Egr2 upregulation induced mitochondrial iron overload implicating in sevo urane-induced cognitive de cits in developing mice


 Background Sevoflurane inhalation initiated cognitive deficits implicated in mitochondrial dysfunction and synaptogenesis impairment. Bioinformatics analysis indicated that Egr2 may play a crucial role in maintaining cognitive function. Therefore, we attempted to clarify the potential mechanism regarding Egr2 expression and cognitive deficits induced by sevoflurane administration.Methods Animals received sevoflurane anesthesia, and the behavioral tests including Morris water maze, novel object recognition test and trace fear conditioning were performed. Then, the immunofluorescent staining was employed to detect the effect of sevoflurane inhalation in hippocampal neurons. Meanwhile, bioinformatics analysis was implemented, and the level of lipid peroxidation, mitochondrial membrane potential, morphology and membrane permeability, and cytoplasm calcium levels were investigated after Egr2 interference by using JC-1 probe, MitoTracker staining, Mitochondrial permeability transition pore (mPTP) assay, and Fluo calcium indicators, respectively. Additionally, Prussian blue staining was used to evaluate the iron content.Results The behavioral tests indicated that the cognitive function was significantly attenuated after sevoflurane administration. The Golgi-Cox staining displayed that the dendritic length, density and nodes were significantly reduced following sevoflurane inhalation. The bioinformatics analysis showed that sevoflurane administration results in the Egr2 expression upregulation. Additionally, the results suggested that sevoflurane administration elevated the cytoplasm calcium levels, reduced the mitochondrial membrane potential and triggered the opening of mPTP. Prussian blue staining showed that the iron deposition was apparently increased. However, Egr2 level downregulation partly reversed these above changes. Moreover, the behavioral performance was effectively improved after deferiprone (DFP) administration.Conclusion These findings demonstrated that sevoflurane administration elicited mitochondrial dysfunction and iron dyshomeostasis, and eventually resulted in cognitive impairments, whereas suppressing Egr2 expression partly improved this pathological process.

All animals were randomly assigned to different groups according to the experimental protocols: Part1) control group (Ctrl) and sevo urane group (SEV); Part2) GFP-Ctrl group, GFP-SEV group, GFP-Egr2 shRNA-Ctrl group, GFP-Egr2 shRNA-SEV group; Part3) Ctrl group, SEV group, DFP group and SEV+DFP group. To induce general anesthesia, the pups were placed in an acrylic anesthetizing chamber with two interfaces including sevo urane vaporizer and multi-gas monitor. The SEV group was exposed to 3% sevo urane delivered in humidi ed 60% O 2 carrier gas for 2 h (2 L/min fresh gas for 3 min, followed by 1 L/min) by using the Datex-Ohmeda anesthesia system (Madison, WI, USA), while the Ctrl group received 60% oxygen (balanced with nitrogen) for the same period at postnatal day 6-8 (P6-8) [9]. Similarly, GFP-Ctrl group and GFP-Egr2 shRNA-Ctrl group received the same process with Ctrl group, and the GFP-SEV and GFP-Egr2 shRNA-SEV group underwent the scheme with SEV group. Moreover, to ensure su cient ventilation, a single sample (100 μL) of arterial blood was obtained at the end of sevo urane anesthesia or sham exposure by cardiac puncture from ve mice of each group. These animals were not used for any other part of the study. Arterial carbon dioxide partial pressure (PaCO 2 ), arterial oxygen pressure (PaO 2 ), blood oxygen saturation (SaO 2 ) and power of hydrogen (pH) were evaluated by using a blood gas analyzer (Kent Scienti c Corp., Torrington, CT, USA) ( Table 1). There was no significant difference in pH, PaCO 2 , PaO2, Glucose and SaO 2 level between the groups.
To inhibit the level of iron, Deferiprone (DFP), an iron chelating agent [21], was administrated to detect the impact of iron chelating agent in mice after sevo urane treatment. The mice in DFP group and SEV+DFP group were given Deferiprone (DFP) (100 mg/kg in 1% DMSO, i.p.) at P9 and the mice in Ctrl group and SEV group were given 50 μL of 1% DMSO by intraperitoneal injection at P9, respectively.

Behavioral test
Morris water maze After the sevo urane exposure, the spatial memory abilities were evaluated at P40 by using the Morris Water Maze (MWM) test as previously described [24]. A circular black pool (diameter: 120 cm; depth: 21 cm) was lled with opaque water using black non-toxic ink to reach 1.0 cm above the platform surface (diameter, 10 cm), and the water temperature was kept at 22 °C. Meanwhile, an invisible platform (diameter, 10 cm) was xed in the pool and submerged 1 cm. In the training phase (P40-44), all animals received four training trials per day for a total of four days. The mice were placed into the pool at a random starting position and allowed to discover the hidden platform for 120 s. Mice were guided to the platform if they could not locate the platform within 2 min. The latency time (the time to reach the hidden platform) was recorded for assessing the spatial learning. In the testing phase (P45), the platform was removed, and the mean distance crossed the original platform site, platform-crossing times, and time spent were recorded for measuring memory function, respectively. After each trial, the mice were wiped dry and a heat lamp was used to faster temperature recovering before returning to home cages.

Novel object recognition test
Cognition was measured by the Novel object recognition (NOR) experiment at P35.
The animals are exposed to two identical objects for 20 min, then trained for 5 min during the familiarization phase. Thereafter, the mice are exposed to a single copy of the familiar object and a novel object (test phase) after 24h. The total distance traveled was recorded and the Recognition index was calculated [25]: A recognition index was calculated for each animal and re-ported as the ratio TB/(TA + TB), where TA = time spent exploring the familiar object A and TB = time spent exploring the novel object B. Recognition memory was evaluated as in the long-term memory test. Exploration was de ned as sni ng or touching the object with the nose or forepaws.

Trace fear conditioning
The fear condition test is extensively used to detect the tone's effect on the hippocampus-dependent memory [26]. Brie y, the mice were placed in a sound attenuating fear-conditioning chamber (ACT-100A, Coulbourn Instrumnets, USA), The mice free explored for 2 min in the chamber, and the freezing was recorded as control. Then, the mice received 30s sound (80dB, 1500HZ) as conditioned stimulus, and foot shock (0.7 mA; 2 s) by the oor's steel rods at last 2 s, and keep the sound and the shock stopped at the same time. The mice stayed in the chamber for another 2 min. The training repeated for 5 times. The next day for contextual fear test, the mice were placed into the same chamber and the freezing was recorded for 5 min. After 2h, the mice were placed into another chamber for 3 min, then received the same conditioned stimulus for 3 min. The freezing of mice was recorded all the time. The data of freezing were recorded by Freeze Frame software.

Tissue harvest
Animals were anesthetized with 2% pentobarbital sodium (40 mg/kg, i.p.) at P42. Then, the right atrium was incised and transcardiac perfusion was performed with heparinized 0.9% saline followed by 4% formaldehyde. The brain tissue was extracted and rinsed using 0.9% sodium chloride at 4 °C. The hippocampus was stripped and xed in 30% sucrose in 0.1 M phosphate buffers (pH 7.4, 4 ℃) for 24-48h, then the specimens were stored in a −80 °C freezer.

TUNEL Assay
A TUNEL assay was performed to detect the DNA fragmentation caused by cell death in the hippocampus of aged rats. After preparation of sections (6 μm), the TUNEL staining was carried out using an in situ cell death detection kit (YEASEN, 40302ES20) according to the manufacturer instructions. Fluorescence signals were visualized under an epi uorescence microscope. Images were captured with the assistance of Image-Pro Plus 5.0 software, and all the parameters used in this experiment were kept consistent during capturing.
Then sections were rinsed with PBS (3×10 min) followed by incubation with Alexa Fluor™ 488 goat antimouse antibody and Alexa Fluor™ 594 goat anti-rabbit antibody for 1 h at room temperature. After rinsing with PBS (6×5 min), uorescence signals were visualized under an epi uorescence microscope. Images were captured with the assistance of Image-Pro Plus 5.0 software, and all the parameters used were kept consistent during capturing.

Golgi-Cox staining
The morphology of neuronal dendrites and dendritic spines was investigated in the hippocampus by using the Hito Golgi-Cox OptimStain TM PreKit (Hitobiotec Corp. Kingsport, TN, USA). The brain tissues were obtained after sacri ce, and rinsed with Milli Q water. The equal volumes of Solutions A and B were used to impregnate the brain tissues, and the impregnation solution was replaced the following day and stored in darkness (Room temperature, 2 weeks). Then, the brain tissues were transferred to Solution C, which was replaced the following day. The brains were stored at 4 °C for 72 h in the dark. The Brain sections (100 μm thickness) were generated using a cryotome with the chamber temperature set at −19 °C . Each section was mounted on gelatin-coated microscope slides using Solution C. Each section was mounted on gelatin-coated microscope slides using Solution C. The excess solution on slide was removed using a Pasteur pipette and absorbed with lter papers, then the sections were allowed to dry naturally at room temperature for 3 days. The dried brain sections were processed according to the manufacturer's instructions. Thereafter, the dendrites of CA1 sub region in the hippocampus were observed by using an Olympus BX61 uorescence microscope (Olympus, Japan).

RNA extraction
For the RNA-Seq analysis, the hippocampal tissues were obtained from Ctrl group and SEV group at P6 and P30, respectively. Total RNA was extracted from different group using RNAiso Plus Reagent (TaKaRa, Japan), and puri ed by RNasey Mini Kit (QIAGEN) based on the manufacturer's protocol. NanoDrop spectrophotometry (Thermo Scienti c, Wilmington, USA) was used to detect the RNA concentration, and the integrity was con rmed through electrophoresis. Subsequently, the cDNA synthesis and antisense RNA (aRNA) ampli cation was performed using Amino Allyl MessageAmp II aRNA Ampli cation Kit (Ambion, USA). The total RNA was stored at -80°C for future use.

RNA-Seq
A total of 1.5 μg RNA was used as the input material. The clustering of the index-coded samples was performed by using a TruSeq PE Cluster Kit v3-cBot-HS (Illumina) based on the manufacturer's instructions. The library were sequenced using an Illumina HiSeq platform, and paired-end reads were generated followed by cluster generation. Thereafter, these raw reads in the fastq format were processed by using in-house Perl scripts. Low-quality data were discarded by using Trim Galore (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The GC-content and sequence duplication level of the clean reads were calculated, and the clean reads were assembled with Trinity software via the default parameters (https://github.com/trinityrnaseq/trinityrnaseq/wiki). Then, the RNAseq data les were deposited in the NCBI Sequence Read Archive (SRA) database (SRA accession: Not uploaded, to date).

Data analysis by using integrated Differential Expression and Pathway analysis (iDEP) tools
The differentially expressed genes (DEGs) acquired from the RNA-seq-Based expression pro ling were analyzed through iDEP (integrated Differential Expression and Pathway analysis) online tools (http://bioinformatics.sdstate.edu/idep/). To date, iDEP seamlessly connects 63 R/Bioconductor packages, 2 web services, and comprehensive annotation and pathway databases for 220 plant and animal species [27]. Brie y, the expression matrix of DEGs (Table S1 and Table S2) was ltered and converted to Ensemble gene IDs, and the exploratory data analysis (EDA) including K-means clustering and hierarchical clustering was performed using the pre-processed data. The pairwise comparison (Ctrl-6d group VS SEV-6d group; Ctrl-30d group VS SEV-30d group) was employed by using the DESeq2 package with a threshold of false discovery rate FDR < 0.05 and fold-change > 2. Additionally, a hierarchical clustering tree and network of enriched GO terms were constructed to visualize the potential connections among DEPs. Gene Set Enrichment Analysis (GSEA) method was used to investigate the related signal pathways activated by sevo urane administration. Therefore, WGCNA was performed to construct co-expression networks and sub-modules, and the corresponding enriched pathways in selected module were exhibited, respectively.

Gene Ontology and KEGG Pathway Analysis of DEGs
Gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were employed to analyze the differentially expressed genes (DGEs) between different groups (Ctrl-6d group VS SEV-6d group; Ctrl-30d group VS SEV-30d group) by using String online tools (https://stringdb.org/cgi/input.pl). GO analysis was used to annotate genes and gene products including biological process (BP), cellular component (CC) and molecular function (MF). KEGG is utilized for systematic analysis of gene function and related high-level genome functional information of DGEs, which consists of a series of genome and enzymatic approaches and genomic information with higher order functional information [28].
Integration of Protein-Protein Interaction (PPI) Network Analysis and related database STRING version 10.0 covers 9, 643, 763 proteins obtained from 2031 organisms [29]. The String database (https://string-db.org/cgi/input.pl) is used to predict the protein-protein interactions comprising direct/indirect associations. To investigate the potential relationships, String tool was constructed according to the function and pathway enrichment analysis. Moreover, the Genecards website (https://www.genecards.org) and The Human Protein Atlas database (https://www.proteinatlas.org/) were separately used to determine the expression of Egr2 in the tissues and organs of the human body.

Cell culture
Primary hippocampal neurons were cultured by using fetal mice (E17) hippocampi according to a previously described protocol [30]. Brie y, the mice pregnancy for 17 days was anesthetized through 1% iso urane, and the uterus was exposed and the fetus was removed. The fetal mice were sacri ced and the hippocampi were obtained under a sterile environment. Then, the hippocampi were treated using 0.125% trypsin in Hank's buffer (in mmol/L: 137 NaCl, 5.4 KCl, 0.4 KH 2 PO 4 , 0.34 Na 2 PO 4 ·7H 2 O, 10 glucose and 10 HEPES) for 12 min at 37 °C and dissociated by repeated passage with Pasteur pipettes.

Sevo urane treatment
The cultured cells were placed in an airtight plastic chamber (MIC-101), which was connected to an acrylic anesthetizing chamber with two interfaces including a sevo urane vaporizer and a multi-gas monitor. The chamber was gassed with 4.1% sevo urane in the carrier gas (95% air/5% CO 2 ) for 15 min, and the concentration of sevo urane was monitored by a gas monitor (PM 8060, Drager, Lübeck, Germany) [31]. Then, the chamber was sealed and incubated for 6 h at 37 °C. The gas was renewed every 3 h, and the concentration of sevo urane was con rmed at the end of the incubation. Meanwhile, the control group received the same procedure with air containing 5% CO 2 .

Iron Levels detection
Iron assay was performed according to the manufacturers' protocol of Iron Assay Kit (Abcam, ab83366) [32]. Brie y, the specimens were incubated with iron reducer at 25 °C for 30 min followed by incubating for 60 min with iron probe at 25 °C. Then, the microplate reader (OD 593 nm) was used to detect the level of iron.

Lipid peroxidation assay
To detect the level of lipid peroxidation, the uorescent reporter molecule C11-BODIPY 581/591 (Invitrogen TM , D3861) was used. Cells were induced with the probes for 30 min (2.5 µM), and the uorescence of C11-BODIPY 581/591 shifted from red to green. The uorescence spectrophotometer was utilized to monitor this kinetics of the reaction, and the uorescence emission intensity at 520 nm was recorded. Images were captured with the assistance of Image-Pro Plus 5.0 software, and all the parameters used were kept consistent during capturing.

Detection of mitochondrial membrane potential and calcium level
Mitochondrial membrane potential was detected by using JC-1 (Thermo Fisher Scienti c, MA, USA) uorescent dye. H4 cells were randomly divided into four groups including Ctrl group, SEV group, Egr2 shRNA group and SEV+ Egr2 shRNA group, and these cells were cultured for 6 h. Then, 10 μM JC-1 reagent was added and stained for 20 min. JC-1 emits uorescence including red uorescent Jaggregates (530 nm excitation/590 nm emission) at high potentials, and green uorescent J-monomers (490 nm excitation/530 nm emission) at low potentials. The cells were visualized immediately after treatment using an epi uorescence microscope. Images were captured with the assistance of Image-Pro Plus 5.0 software. Additionally, the cytoplasmic calcium level was measured using Fluo calcium indicators (Fluo-4, AM, YEASEN, 40704ES50). An aliquot of DMSO stock solution (5 mM) was diluted to a nal concentration of 5 µM in buffered physiological medium. H4 cells were washed with indicator-free medium after treating for 6 h. The uo acetoxymethyl ester was used for cell incubation (30-60 min, 37°C). Cells were washed in indicator-free medium once again prior to uorescence was measured. The uorescence signals were visualized by an epi uorescence microscope after treatment. Images were obtained with the assistance of Image-Pro Plus 5.0 software.

MitoTracker Imaging
H4 cells were randomly divided into four groups consisting of Ctrl group, SEV group, Egr2 shRNA group and SEV+Egr2 shRNA group, and these cells received corresponding treatment for 6 h. The medium was replaced with pre-warmed (37°C) 50 nM MitoTracker (Invitrogen) medium for 10 min. Thereafter, the loading solution was replaced with fresh medium once again. The cells were visualized after treatment by using an epi uorescence microscope. Images were captured with the assistance of Image-Pro Plus 5.0 software.
Mitochondrial permeability transition pore assay Determination of Malondialdehyde (MDA) and GSH Levels Tissue proteins were prepared as described in the Lipid Peroxidation MDA assay kit (Beyotime, S0131). The MDA concentration of each sample was evaluated by multimode microplate readers (SpectramMax M5) at 532 nm, and using 490 nm served as a control. Additionally, the level of GSH were measured according to the requirements of the instructions in reagent kits (Beyotiome, S0052), and the protein concentration was determined with BCA protein assay reagent kit. The values were normalized to total protein in tissue samples.
Reactive oxygen species determination H4 human neuroglioma cells were randomly divided into four groups including Ctrl group, SEV group, Egr2 shRNA group and SEV+ Egr2 shRNA group, and the regional mitochondrial ROS accumulation was measured by using the Mito-SOX reagent (M36008, Thermo Fisher, USA

Mitochondrial respiration analysis
The oxygen consumption rate (OCR) was measured by using a Seahorse XF96 analyzer (Seahorse Agilent, USA) combined with the Agilent Seahorse XFe96 Extracellular Flux Assay Kit according to the manufacturer's recommendations. Brie y, H4 cells were seeded in 96-wells of a Agilent Seahorse XF96 cell culture microplate (101085-004) and received corresponding treatment for 6 h. The culture medium was replaced with 175 μL assay medium, supplemented with 25 mM glucose, 2 mM glutamine, and 2 mM pyruvate on the day of the assay. Prior to the assay, plates were incubated at 37°C for approximately 1 h in a non-CO2 incubator. Afterwards, the basal OCR was determined followed by the automated injection of 25 μl oligomycin (8 μM), and mixing for 3 min and measurement for 2 min. Next, 25 μl carbonyl cyanide 4-(tri uoromethoxy) phenylhydrazone (FCCP) (9 μM) was injected, and OCR was measured for 2 min. Finally, a combination of 25 μl rotenone (20 μM) + antimycin A (100 μM) was injected, followed by the same mixing and measurement steps. ECAR was automatically recorded by the Seahorse XFe96 software, and the respiration rate was calculated by the Seahorse analyzer.
MitoTracker and ERTracker Imaging H4 cells were randomly divided into four groups (Ctrl group, SEV group, Egr2 shRNA group and SEV+ Egr2 shRNA group), and these cells received corresponding treatment for 6 h. Thereafter, the medium was replaced with prewarmed (37 °C) MitoTracker medium (50 nM, Invitrogen) for 5 min and ERTracker medium (100 nM, invitrogen) for 30 min. cells were permeabilized with 0.2% Triton® X-100 for 10 minutes and incubated with antibody diluent containing goat antibodies against Drp1 (1:100; ABclonal, A17069) overnight at 4 °C. Then sections were rinsed with PBS (3×10 min) followed by incubation with Alexa Fluor™ 594 goat anti-rabbit antibody for 1 h at room temperature. After rinsing with PBS (6 × 5 min), uorescence signals were visualized under an epi uorescence microscope. Images were captured with the assistance of Image-Pro Plus 5.0 software, and all the parameters used were kept consistent during capturing.

Morphological observation of mitochondria
The hippocampal tissues were xed with 2.5% glutaraldehyde overnight at 4 °C, and post-xed with 1% osmium tetraoxide for 2 h after rinsing for three times with phosphate-buffered saline (PBS). Then, the specimens were rinsed with distilled water followed by a graded ethanol dehydration series ending with propylene oxide. After in ltration in a mixture of one-half propylene oxide and one-half resin, the tissues were embedded in resin. Cross sections (120 nm) were made, which were stained with 4% uranylacetate for 20 min and 0.5% lead citrate for 5 min. The morphology of mitochondria in the hippocampal neurons was observed by using a transmission electron microscope (TEM) (Phliphs Tecnai 10, Holland) in the Center of Cryo-Electron Microscopy at Zhejiang University.
Prussian blue staining Sections (5 μm) were stained for Prussian blue reaction through an Iron Stain Kit (YEASEN, 60533ES20) according to the manufacturer's instructions. Brie y, slides were depara nized and hydrated to deionized water. Then, the samples were immersed in a freshly prepared solution of equal parts 5% potassium ferrocyanide and 5% hydrochloric acid for 10 min. Meanwhile, the samples were rinsed using deionized water, immersed in 2% pararosaniline solution for 5 min, and rinsed with deionized water once again, and immediately dehydrated and coverslipped. Images of positively stained sections were captured via an Olympus BX61 microscope.
Statistical analysis SPSS 19.0 software was used to process the data. All data are represented as mean ± standard deviation, and analyzed by one-way analysis of variance (ANOVA) and Tukey's post hoc test. P<0.05 was considered statistically signi cant.

Sevo urane Exposures induced Cognitive Impairment in developing mice
To determine the effect of sevo urane administration on cognitive function, behavioral tests including Morris Water Maze, novel object recognition test and trace fear conditioning were performed. Brie y, animals were exposed to 3% sevo urane for 2 h at P6-P8. Cognition was measured by the Novel object recognition (NOR) experiment at P30.The spatial memory abilities were evaluated at P37 by using the Morris Water Maze (MWM) test. And auditory trace fear conditioning, as a hippocampus-dependent learning task, was performed at P30. The results showed that the time course of mean escape latency was longer at third and fourth day in SEV group than that of in Ctrl group (Fig. 1A, P < 0.05). The time spent in the target quadrant and the total number of platform area crossings were signi cantly decreased in SEV group compared to Ctrl group (Fig. 1B-D, P < 0.05). The fear condition test showed that the freezing time was dramatically reduced in SEV group than that of in Ctrl group (Fig. 1E-G, P < 0.05). The novel object recognition test suggested that the overall distance of traveling was no signi cant different ( Fig. 1H, P > 0.05), while the recognition index was decreased in SEV group compared to Ctrl group ( Fig. 1I, P < 0.05).
Sevo urane administration induced the neuronal death and reduced the synapse formation in hippocampus The hippocampal tissues of mice were harvested at P42 after sevo urane administration at P6-P8, and the immuno uorescence staining was performed. The results showed that the relative uorescence intensity of Tuj1 were obviously increased, while the GFAP and NeuN uorescence intensity were decreased in SEV group including DG, CA1 and CA3 region compared to Ctrl group (Fig. 1J, K, M, N; P < 0.05). The TUNEL assay was used to visualize the apoptosis, and the results showed that the apoptotic cells were signi cantly increased in SEV group than that of in Ctrl group (Fig. 1L, O; P < 0.05). Meanwhile, the results displayed that the dendritic length, density and nodes were signi cantly reduced after sevo urane administration compared with Ctrl group indicated by Golgi-Cox staining (Fig. 1P, Q; P < 0.05).

Bioinformatics analysis indicated that Sevo urane administration resulted in the Egr2 expression signi cantly upregulated
The hippocampal tissue of mice was obtained at P30 after sevo urane treatment, and the RNA-Seq was performed. Then, to further clarify the potential mechanism, the data were analyzed by using integrated Differential Expression and Pathway analysis (iDEP) tools and String database. The results showed that the differential expression genes (DEGs) including 3 upregulated and 18 downregulated genes were screened by using IDEP tools with FDR < 0.05, Fold Change > 2 ( Fig. 2A), and these genes were enriched in various signal pathways consisting of learning or memory (Fig. 2B). Then, these hub genes were further analyzed through String database. Brie y, the GO function analysis as a dynamic controlled vocabulary is utilized to describe the role of gene with three categories information comprising of biological process (BP), cellular component (CC), and molecular function (MF). GO term enrichment analysis indicated that the BP was involved in 88 categories, and the top 10 BP was presented in Fig. 2C; the CC consists of 2 categories showed in Fig. 2D; and the MF includes 17 categories, and the top 10 MF was showed in Fig. 2E. The PPI network analysis was constructed for DEGs, and the results were showed in Fig. 2F, and the volcano plot showed the distribution of DEGs according to the Fold Change and P value (Fig. 2G, Table S3). Meanwhile, the bubble diagram showed the enriched pathways of DEGs (Fig. 2H). Gene coexpression was visualized by String tools, which revealed that Arc and Egr2 were coexpression in human beings and Mus nusculus (Fig. 2I), and may participated in the regulation of synapse formation [33,34]. Moreover, the genes related cognitive function and mitochondrial function were analyzed by String and KEGG database, and the results displayed that the genes mediating cognitive function were closely correlated to the mitochondrial related genes, and enriched in the same signaling pathway (Fig. 2J, K). Additionally, the Genecards database (https://www.genecards.org) was used to search the mRNA expression in normal human tissues from GTEx, Illumina, BioGPS, and SAGE (Serial Analysis of Gene Expression) for Egr2 Gene. Meanwhile, The Human Protein Atlas website (https://www.proteinatlas.org/) was utilized to retrieve this Egr2 expressional data separately obtained from Consensus Human Brain Dataset, GTEx Human Brain RNA-Seq Dataset and FANTOM5 Human Brain CAGE Dataset. And the results showed that the expression level of Egr2 was low specially in hippocampal formation (Fig. 2N-Q). The RNA-Seq results showed that the expression level of Egr2 was signi cantly upregulated in SEV group compared to Ctrl group at P6 and P30 (Fig. 2L). Similarly, the WB results indicated that the relative protein level of Egr2 was signi cantly increased in SEV group when compared with Ctrl group in animal hippocampus and primary cultured neuron, respectively (Fig. 2M, P < 0.05). These above data suggested that Egr2 played a vital role in sevo urane-induce cognitive dysfunction in developing mice.

Egr2 downregulation alleviated the cognitive de cits induced by sevo urane administration in developing mice
To determine the effect of Egr2 downregulation in cognitive function, the recombinant adeno-associated virus (AAV) was transfected into the hippocampus by intracerebroventricular injection assisted with stereotaxic apparatus. Six-day-old mice received anesthesia with 3% sevo urane 2 hours daily on postnatal days 6, 7, and 8. The mice were received Egr2 shRNA at P9 and sacri ced at P42 after behavior detection. The green uorescence suggested that the AAV successfully reached the ventricular injection sites (Fig. 3J). Then, the WB assay was performed and the results showed that the Egr2 protein level was signi cantly elevated in GFP-SEV group than that of in GFP-Ctrl group (Fig. 3K, P < 0.05). Whereas, the Egr2 protein level was obviously reduced in GFP-Egr2 shRNA-SEV group compared to GFP-SEV group (Fig. 3K, P < 0.05). Meanwhile, the behavioral tests including Morris Water Maze, novel object recognition test and trace fear conditioning were performed. The results showed that there were no signi cant different among the time course of mean escape latency, the time spent in the target quadrant and the total number of platform area crossings in GFP-Egr2 shRNA-Ctrl group when compared with GFP-Ctrl group (Fig. 3A-C; P > 0.05). However, the time course of mean escape latency was reduced at third and fourth day ( Fig. 3A; P < 0.05), and the time spent in the target quadrant and the total number of platform area crossings were obviously increased in GFP-Egr2 shRNA-SEV group when compared with GFP-SEV group (Fig. 3B, C, I; P < 0.05). The memory retrieval test showed that the freezing time was signi cantly increased in GFP-Egr2 shRNA-SEV group compared to GFP-SEV group (Fig. 3D-F; P < 0.05). The novel object recognition test displayed that the recognition index was reduced in GFP-SEV group when compared to GFP-Ctrl group, while which was increased in GFP-Egr2 shRNA-SEV group compared with GFP-SEV group (Fig. 3H, P < 0.05). Moreover, the results of Golgi-Cox staining indicated that the dendritic length, density and nodes were signi cantly increased in GFP-Egr2 shRNA-SEV group compared to that of in GFP-SEV group (Fig. 3L).
Egr2 downregulation partly improved mitochondrial dysfunction induced by Sevo urane administration Mitochondria are the cellular structures responsible for energy metabolism, and participate in various cellular biological processes. Previous study suggested that sevo urane inhalation may result in mitochondrial dysfunction by inducing reactive oxygen species formation [35]. Bioinformatic analysis displayed that Egr2 was closely correlated to the mitochondrial related genes, and enriched in the same signaling pathway associated with mitochondria (Fig. 2J, K). In this study, the mitochondrial membrane potential, morphology, membrane permeability and cytoplasm calcium level were detected in H4 cells by using JC-1 probe, MitoTracker staining, mitochondrial permeability transition pore (mPTP) assay and Fluo 4-AM calcium indicators, respectively. Meanwhile, the mitochondrial ultrastructure was observed by cryo-SEM, and the MDA and GSH level was measured in hippocampal tissues. Additionally, the OCR was evaluated and ECAR was automatically recorded by using a Seahorse XF96 analyzer to determine the respiration rate in H4 cells by using the Seahorse analyzer. The results of mitochondrial membrane potential showed that the red uorescence was decreased and green uorescence was increased in SEV group compared with Ctrl group (Fig. 4A, B; P < 0.05). Whereas, the red uorescence was increased and green uorescence was decreased in SEV + Egr2 shRNA group compared with SEV group (Fig. 4A, B; P < 0.05). Mitochondrial morphology was detected by MitoTracker staining and cryo-SEM, and the representative imaging showed that sevo urane administration accelerated the formation of fragmentation, reduced volume, intercristal space, and length in mitochondria compared with Ctrl group, while this status was partly reversed in SEV + Egr2 shRNA group when compared to SEV group (Fig. 4A, E; P < 0.05). mPTP assay was employed to detect the membrane permeability by observing the uorescence quenching. The results showed that the green uorescence was reduced in SEV group than Ctrl group, whereas which was effectively enhanced in SEV + Egr2 shRNA group compared with SEV group (Fig. 4A, C; P < 0.05). Meanwhile, the Fluo 4-AM calcium indicators displayed that the cytoplasm calcium levels were upregulated in SEV group compared to Ctrl group, while which was downregulated in SEV + Egr2 shRNA group compared with SEV group (Fig. 4A, D; P < 0.05). Moreover, the results showed that the concentration of MDA in hippocampus was no signi cant different between GFP-Egr2 shRNA-Ctrl group and GFP-Ctrl group, but which was obviously decreased in GFP-Egr2 shRNA-SEV group compared to GFP-SEV group (Fig. 4F, P < 0.05). Similarly, the level of GSH of hippocampuswas decreased in SEV group than Ctrl group, but which was effectively evaluated in SEV + Egr2 shRNA group compared to SEV group ( Fig. 4G, P < 0.05). Additionally, the DHE and mito-SOX staining were used to detect the level of intracellular and mitochondrial ROS in H4 cells, respectively. The data suggested that sevo urane administration obviously enhanced the the intracellular ROS level and the mitochondrial ROS level, while Egr2 gene silencing effectively reversed the ROS level in intracellular and mitochondrial compared with SEV group (Fig. 4H, P < 0.05). Furthermore, the OCR and ECAR assay were separately performed to investigate the mitochondrial respiratory function including anaerobic glycolysis and aerobic respiration in H4 cells. Sevo urane administration signi cantly suppressed mitochondrial respiration consisting of reducing aerobic respiration and enhancing anaerobic glycolysis (Fig. 4I, J). Particularly, the ATP production, basal respiration and maximum respiration were obviously decreased in SEV group than Ctrl group, whereas which were signi cantly elevated in SEV + Egr2 shRNA group compared with SEV group (Fig. 4K-M, P < 0.05).

Egr2 downregulation alleviates iron overload-induced ER-mediated mitochondrial ssion in hippocampal primary neurons
The mitochondrial de cits would elicit metabolic disequilibrium and eventually result in neurogenetic abnormal. Accumulating evidence revealed that mitochondria was closely correlated to iron homeostasis [36]. To investigate the interrelation between mitochondria dysfunction and iron homeostasis, the related protein and iron content were detected in H4 cells and hippocampal tissues. The results showed that the protein level of ACSL4 and COX2 was upregulated, but the GPX4 and FTH1 protein level was downregulated in SEV group compared with that of in Ctrl group (Fig. 5A). Meanwhile, the protein level of ACSL4 and COX2 was upregulated, while the GPX4 and FTH1 protein level was downregulated in GFP-Egr2 shRNA-SEV group compared with GFP-Ctrl group in neuron and hippocampus, respectively (Fig. 5B, C). Whereas, the protein level of ACSL4 and COX2 was reduced, while the GPX4 and FTH1 protein level was increased in GFP-Egr2 shRNA + SEV group compared to that of in GFP-SEV group in neuron and hippocampus, respectively (Fig. 5B, C). The iron assay showed that sevo urane administration signi cantly induced iron overload compared to Ctrl group in animal hippocampus (Fig. 5D, P < 0.05).
However, the iron content was reduced in GFP-Egr2 shRNA-SEV group compared to that of in GFP-SEV group in animal hippocampus (Fig. 5E, P < 0.05). Meanwhile, iron distribution was assessed histologically in hippocampus of animals by using Perls' Prussian blue staining. The results showed that the iron deposition was apparently elevated in SEV group than Ctrl group, but which of in SEV + Egr2 shRNA group were effectively reversed when compared to SEV group (Fig. 5H). To detect the level of lipid peroxidation, the uorescent reporter molecule C11-BODIPY 581/591 was employed in primary cultured neurons. The results showed that the uorescence partly shifted from red to green after sevo urane administration (Fig. 5G). The quantitative results showed that the red uorescence was decreased and green uorescence was increased in SEV group compared with that of in Ctrl group (Fig. 5G, K; P < 0.05). However, the red uorescence was increased and green uorescence was decreased in SEV + Egr2 shRNA group compared to SEV group (Fig. 5G, K; P < 0.05).
Moreover, to investigated the correlation between Egr2 expression and iron overload-induced mitochondrial ssion, the immuno uorescence co-localization analysis including MitoTracker (Red), ERTracker (Green), and Drp1 (Blue) were employed in hippocampus. The results showed that sevo urane administration induced iron overload, and thereby increases the co-localization of Drp1 puncta and expanded ER on mitochondria/fragmented mitochondria in hippocampal neurons when compared with Ctrl group; whereas, the co-localization among Drp1 puncta, expanded ER, and mitochondria/fragmented mitochondria was effectively reduced in SEV + Egr2 shRNA group than SEV group (Fig. 5F, J; P < 0.05). Additionally, the WB assay was performed for detecting iron metabolism and mitochondrial ssionrelated proteins, and the results showed that the protein level of DRP1, DMT1 and Ferroportin-1 was upregulated in SEV group than that of in Ctrl group, while which was downregulated in SEV + Egr2 shRNA group when compared with SEV group (Fig. 5I, P < 0.05).

The Behavioral Performance Was Effectively Improved After Deferiprone Administration
To investigate the effect of iron overload on cognitive de cits following sevo urane inhalation, the deferiprone (DFP) was administrated to evaluate the impact of iron chelating agent in mice after sevo urane treatment. Six-day-old mice received anesthesia with 3% sevo urane 2 hours daily on postnatal days 6, 7, and 8. The mice received deferiprone at P9 and were sacri ced at P42 after Morris Water Maze test. The results of Morris Water Maze showed that the time course of mean escape latency was increased in SEV group than Ctrl group, while which was reduced in SEV + DFP group compared with SEV group at third and fourth day ( Fig. 6A; P < 0.05). The time spent in the target quadrant and the total number of platform area crossings were obviously decreased in SEV group when compared with Ctrl group, but these of increased in SEV + DFP group compared to SEV group (Fig. 6B-D; P < 0.05). Meanwhile, the hippocampus was harvested for the iron content detection and western immunoblotting assay. The iron assay indicated that sevo urane administration signi cantly induced iron overload compared with Ctrl group in animal hippocampus (Fig. 6E, P < 0.05). Whereas, the iron content was reduced in SEV + DFP group than SEV group in animal hippocampus ( Fig. 6E; P < 0.05). Moreover, the WB assay showed that the protein level of ACSL4 and Cox2 was signi cantly increased in SEV group than Ctrl group, while which was reduced in SEV + DFP group when compared to SEV group ( Fig. 6F; P < 0.05). And the protein expression of FTH1 and GPX4 was decreased in SEV group than Ctrl group, while which was elevated in SEV + DFP group compared with SEV group ( Fig. 6F; P < 0.05)

Discussion
Cognitive dysfunction is a common complication involving in learning and memory de cits, attention and information processing anomalies, and personality and social ability disorders [37,38]. Related researches revealed that anesthesia accelerated the formation of extracellular amyloid plaques and intraneuronal neuro brillary tangles, elevated the generation and accumulation of Aβ, and subsequently exacerbates neuro brillary degeneration and induced nerve impairment [39]. Meanwhile, previous studies documented that anesthesia inhalation would result in direct toxic effects in neurons by inducing the dysregulation of calcium homeostasis and neurotransmitter release, aggravating the endogenous neurodegeneration processes, and inhibiting the physiological functions of neural stem cells [40]. Currently, sevo urane was extensively used in pediatric practice as an inhaled volatile anesthetic agent.
Emerging evidence demonstrated that sevo urane administration induced the neuroin ammation and neuronal damage, reduced the synaptic plasticity, and eventually resulted in cognitive impairment [41]. In this study, our results showed that the the learning and memory performance were signi cantly impaired following sevo urane inhalation in developing mice. Immuno uorescent staining displayed that the immature neurons were increased and the mature neurons were reduced in hippocampus. Golgi-Cox staining displayed that the dendritic length, density and nodes were obviously reduced after sevo urane administration. Most intriguingly, bioinformatics analysis showed that the Egr2 expression was signi cantly elevated, and closely correlated to mitochondrial function. Thereafter, the level of lipid peroxidation, mitochondrial membrane potential, morphology and membrane permeability, and cytoplasm calcium levels were investigated after Egr2 expression silence, respectively. Our results provide abundant evidence for clarifying the underlying mechanism regarding the cognitive dysfunction induced by sevo urane administration, which implicated in destroying mitochondrial respiratory network, elevating mitochondria ROS and reducing membrane potential, disturbing calcium homeostasis and iron content, and eventually resulting in cognitive de cits.
Mitochondria, as one of the chief sources of reactive oxygen species (ROS), play a crucial role in maintaining normal physiological activities. Excessive stimulation of NAD(P)H and electron transport chain would disrupt the normal redox state of cells and cause the overproduction of peroxides and free radicals, and thereby lead to the oxidative damage and mitochondrial dysfunction [42]. Meanwhile, the mitochondrial DNA impairment derived from accumulation of superoxide radicals would further amplify oxidative stress by mediating critical proteins, and initiating a vicious circle of ROS production to destroy the organelle, induce metabolic disequilibrium and genomic instability, and eventually result in cognitive impairment [43][44][45]. Hippocampus is one of the most vulnerable brain regions to oxidative damage, which is critical for the formation of long-term memory and learning [46]. Previous studies showed that Egr2 was intimately connected with the formation of myelination [47]. Egr2 expression was reduced after microneme-mediated attachment was blocked using a calcium chelator [48]. Meanwhile, Egr2 may participate in increasing adipocyte mitochondrial respiration and dampening oxidative stress reaction [49]. In this study, bioinformatics analysis revealed that Egr2 may implicate in the mitochondrial metabolic process. And the results showed that sevo urane administration accelerated the formation of fragmentation, reduced volume and intercristal space, elevated the intracellular ROS level in hippocampal mitochondria, and consequently resulted in the impairment of behavioral outcomes. Interestingly, we found that Egr2 expression silencing could effectively reverse the elevation of cytoplasm calcium content, reduction of the mitochondrial membrane potential and the opening of mPTP following sevo urane administration.
Iron is a crucial component for biochemical reactions including cellular metabolism, synthesis of DNA, RNA and proteins, enzymatic reactions, and synthesis of myelin [50]. Accumulating evidence revealed that neurons were particularly vulnerable to the alteration of iron content, and iron homeostasis disorder triggers a serial cascade of pathophysiologic reactions including neurogenetic abnormality, disturbing neurotransmitter synthesis and release, and mitochondrial dysfunction [51]. Previous studies showed that iron participated in the generation of ROS, and caused aggregation and phosphorylation of tau, and thereby aggravated the toxicity by mediating DNA oxidation, lipid peroxidation, accumulation of advanced glycation end products, malondialdehyde and peroxynitrite in Alzheimer patients [52]. Recently, related research proved that sevo urane disrupted iron homeostasis by affecting the protein expression and mitochondrial iron accumulation [53]. Mitochondria are major generators of iron-sulfur clusters (ISC), and tightly regulated iron uptake and utilization [54]. Additionally, mitochondrion could effectively catalyze electron transport through heme-and ISC -containing proteins to process energy transduction owing to the reversible oxidation states of iron. The prevailing hypothesis indicated that the mitochondrial iron content was affected by the labile iron pool in the cytosol, and Fe 2+ was transported into the mitochondria by binding with hydrophobic pockets of chaperone proteins [55]. Meanwhile, Egr2 gene may correlated with the HO-1 expression, which could convert heme to iron, and participated in memory, cognition and synaptic plasticity [56]. Furthermore, Egr2 may implicate in stimulating iron acquisition in pro-in ammatory conditions [57]. Consequently, we speculated that Egr2 may play a crucial role in cognitive function correlated to mitochondrial iron metabolism. In this study, the WB assay showed that the protein level of ACSL4, COX2, Ferroportin1 and DMT1 was upregulated, and the GPX4 and FTH1 protein level were downregulated after sevo urane administration, while which were reversed when Egr2 expression was suppressed. Meanwhile, the iron assay showed that sevo urane administration signi cantly induced iron overload, and the iron deposition was apparently elevated in hippocampus indicated by Perls' Prussian blue staining.
Collectively, these ndings documented that sevo urane administration reduced the mitochondrial membrane potential, and elevated the mitochondrial membrane-permeability, and further initiated iron dyshomeostasis. These changes facilitated the neuronal dysfunction and eventually resulted in cognitive de ciency, whereas suppressing Egr2 expression partly reversed this pathological process.  There is no significant difference in Ph, PaCO2, PaO2, Glucose and SaO2 between the groups. PaCO2 = arterial carbon dioxide tension; PaO2 = arterial oxygen tension; SaO2 = arterial oxygen saturation.