2.1 Delirium-Like Behavior and Neuroinflammation Were Induced by Tibial Surgery in T2DM Rats

We found that normal rats who received 30 min of sevoflurane anesthesia or tibial surgery did not exhibit altered behavior (Figure S1A–E). Regarding T2DM rats, significantly increased latency to find pellets was observed at 1 and 3 days after tibial surgery (Figure 1B). On the third day after surgery, comparable movement capacity among diabetic groups suggested that the poorer performance (Figure 1C) in the DMAS group resulted from impaired natural behavior or olfactory function.

On Days 1 and 3 following tibial surgery, freezing time was significantly reduced in the context test. A decrease in freezing time was not observed in the DMAS group during the tone test (Figure 1E,F). Thus, only hippocampal-dependent spatial memory was impaired in the DMAS group. Moreover, 30 min of sevoflurane anesthesia did not cause any behavioral changes in normal adult or T2DM rats. Hence, our results suggest that orthopedic surgery plus anesthesia could induce delirium-like behavior in T2DM rats but not in normal adult rats.

On Days 1, 3, and 7 postoperatively, hippocampal samples were collected to determine pro-inflammatory cytokine levels. On Day 1 post-surgery, the levels of interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) were significantly increased in surgical groups compared with nonsurgical groups. On Postoperative Day 3, the DMAS group had a higher level of neuroinflammation, which persisted until 7 days after tibial surgery. Furthermore, 30 min of sevoflurane anesthesia alone did not induce any changes in pro-inflammatory cytokines in normal (Figure S1F–H) or T2DM rats (Figure 1G,L,M).

Tibial surgery induced significant morphological changes in astrocytes and microglia in the hippocampus, which displayed amoeba-like morphology following tibial surgery. GFAP and Iba-1 expression levels were significantly increased in the DMAS group, but there were no significant changes in the DMA group (Figure 1H–K,N,O) on Day 3 post-surgery.

2.2 Mitochondrial Dysfunction Was Induced in the Hippocampus of T2DM Rats After Tibial Surgery

In the hippocampal tissue of normal adult rats, there were no significant changes in the ATP content or ΔΨ_m level (Figure S1I–K), except that the reactive oxygen species (ROS) level in the ConAS group was higher than that in the Con group on the first day after surgery (Figure S1J). Meanwhile, T2DM rats showed a significant decrease in ATP production and the ΔΨ_m level, along with a significant increase in ROS production, the ΔΨ_m level, and a significant increase in ROS production after tibial surgery. Moreover, the reduced ΔΨ_m level and high ROS production in the hippocampus of diabetic rats persisted until 7 days after surgery (Figure 2A–C).

Hippocampal protein and mRNA levels of molecules related to mitochondrial biogenesis and dynamics of normal adult rats that underwent tibial fracture surgery were not significantly altered compared with those of the nonsurgical group (Figure S1L,N,O), and there were no statistically significant group differences in mtDNA/nuclear DNA (nDNA) (Figure S1M). mtDNA/nDNA in the DMAS group showed a significant reduction compared with nonsurgical groups. However, 30 min of sevoflurane anesthesia did not induce significant changes (Figure 2G). In T2DM rats, the mRNA levels of all mitochondrial biogenesis-related proteins were significantly reduced in the hippocampus after tibial surgery compared with the nonsurgical group, where the levels of each protein indicate mitochondrial biogenic dysfunction (Figure 2D–F).

In normal adult rats, mitochondrial function was sufficient to overcome surgical stress (Figure S1I–K), whereas in T2DM rats, impaired mitochondrial function exacerbated dysfunction after surgical stress. A significant increase in mitochondrial number and damage, along with a decrease in length (Figure 2J–L), occurred in the DMAS group. Fragmented mitochondria with vacuoles and no cristae structure indicated mitochondrial damage after tibial surgery in T2DM rats. A more detailed analysis of single mitochondria was then performed using transmission electron microscopy (TEM) images, and form factor (FF) and aspect ratio (AR) values were calculated (Figure 2I). FF has a value of 1 when the mitochondria form a perfect circle. A value of FF > 1 indicates that mitochondria are longer and have more complex shapes [22]. AR is the ratio between the major and minor axes of the mitochondrial ellipse, indicating elongation and fragmentation of the mitochondria. After tibial surgery, there was a significant reduction in the FF and AR in the DMAS group compared with the nonsurgical group (Figure 2M,N).

2.3 Swimming Exercise Attenuated Delirium-Like Behavior and Enhanced Mitochondrial Function Following Tibial Surgery in T2DM Rats

Four weeks of swimming exercise alleviated insulin resistance and lipoatrophy in T2DM rats. Specifically, the fasting serum insulin and fasting glucose levels of the DMEAS group significantly decreased, whereas body weight increased, compared with the sedative group (Figure 3B–D). T2DM rats that completed 4 weeks of swimming exercise displayed improvement in POD-like behavior. The latency to find pellets was significantly decreased (Figure 3E), and the freezing time in the context test was significantly increased (Figure 3H) in T2DM rats that engaged in swimming exercise compared with sedated rats. In the CA1 region of T2DM rats, the immunoreactive areas of GFAP and Iba1 were reduced after swimming exercise in comparison with sedentary rats following tibial surgery (Figure 3J–O). On the third postoperative day, the levels of pro-inflammatory cytokines were also significantly reduced after exercise pretreatment (Figure 3P).

T2DM rats that engaged in swimming exercise showed a significant increase in ATP production and ΔΨ_m levels, as well as a significant decrease in ROS levels compared with those without exercise (Figure 4A–C). Polymerase chain reaction (PCR) results showed that Ppargc1a, Nrf1, and Tfam expression levels were significantly upregulated in the exercise group compared with the sedative group (Figure 4D). Consistently, WB analysis confirmed that the protein levels of PGC-1α, NRF1, and TFAM were significantly increased in the DMEAS group compared with the DMAS group (Figure 4E–G). TEM showed that swimming induced a significant increase in mitochondrial length and cristae structure (Figure 4H). After exercise and surgical stress, the mitochondrial number did not change significantly, but mitochondrial fragmentation and damage percentage were reduced, the length was extended, and mitochondrial cristae were clearer than in the DMAS group (Figure 4I–K). Further, swimming pretreatment improved the FF and AR, indicating better-preserved mitochondrial morphology and elongation of the mitochondrial structure (Figure 4L,M).

2.4 SIRT2 Deficiency Reversed Exercise-Mediated Cognitive Improvement and Mitochondrial Function in T2DM Rats Following Tibial Surgery

SIRT2 was screened as the target molecule that may ameliorate POD-like behavior in T2DM rats (Figure S2A). To explore whether the ameliorative effect of swimming exercise pretreatment is dependent on the effect of SIRT2 on mitochondrial biogenesis-related proteins, we delivered a SIRT2-specific inhibitor, AGK2 (IC₅₀ = 3.5 µM), via stereotactic injections into the CA1 hippocampal brain of T2DM rats to downregulate SIRT2 following swimming exercise. The DMEAS-AGK2 group took longer to find pellets and had shorter freezing times in the context test (Figure 5A–E). These results suggest that inhibition of SIRT2 expression in the CA1 region of the hippocampus suppressed the ameliorative effect of swimming exercise pretreatment on POD-like behavior in T2DM rats.

Mitochondrial function results revealed decreased ATP content and ΔΨ_m levels and increased ROS production in hippocampal tissue in the DMEAS-AGK2 group compared with the DMEAS group (Figure 5F–H). ELISA showed that the injection of solvents into the CA1 hippocampal area did not significantly change the levels of the pro-inflammatory factors (Figure 5I). The PCR results showed that the expression of Ppargc1a, Nrf1, and Tfam was significantly reduced in the DMEAS-AGK2 group compared with the DMEAS group (Figure 5J). WB showed that the expression of SIRT2, NRF1, and TFAM was downregulated in the DMEAS-AGK2 group compared with the DMEAS group, whereas the expression of PGC-1α was not significantly altered (Figure 5K,L). The mtDNA/nDNA ratio in the DMEAS-AGK2 group was considerably lower than in the DMEAS group (Figure 5M).

2.5 Overexpression of SIRT2 Ameliorated High Glucose Plus Conditioned Culture Media-Induced Impairment of Neuronal Mitochondrial Biogenesis

The extraction process of conditioned culture media (CCM) and the optimal stimulation conditions for HT22 cells were described in Supporting Information. ELISA of pro-inflammatory factors showed that the levels of IL-1β, IL-6, and TNF-α were significantly higher in the high glucose plus conditioned medium (HG+CCM) groups compared with the normal medium (Ctrl) groups. The expression levels of IL-1β, IL-6, and TNF-α were significantly lower in the Lv5-SIRT2-HG+CCM group compared with the HG+CCM group (Figure 6B). The protein expression of NRF1 and TFAM was significantly increased in the Lv5-SIRT2-HG+CCM group compared with the HG+CCM group, while PGC-1α expression was not significantly different between the groups (Figure 6E,F). The mtDNA/nDNA levels were substantially lower in the HG+CCM group than in the Ctrl group, whereas the mtDNA/nDNA levels were significantly higher in the Lv5-SIRT2-HG+CCM group compared with the HG+CCM group (Figure 6G).

2.6 SIRT2 Ameliorated Neuronal Mitochondrial Biogenesis Dysfunction by Upregulating PGC-1α Deacetylation

In HT22 cell lines, immunofluorescence results showed that SIRT2 and PGC-1α were significantly colocalized in the cytoplasm, neuronal SIRT2 and PGC-1α expression was downregulated in the HG+CCM treated group, and PGC-1α expression was not significantly altered in the Lv5-SIRT2-HG+CM group. These results were consistent with those of WB (Figure 6C). Immunoprecipitation was applied to detect the effect of overexpression of SIRT2 on the acetylation level of PGC-1α in the HG+CCM group. The ratio between total acetylated PGC-1α in IB and total PGC-1α in Input showed that the total acetylation level of PGC-1α in the Lv5-SIRT2-HG+CCM group was significantly reduced compared with that in the HG+CCM group (Figure 6D).

The mitochondrial function results showed that ATP content and MMP significantly decreased, whereas MitoSOX levels increased dramatically in the HG+CCM group compared with the Ctrl group. Moreover, ATP content and MMP were increased, whereas MitoSOX levels were decreased in the Lv5-SIRT2-HG+CCM group compared with the HG+CCM group (Figure 6H–J). The mitochondrial function results are shown in fluorescence images of JC-1 monomer/polymer and MitoSOX; the representative fluorescence images were consistent with the functional assay results (Figure 6K).

5.1 Animals

Animal Ethics Committee approval was obtained (LA2021041), and the study was conducted following the ARRIVE 2.0 guidelines for the Care and Use of Animals in Research [48]. Adult male Sprague Dawley rats (weight: 200 ± 20 g) were purchased from Huafukang Bioscience Co. Ltd. (Beijing, China). Group-housed (three per cage) rats were maintained at room temperature (24 ± 1°C), with humidity of 50%–60% and free access to food and water. Rats were given 1 week to acclimate to their housing conditions before the experiment began.

Induction of T2DM in rats was achieved using a high-fat diet (HFD) and low-dose streptozotocin (STZ) [49]. Three groups of rats were randomly divided after 1 week: a control group (n = 60), an HFD group (n = 90), and an HFD + exercise group (n = 60). The rats in the HFD group were fed an HFD (D12492, Research Diets Inc., New Brunswick, NJ, USA) for 4 weeks. In the fourth week of HFD, rats were fasted for 12 h with free access to water, and each rat was injected intraperitoneally with 30 mg/kg STZ (Sigma-Alrich, St. Louis, MO, USA). A regular diet was fed to rats in the control group, and the same dosage of citrate buffer was injected. Blood glucose levels were measured 3 days after STZ injection from the tail vein. Further investigations were conducted on rats with random blood glucose > 16.7 mmol/L in the HFD group [50]. A total of 160 rats were used for T2DM modeling, with a success rate of 93.75%, and 150 rats ultimately met the criteria for successful induction. Rats that failed to meet the diagnostic criteria after injection were excluded from the study.

5.2 Swimming Exercise

We trained T2DM rats five times a week in a swimming pool (160 cm diameter, 50 cm height) filled with water at 30 ± 1°C. The swimming program included an adaptive phase and an exercise phase. To decrease water-induced stress, swimming exercises of gradually increasing duration (10, 20, 30, 40, and 50 min/day) allowed for acclimation in the adaptive phase. Three 20-min swimming sessions were followed by a 10-min rest period. Animals were dried with towels after swimming and kept warm with electric heaters for 15 min afterward.

5.3 Surgical Trauma

We performed an open tibial fracture with intramedullary fixation of the left hind paw in aseptic surgical conditions, with a modification of the anesthesia protocol [51]. Animals were given a general anesthetic (3% sevoflurane in 0.30 FiO₂). Shaving and disinfection were performed on the rats' left tibias. Following a median incision on the lower two-thirds of the tibia, a 20-G pin was inserted into the intramedullary canal. A periosteal stripping and osteotomy were then performed. After producing a tibial fracture, the wound was irrigated, and the skin was sutured with 8/0 Prolene sutures. A warming pad was used to maintain the temperature while the animals spontaneously recovered from the anesthetic. The total duration of anesthesia and surgery was fixed at 30 min, a 2% lidocaine solution was applied locally before the incision, and 1% a tetracaine hydrochloride mucilage was applied to the wound twice daily until Day 3.

5.4 Behavioral Tests

POD is characterized by acute concurrent disturbances at diverse cognitive levels affecting natural and learned behaviors [52]. Animal models of delirium have mainly been used to test natural and learned behaviors [10, 27, 53]. The buried pellet test (BPT) [11], open field test (OFT) [54], and FCT [55] were used to assess POD. The BPT and OFT assess changes in the natural behavior of rats, while the FCT assesses changes in learned behavior.

Using the BPT, rats were assessed for their natural tendency to locate olfactory cues [56]. Three pieces of sweet cereal were given to each rat 2 days before the test. Rats were habituated to the testing environment for at least 1 h before being tested in their home cages. During habituation, the test cage (48 cm length × 25 cm width × 20 cm height) was prepared with 5-cm-deep clean bedding, and a piece of sweetened cereal was randomly buried 3.8 cm below the surface of the bedding. After the rat had been in the cage for 5 min, its latency to eat the food was measured as the time it took for it to uncover the pellet and grasp it in its forepaws or teeth. A latency of 300 s was recorded if the rat failed to find the pellet within 5 min.

After the BPT and 30 min before the FCT, the OFT was conducted. Under dim lighting, rats were gently placed in an open field chamber and allowed to move around freely for 5 min. Using SMART 3.0.1 (San Diego Instruments, San Diego, CA, USA), movement parameters were automatically calculated. Rat locomotor activity and anxiety are reflected in the total distance traveled and time spent in the central zone. To prevent the presence of olfactory cues, the chamber floor was cleaned with 75% ethanol after each test.

There are two phases to the FCT test: a training phase before surgery and a test phase after surgery at 1, 3, and 7 days. The rats underwent fear conditioning training 1 day before surgery, which involved being placed in a conditioning chamber for 180 s and exposed to three pairs of conditional–unconditional stimuli, followed by an additional 60 s in the chamber. The pairs of conditional stimuli consisted of a 60 s, 80 dB sine wave tone (conditional stimulus) and a 1 s, 0.8 mA electrical foot shock. FCT tests hippocampal-dependent memory in the context test and hippocampal-independent memory in the tone test [57]. Rats were simply placed back into the conditioning chamber without any tones or foot shocks during the context test. During the tone test, the rats were placed in a novel chamber with a different painting for 5 min, followed by 3 min of tone delivery. A tracking software recorded the percentage of rat freezing time.

5.5 Levels of Inflammatory Factors

Hippocampal levels of the pro-inflammatory markers IL-1β, IL-6, and TNF-α were quantified via enzyme-linked immunosorbent assay (ELISA) kits (Neobioscience, Shenzhen, China), following the guidelines provided by the manufacturer. Results are expressed as pg of cytokine per mg of tissue.

5.6 Western Blot

Using RIPA buffer containing phosphatase and protease inhibitors, the hippocampus and cells were lysed. Following the extraction of proteins, the supernatants were centrifuged for 15 min at 4°C and quantified with the BCA protein assay kit (Beyotime Biotechnology, Shanghai, China). Proteins were separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes in equal amounts. 5% fetal bovine serum (FBS) in Tris-buffered saline containing 0.1% Tween 20 was used to block the membranes, and the following primary antibodies were incubated overnight at 4°C: anti-PGC-1α (Abcam, ab106814), anti-NRF1 (Abcam, ab175932), anti-TFAM (Abcam, ab272885), anti-Drp1 (Santa Cruz, sc-271583), anti-Fis1 (Santa Cruz, sc-376447), anti-OPA1 (Santa Cruz, sc-393296), anti-Mfn1 (Santa Cruz, sc-166644), anti-Mfn2 (Santa Cruz, sc-515647), anti-SIRT2 (Abcam, ab211033), and anti-β-actin (Santa Cruz, sc-47778). Secondary antibodies were incubated for 1 h at room temperature, followed by an enhancement of chemiluminescence and subsequent analysis with Gel-Pro software (Media Cybernetics, Rockville, MD, USA).

5.7 Quantitative Real-Time PCR

With the help of the TRIzol reagent, total RNA was extracted from the hippocampus or cells, and complementary DNA was synthesized with a reverse transcription kit (Invitrogen, Carlsbad, CA, USA). An ultraviolet spectrophotometer was used to measure total RNA concentration (Nanodrop ND-100, Thermo Fisher Scientific, Wilmington, DE, USA). A260/A280 values all fell between 1.8 and 2.0, indicating high purity. Shenggong Biotech Co. Ltd. (Shanghai, China) designed and synthesized the primers. PCR was performed using SYBR Green qPCR Master Mix (Tiangen, Beijing, China). Expression levels were normalized to Actb levels and calculated using the formula 2^−ΔΔCt.

5.8 Measurement of mtDNA Copy Number

We extracted genomic DNA from hippocampal tissue and HT22 cells using the DNeasy Blood & Tissue Kit (Qiagen, Shanghai, China) as directed by the manufacturer [58]. A real-time quantitative PCR was used to determine the copy number of mtDNA/nDNA. The ABI Prism 7500HT system (Applied Biosystems, Waltham, MA, USA) was used to perform real-time quantitative PCR experiments with TB Green Premix Ex Taq II (TAKARA, Shiga, Japan). The primers used to amplify mtDNA and nDNA were listed in Supplementary [3, 59–61]. The relative mtDNA copy number was calculated with the 2^−ΔΔCt method [62].

5.9 Immunofluorescence

Transcardial injections of phosphate-buffered saline (PBS, pH 7.3) and 4% paraformaldehyde were administered to rats. The tissues were embedded in OCT, and 10 µM sections were prepared. Iba-1 (Merck Millipore, 019–19741) and GFAP (CST, 3670S) expression levels were investigated to elucidate the activation of microglia and astrocytes. Antibodies against GFAP and Iba-1 were incubated with sections, followed by 1 h of secondary antibody incubation with Alexa Fluor CY3 (Jackson, 111-165-003, 115-165-003). Images of the slides were then taken (Leica Microsystems, Wetzlar, Germany).

5.10 Adenosine Triphosphate Measurement

An ATP assay kit (Beyotime Biotechnology) was used to measure hippocampal ATP. Each sample was tested for protein concentration using a BCA protein assay kit (Beyotime Biotechnology). Amounts of ATP are expressed as nmol/mg protein.

5.11 ROS Measurement

ROS levels were quantified by dichlorodihydrofluorescein diacetate (DCFH-DA) [63]. Using cool Locke's buffer at 1:20, hippocampal homogenates were diluted to 2.5 mg tissue/500 µL. Lock's buffer (1 mL, pH 7.4) and 200 mL of homogenate were mixed in the reaction mixture, and 10 mL of DCFH-DA (5 mM) was incubated at room temperature for 15 min, followed by analysis with a spectrofluorometer (Promega, Madison, WI, USA) with excitation at 484 nm and emission at 530 nm. Parallel blanks were used to measure background fluorescence. The expression of ROS was quantified on a histogram.

5.12 Isolation of Mitochondria

The mitochondrial fraction was isolated using the Qproteome Mitochondria Isolation Kit (Solarbio Science & Technology Co., Beijing, China). Fresh mitochondria were used for membrane potential detection.

5.13 Measurement of Mitochondrial Membrane Potential

By staining hippocampal tissue with JC-1, we quantified the red/green ratio. A mitochondrial extraction kit (Solarbio) was used to extract the mitochondria from hippocampal tissue. The MMP of mitochondria in each group was determined using an MMP assay kit (Beyotime Biotechnology) with JC-1 according to the manufacturer's instructions. The isolated mitochondria suspension was incubated with 1.5 µM of the JC-1 dye for 30 min at 37°C, rinsed twice with PBS, and then detected with a fluorescence spectrophotometer (F-4500, Hitachi, Tokyo, Japan). The red/green fluorescence intensity ratio was used to determine MMP.

5.14 Transmission Electron Microscopy

TEM was performed [64]. Briefly, rats were perfused with PBS and injected with a mixed fixative solution (25% glutaraldehyde and 0.2 M PBS). Hippocampal tissue was cut into 1 mm³ pieces and fixed in 2.5% glutaraldehyde for 24 h, followed by 1% osmic acid for 2 h. The tissue was then dehydrated with acetone and embedded in epoxy. Using an ultramicrotome, ultrathin slices (70–80 nm) were cut, and hippocampal mitochondria were examined using JEOL JEM-1400 at 1000× magnification.

5.15 Stereotaxic Injection

Sevoflurane was used to anesthetize the rats, which were then positioned in an animal stereotaxic apparatus (RWD Life Science, Shenzhen, China). The skull hole was made with a microsurgical drill and injected with a 5 µL Hamilton syringe (Hamilton, Bonaduz, Switzerland). Drug administration was controlled by a micro syringe pump controller (RWD Life Science, Shenzhen, China), and the syringe was pulled out 10 min after drug administration. All drugs were injected using the exact stereotaxic coordinates (measured from bregma: −3.96 mm anteroposterior, ±2.5 mm mediolateral, and −3.15 mm dorsoventral). SIRT2 inhibitor AGK2 (MedChemExpress, HY-100578) was dissolved in 4% dimethyl sulfoxide and diluted in PBS. The volume of AGK2 (200 µM) was 2.5 µL per side [65]. To avoid surgical damage, control rats received an injection of vehicle solution (2.5 µL 0.1 M PBS per side). Tibial fracture surgery was performed following the stereotaxic injection.

5.16 Lentivirus Transfection

To overexpress SIRT2 in HT22 cells, we employed lentivirus transfection technology. HT22 cells were transfected with Lentivirus5-Sirt2 or Lentivirus5-NC (GenePharma, Shanghai, China) according to relevant instructions.

5.17 Mitochondrial Superoxide Determination

HT22 cells were grown in six-well glass-bottom plates at a density of 2 × 10⁴/well. A 5 mM MitoSOX reagent stock solution (Thermo Fisher Scientific, M36008) was prepared using 2 mL of 5 µM MitoSOX reagent working solution. The cells adhered to coverslips were covered and incubated at 37°C for 10 min in the dark. Cells were then washed gently three times with a warm buffer and mounted in the warm buffer for imaging. MitoSOX was detected using a confocal fluorescence microscope (Axio Observer Z1; Zeiss, Oberkochen, Germany) at 510/580 nm excitation/emission.

5.18 Immunocytochemistry

HT22 cells were washed with PBS, fixed for 10 min (4% paraformaldehyde solution), and then washed three times with PBS for 15 min. The cells were then permeabilized for 5 min with 0.2% Triton X-100 and blocked with 10% bovine serum albumin in PBS. The slides were incubated with anti-PGC-1α (1:100) and anti-SIRT2 (1:100) overnight at 4°C. After washing three times, the cells were then incubated for 1 h with Alexa Fluor 488 (Abcam, ab150129) and Alexa Fluor CY3 (Jackson, 111-165-003) secondary antibodies at 37°C, and the nuclei were stained with DAPI for 8 min. Fluorescence images were obtained with a confocal fluorescence microscope.

5.19 Coimmunoprecipitation

The total protein in the cells was extracted and incubated with protein A/G-coupled magnetic beads (Thermo Fisher Scientific, 20421) with PGC-1α antibodies (Santa Cruz, sc-518025) at 4°C overnight. Using Tris-buffered saline buffer, the beads were washed four times and the supernatant was discarded, boiled at 95°C for 10 min, and eluted with sodium dodecyl sulfate loading buffer. Using antiacetylated lysine antibodies (CST, 9441S), immunoblot analysis was performed.

5.20 Statistical Analysis

The statistical analysis was conducted using GraphPad Prism 7 software (GraphPad Software Inc., La Jolla, CA, USA). The Shapiro–Wilk test was used to confirm the normality of the data distribution. Mean ± standard deviation is used to present quantitative data. Statistical significance was determined using analysis of variance followed by Bonferroni's post hoc test. The means of the two groups were compared using two-way unpaired Student's t-tests. p < 0.05 indicated statistical significance.