can ameliorate the sevoflurane-induced cognitive impairment in young mice.

Finally, we asked whether the sevoflurane-induced cognitive impairment was dependent on the 3-day consecutive anesthesia exposures. The effects of sevoflurane anesthesia on cognitive function in young mice were determined using different anesthesia methods. Specifically, the young mice received 3% sevoflurane 2 h daily for 3 nonconsecutive days (P6, P8, and P10) instead of 3 consecutive days (P6, P7, and P8). We found that the nonconsecutive 3-day sevoflurane anesthesia also induced cognitive impairment in the young mice, as evidenced by a significant interaction of treatment (control and sevoflurane) and time (P31 to P37) on escape latency of MWM (fig. 8A) and significant difference of platform-crossing times between control and sevoflurane (fig. 8B). These data suggest the sevoflurane-induced cognitive impairment in young mice is not dependent on the 3 consecutive day exposures of anesthesia.

In this mechanistic investigation of anesthesia-induced neurotoxicity and neurobehavioral deficits in developing brain, we demonstrated that sevoflurane activated the ubiquitination–proteasome pathway by stimulating ubiquitination–proteasome pathway and promoting the interaction of PSD-95 and MDM2. This interaction facilitates the ubiquitination of PSD-95, leading to a reduction in protein levels of PSD-95. MG132 (a proteasome inhibitor of the ubiquitination–proteasome pathway) and Nutlin-3 (an inhibitor of MDM2) blocked the sevoflurane-induced PSD-95 metabolism and ameliorated the sevoflurane-induced cognitive impairment in young mice.

The sevoflurane anesthesia decreased the protein level, but not the mRNA level, of PSD-95 (fig. 1). These results suggest that sevoflurane may not regulate the generation of PSD-95 but facilitate PSD-95 degradation. Then the sevoflurane-induced changes in PSD-95 levels were inhibited by MG132 (fig. 2) and Nutlin-3 (fig. 3). These data suggest that sevoflurane decreases PSD-95 levels by activating the ubiquitination–proteasome pathway and specifically through acting on MDM2 and proteasome. The future studies should assess the underlying mechanism by which sevoflurane acts on MDM2 by investigating whether sevoflurane regulates the trafficking of MDM2. Note that the interactions of sevoflurane and MG132 or Nutlin-3 were examined in both neurons and synaptosomes, which would help us to determine age-dependent changes of sevoflurane on the interaction of PSD-95 and MDM2 or proteasome by using synaptosomes obtained from mice with different ages.

There was an interaction of the sevoflurane and MG132 on the levels of ubiquitinated PSD-95 (fig. 4). The original hypothesis was that sevoflurane would increase the ubiquitinated PSD-95 levels by promoting the ubiquitination of PSD-95. Interestingly, we found that sevoflurane decreased the ubiquitinated PSD-95 levels in the hippocampus neurons. These data suggest that sevoflurane may have stronger effects on stimulating proteasome-associated degradation of ubiquitinated PSD-95 than on promoting MDM2-regulated generation of ubiquitinated PSD-95 in a time-dependent manner. This hypothesis was supported by the observation that proteasome inhibitor MG132 blocked the sevoflurane-induced changes in ubiquitinated PSD-95 levels (fig. 4). Moreover, these findings are consistent with the results obtained from the previous studies that treatment with NMDA induces time-dependent changes in ubiquitinated PSD-95 levels.³⁶  Specifically, there was a reduction and then an increase in ubiquitinated PSD-95 levels at 5 and 10 min after the treatment of NMDA, respectively, in hippocampus neurons.³⁶  Future studies should test whether sevoflurane increases ubiquitinated PSD-95 levels at shorter time and decreases its levels at longer time after the treatment.

Moreover, sevoflurane promoted the interaction of PSD-95 and MDM2 (top panel in fig. 5) in neurons and decreased PSD-95 levels in total cell extract (bottom panel in fig. 5). These data suggest that sevoflurane can promote the binding and interaction of MDM2 with PSD-95, leading to ubiquitination of PSD-95. These findings further suggest that sevoflurane may act on both MDM2 and other parts in the ubiquitination–proteasome pathway.

Finally, both MG132 and Nutlin-3 ameliorated the sevoflurane-induced cognitive impairment in young mice (fig. 6), as well as the sevoflurane-induced changes in PSD-95 levels. Taken together, these findings suggest that the sevoflurane-induced reduction in PSD-95 may lead to cognitive impairment in young mice.

PSD-95 is a major scaffolding protein in excitatory postsynaptic density²⁵  that enhances maturation of the presynaptic terminal, increases the number and size of dendritic spines,⁵⁸  and contributes to synaptic organization.⁵⁹  Experiments demonstrate that PSD-95 can orchestrate synaptic development and play an important role in synapse stabilization and plasticity.⁶⁰  Our previous studies reported that sevoflurane reduced the levels of PSD-95 in vitro and in vivo.^12,18  The findings from current studies reveal that sevoflurane may decrease PSD-95 levels by activating the ubiquitination–proteasome pathway.

Earlier studies have demonstrated that PSD-95 can be degraded by the ubiquitination–proteasome pathway.^36,37  Specifically, Colledge et al.³⁶  showed that treatment with NMDA induced a rapid ubiquitination of PSD-95, which reached peak levels at 10 min after the treatment, leading to reduction in PSD-95 levels in the neurons. Tsai et al.³⁷  reported that activation of the activity-dependent transcription factor myocyte enhancer factor 2 in the nucleus promoted the ubiquitination of PSD-95, which consequently decreased the levels of PSD-95 and synapse in hippocampus neurons.

Consistently, sevoflurane decreased PSD-95 levels by promoting PSD-95 ubiquitination in neurons, synaptosomes, and brain tissues of mice. However, the current studies are different from the previous studies in revealing: (1) the ubiquitination–proteasome pathway is one of the underlying mechanisms of the cognitive impairment in young mice; (2) sevoflurane induces cognitive impairment in young mice through stimulation of the ubiquitination–proteasome pathway and promoting interaction of MDM2 and PSD-95; (3) the sevoflurane-induced reduction in PSD-95 levels contributes, at least partially, to the sevoflurane-induced cognitive impairment; and (4) MG132 and Nutlin-3 may prevent and treat the sevoflurane-induced neurotoxicity and neurobehavioral deficits in young mice. Collectively, these findings could help us to further understand cognitive impairment and anesthesia neurotoxicity in the developing brain.

It is possible that different mechanisms may contribute to the in vitro and in vivo findings. However, in the present studies, the in vitro and in vivo findings likely share, at least partially, the same mechanism: the ubiquitination-proteasome degradation pathway. First, the effects of MG132 and Nutlin-3 on attenuating the sevoflurane-induced reduction in PSD-95 levels were similar among neurons and synaptosomes (in vitro studies) and brain tissues (in vivo studies). Second, the control conditions in the in vitro studies, which did not affect PSD-95 levels, had the same direct physical interference of airflow rate as compared to the anesthesia condition.

The study has several limitations. First, we only assessed the effects of sevoflurane on the levels of PSD-95, but not the number of synapse or dendritic spine, in hippocampus of young mice. This is because previous studies have reported that reductions in PSD-95 levels are associated with reductions in synapse numbers.³⁷  Future studies should determine whether sevoflurane can reduce the number of synapses and dendritic spines in the hippocampi of mice. Second, we only performed the interaction of PSD-95 and MDM2 studies in neurons, but not in synaptosomes or brain tissues, due to technical limitations of immunoprecipitation experiments in synaptosomes. Third, the anesthesia used in the current studies was 3% sevoflurane 2 h daily for 3 days on P6, P7, and P8 in the young mice. This is because our previous studies showed that anesthesia with 3% sevoflurane 2 h daily for 3 days (P6, P7, and P8), but not for 1 day, in young mice was able to induce cognitive impairment at P31 to P37 in the mice.^11,12  This anesthesia may not be exactly relevant to pediatric anesthesia practice. In addition, the anesthetic exposure on 3 consecutive days could represent a severe form of toxicity. However, this sevoflurane anesthesia did not increase mortality or induce abnormal appearance in the mice, e.g., ruffled fur guarding, loss of appetite, reluctance to move, mutilation, restlessness, recumbence, and vocalization, as compared to the control group at P31. Moreover, anesthesia with 9% desflurane 2 h daily for 3 days (P6, P7, and P8) in the young mice did not induce cognitive impairment tested from P31 to P37 in the mice.¹¹  These data suggest that the 3 consecutive days of anesthesia may not cause a severe form of toxicity. Finally, the findings that anesthesia with 3% sevoflurane 2 h daily for nonconsecutive days (P6, P8, and P10) in young mice still induced cognitive impairment (fig. 8) suggest that the sevoflurane-induced cognitive impairment in young mice may not be dependent on the 3 consecutive day anesthesia exposures. Fourth, we did not specifically compare the effects of the anesthesia in male and female mice in the present studies. This is because our previous studies demonstrated that sevoflurane was able to induce cognitive impairment in both male and female young mice, and the objective of the present study was to use the established system to determine the upstream ubiquitination–proteasome mechanism by which the sevoflurane anesthesia decreases the level of PSD-95 and the downstream cognitive consequence of such reduction in PSD-95 levels. Finally, there were dam’s rearing behaviors in the young mice because the mice were separated from mothers.⁶¹  However, the control mice also had the same time of separation from the mother. Moreover, the separation only occurred 2 h daily for 3 days.

In conclusion, we found that sevoflurane reduced levels of PSD-95 in vitro (neurons and synaptosomes) and in vivo (brain tissues of young mice) by activating the ubiquitination–proteasome pathway. Specifically, sevoflurane stimulated the proteasome and promoted interaction of MDM2 and PSD-95, which caused ubiquitination of PSD-95. The ubiquitinated PSD-95 then was degraded by proteasome, leading to reduction in PSD-95 levels. MG132, a proteasome inhibitor of the ubiquitination–proteasome pathway and Nutlin-3, an inhibitor of MDM2, mitigated these effects. These results illustrated the potential underlying mechanisms of neurotoxicity and neurobehavioral deficits induced by anesthesia or other external stressors in developing brain. These findings should promote further investigation of cognitive impairment in young mice, which will ultimately improve the safety of anesthesia care and lead to better outcomes of anesthesia and surgery in children.

Sevoflurane was generously provided by the Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital and Harvard Medical School (Boston, Massachusetts).

Supported by National Institutes of Health (Bethesda, Maryland) grant Nos. HD086977, GM088801, and AG041274 (to Dr. Xie) and in part by the Chinese National Natural Science Foundation (Beijing, China) grant Nos. 81102513 (to Dr. Lu) and 81373492 (to Dr. Yu).

The authors declare no competing interests.