Isoflurane and Sevoflurane Induce Cognitive Impairment in Neonatal Rats by Inhibiting Neural Stem Cell Development Through Microglial Activation, Neuroinflammation, and Suppression of VEGFR2 Signaling Pathway

All animal experimental protocols were approved by the Animal Care and Use Committee of Wenzhou Medical University (Wenzhou, Zhejiang, China), and all procedures were performed following the National Institutes of Health (NIH, Bethesda, MD, USA) guidelines of animal care. The P7 rats were housed under a 12:12 h light–dark cycle at 22–24 ℃ ambient temperature with their parents. They were randomly assigned to the neonatal rats control (CON), neonatal rats anesthetized with isoflurane (ISO), and sevoflurane (SEV) groups.

The P7 rats were anesthetized at 1 minimum alveolar concentration (MAC) as determined by tail clamping. The 1 MAC of isoflurane or sevoflurane was determined to be 1.1% or 2.0% concentrations, respectively. The ISO or SEV group was flushed continuously with isoflurane or sevoflurane and 30% oxygen for 4 h. The CON group received 30% oxygen for 4 h at the identical condition. During anesthesia, neonatal rats breathed spontaneously. The temperature of the chamber floor was kept at 37 ℃ and was covered with soda lime. The concentration of isoflurane was continuously monitored using a gas analyzer (ARYM-0054 Vamos, Dräger, Germany). When the rats were exposed to anesthetics in chambers, respiratory rate and invasive arterial blood pressure were monitored and blood was sampled for blood gas analysis. Anesthesia was terminated by discontinuing isoflurane or sevoflurane. Then the rats were kept in a chamber containing 30% oxygen until the return of righting reflex. Subsequently, they were returned to their home cages with their parents.

The rats were administered bromodeoxyuridine (BrdU; Sigma, St. Louis, MO; 20 mg/ml) at 100 mg/kg/d for 4 d (intraperitoneally; dissolved in PBS) after isoflurane or sevoflurane exposure.

BrdU has been a principal marker for mitotic cells in studies of the neurogenesis (Gratzner 1982). This method labels freshly divided neural stem/progenitor cells and benefits from its long-term retention in divided cells and its passage to their daughter cells (Schmuck et al. 2014). This feature can be used to trace the cell lineage and cell survival.

Blood pressure and blood gases were measured in a separate cohort (n = 8/group) as previously described to confirm whether such an anesthesia regimen affects cardiorespiratory function. Briefly, arterial blood was sampled in the ISO or SEV rats via a 24-gauge arterial puncture needle (IntroCan®-W, Braun Medical Inc., Bethlehem, PA, USA) through the abdominal aorta using a dissecting microscope (PS100, Nikon, Tokyo, Japan). The mean arterial blood pressure (MAP) was measured by an anesthesia monitor (M3046, Philips Medical System, Boeblingen, Germany). The blood sample (0.2-0.03 ml) was immediately analyzed to determine pH, arterial oxygen, and carbon dioxide using the blood gas analyzer (GEM Premier 3000, Bedford, MA, USA).

The open-field test (OFT) was done at 2 weeks after the isoflurane or sevoflurane exposure. The animals were brought to the experimental room 5–20 min before testing to allow habituation. Each animal was placed in a corner square of the open field and faced the corner. Subsequently, each animal was observed for 5 min each time for three times. After 5 min, the animal was removed. Then the movement distance, time spent, and number of entries to the central region of the animal were recorded.

The MWM was done at 2 weeks (P21) and 6 weeks (P49) after the isoflurane or sevoflurane exposure. Before the trials, all rats were placed in the water of the swimming arena with a 6 cm diameter platform submerged 0.5–1 cm above the surface of the water on day 0. Each rat was allowed to swim for 120 s to locate the platform. The rats that had vision problems or did not swim were removed from the arena and excluded from further experiments. Subsequently, spatial acquisition trials were conducted, where the platform was submerged 0.5–1 cm below the surface of the water. The animals underwent four trials each day in the pool at four different starting positions facing the tank wall. A time limit of 2-min per trial allowed for rats to find the platform within a 30-min inter-trial interval. The animals not finding the platform within the time allotted were guided onto the platform for 15 s. The swim speed, distance (path length), and time (escape latency) in finding the platform were calculated from the recorded videos using the MWM software (SLY-WMS Morris water maze, Shuolinyuan, Beijing, China). On day 6, a probe trial was performed, during which the platform was removed, and animals were placed in a novel start position 180° from the original platform position to swim freely for 30 s. The percentage of time spent in the target quadrant and the time on platform–site crossovers were recorded.

The T maze was done at 2 weeks and 6 weeks after the isoflurane or sevoflurane exposure. Before trials, each animal was maintained at 90%-95% of its free-feeding body weight. Then the animals were placed into a T maze for 3 min each day for successive four days. For the forced trials, the reward food (milk) was placed in the food well at one goal arm, and the other goal arm door was blocked. After opening the central partition at the start arm, each animal, which was placed in the start area, ran for the reward food. When the animal consumed all the reward food, the animal was returned to the start arm and the start arm door was then closed. Then, for rewarded alternation trials, each animal was stood in the start area facing away from the goal arms for 15 s. After 15 s, the central partition and the doors of the goal arms were removed. The animal was allowed to choose between the two open goal arms. The time was allowed to consume the reward if correct. If the animal chose incorrectly the animal was removed after a time period equivalent to the time normally used to consume the reward to ensure that it had definitely discovered that the sample well was empty. As with rewarded alternation, each trial took no more than 2 min. The two trials (forced trials and rewarded alternation trials) were alternated, and the reward food at one goal arm was random, and the numbers of reward food were equal for two goal arms. The numbers of correct alternations were recorded for each animal.

Eight animals from each group were deeply anesthetized with 5% chloral hydrate and transcardially perfused with normal saline 5 days after anesthetics exposure. The brains were quickly removed. The hippocampus tissue was homogenized to a mixture composed of RIPA lysis buffer, phosphatase, and protease inhibitors, and incubated for 30 min on ice. The lysate was then sonicated and centrifuged at 12,000 rpm for 30 min at 4 ℃. The protein samples were quantitated using bicinchoninic acid (BCA) protein assay kit (Thermo Scientific, Eugene, OR, USA) and the concentrations were measured using a spectrophotometer (Multiskan MK3, Thermo scientific). Subsequently, the samples were admixed with 5 × sample buffer, equalized using double-distilled H₂O, and heated for 5 min at 100 ℃. An equal amount of protein from each sample was separated using sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) on a 10% gel and then electrophoretically transferred to a polyvinylidene fluoride (PVDF) membrane (Bio-Rad Laboratories Inc., Hercules, CA, USA). The blots were blocked with 10% skim milk in Tris-buffered saline and Tween 20 (0.1%) (TBST) for 2 h at room temperature (RT) and incubated at 4 ℃ overnight with mouse Nestin antibody (1:500, MAB353, Millipore), rabbit Sox2 antibody (1:1000, ab97959, Abcam), rabbit cd11b antibody (1:500, DF6476, Affinity), rabbit IL-6 antibody (1:500, DF6087, Affinity), rabbit IL-1β (1:500, AF5103, Affinity), rabbit TNF-α (1:500, AF7014, Affinity), rabbit VEGFR2 (1:500, AF6281, Affinity), rabbit p-VEGFR2 (1:500, AF8022, Affinity), or rabbit β-actin antibody (1:1000, AP0060, Bioworld Technology). After incubation with the primary antibody, the blots were incubated for 2 h at RT with secondary antibodies [goat anti-rabbit antibodies (1:5000, 111–035-003), and goat anti-mouse antibodies (115–035-003), Jackson]. Between steps, the blots were washed with TBST. The blots were visualized using ECL western blot detection system (ImageQuant LAS 4000 Mini). The bands were analyzed by Quantity One software version 4.6.2 (Bio-Rad Laboratories Inc.).

Six animals from each group were deeply anesthetized with 5% chloral hydrate and transcardially perfused with normal saline with simultaneous exsanguination from the right atrium, and then with 4% paraformaldehyde in 0.1 M phosphate buffer at a pH of 7.4, 5 days after anesthesia exposure (P12) and 24 h after the MWM (P28 and P56). The brains were removed and kept in 4% paraformaldehyde at 4 ℃ overnight. Subsequently, the brains were consecutively dipped into 70%, 85%, 95%, and 100% ethanol. Serial coronal 5 µm sections were cut in paraffin blocks using a microtome (Leica R2016), and at least three slides from each animal were used for staining. Briefly, sections were deparaffinized with xylene, rehydrated with a series of graded ethanol, and washed in distilled water and then in PBS. The antigen retrieval was performed incubating in 10 mM sodium citrate buffer at a pH of 6.0 for 20 min in a microwave oven at 100 ℃. The sections were blocked in 5% BSA in PBS for 30 min and were then incubated with primary antibodies (mouse anti-Nestin, 1:100, Millipore; rabbit anti-Sox2, 1:100, Millipore; rabbit anti-GFAP, 1:500, Dako; sheep anti-BrdU, 1:500, Abcam; rabbit anti-DCX, 1:400, Abcam; rabbit anti-NeuN, 1:200, Abcam; rabbit anti-IBA1, 1:400, Wako) overnight at 4 ℃. Subsequently, the sections were washed at RT, incubated in fluorophore-conjugated secondary antibodies (donkey anti-sheep, 1:500, ab150177, Abcam; donkey anti-mouse, 1:500, ab150112, Abcam; donkey anti-rabbit, 1:500, ab150068, Abcam; donkey anti-rabbit, 1:500, ab150073, Abcam) for 50 min, and then washed. The tissue sections were then incubated in DAPI (1:1000, Sigma) and then washed with PBS. Finally, the sections were mounted, viewed and quantitated using a microscope (Nikon E100) captured with 200 × magnification. The NSC proliferation (BrdU/Nestin, BrdU/Sox2), numbers (GFAP/Sox2) and differentiation (BrdU/DCX, BrdU/NeuN) of the NSC, microglia activation (IBA1⁺), and the number of neuronal nuclei in the DG (DAPI) were examined. Five random sections per brain were immunostained, and photomicrographs were captured at × 200 magnification. The average of positive cell numbers of five random sections represented one brain. The total number of cells in the hippocampus was assessed by counting the neuronal nuclei stained with DAPI. All counting by investigators was blinded. All quantifications were determined by calculating the percentage of IBA1⁺ or double-labeled positive cells to the total cells of cross-sectional hippocampus area in five random sections per rat using Image-Pro Plus software version 6.0. A ratio was calculated and results presented as levels of expression (positive/total cells). The methods were analyzed as previously described (Shen et al. 2013).

We analyzed the effects of isoflurane or sevoflurane on cardiorespiratory functions by exposing neonatal rats (n = 8 each per group) to 1.1% isoflurane (ISO) or 2% sevoflurane (SEV) for 4 h. However, we did not observe any significant differences in the respiratory rates between the control (CON), ISO and SEV groups of neonatal rats (data not shown). Arterial blood gas analysis demonstrated that partial pressure of oxygen (PaO₂), partial pressure of carbon dioxide (PaCO₂), and pH were all similar and within the normal range for the CON, ISO, and SEV groups of neonatal rats; moreover, mean arterial pressure (MAP) was similar for the CON, ISO, and SEV groups of neonatal rats (Table 1).

We then assessed the effects of isoflurane and sevoflurane on the general activity and behavior of rats (n = 12 each per group). The open-field test results at 2 weeks after exposure to the anesthetics showed that the total distance traveled within the test arena, time spent, and number of entries to the central region of the open-field arena were similar for the CON, ISO, and SEV groups of rats (Fig. 1). These data suggested that the inhaled anesthetics did not affect activity and behavior of rats.

We performed Morris water maze (MWM) tests to determine the effects of isoflurane and sevoflurane on the learning and memory of rats. The ISO and SEV group rats (n = 15 each per group) showed higher MWM escape latencies and distances compared to the CON group rats (n = 15) on days 1, 2, and 3 [Fig. 2A,B; MWM escape latency: p = 0.011 (ISO vs. CON); p = 0.028 (SEV vs. CON); MANOVA (F = 4.102, df = 2) for day 1; p = 0.015 (ISO vs. CON); p = 0.030 (SEV vs. CON); MANOVA (F = 3.882, df = 2) for day 2; p = 0.013 (ISO vs. CON); p = 0.038 (SEV vs. CON); MANOVA (F = 3.867, df = 2) for day 3; MWM escape distance: p = 0.000 (ISO vs. CON); p = 0.001 (SEV vs. CON); MANOVA (F = 11.339, df = 2) for day 1; p = 0.002 (ISO vs. CON); p = 0.053 (SEV vs. CON); MANOVA (F = 5.408, df = 2) for day 2; p = 0.002 (ISO vs. CON); p = 0.033 (SEV vs. CON); MANOVA (F = 5.581, df = 2) for day 3]. The ISO group (n = 15) showed significantly increased escape distance compared to the CON group (n = 15) on day 4 [Fig. 2A,B; p = 0.017; MANOVA (F = 3.110, df = 2)].

The MWM escape latencies of both ISO and SEV groups of rats (n = 15) were higher than the CON group (n = 15) on day 5, but the differences were significant only between the ISO and CON groups [Fig. 2A,B; p = 0.045; MANOVA (F = 2.658, df = 2)]. However, escape distances of both the ISO and SEV groups of rats were significantly higher than the CON group rats on day 5 [Fig. 2B; p = 0.013 (ISO vs. CON); p = 0.039 (SEV vs. CON); MANOVA (F = 3.829, df = 2)]. During the probe trial on day 6, time spent at the platform–site crossovers and percentage time in the target quadrant was lower for the rats from the ISO and SEV groups compared to the CON group, however, the differences were not statistically significant [Fig. 2c,d; one-way ANOVA (F = 1.627; df = 2), p = 0.209 for the percentage of time in the target quadrant; one-way ANOVA (F = 0.204; df = 2), p = 0.816 for platform–site crossovers]. The MWM test results6 weeks after exposure to the anesthetics did not show any significant differences between the CON, ISO, and SEV groups of rats (Fig. 3A–D).

These results suggested that cognitive function was impaired in the rats 2 weeks after exposure to isoflurane or sevoflurane. However, cognitive function was restored 6 weeks after exposure to the anesthetics.

We then analyzed the working memory of rats (n = 15 each per group) using the T maze test 2 weeks after exposure to the anesthetics. The percentages of correct alternations were significantly lower on days 1 and 8 for the ISO and SEV group rats compared to the CON group rats [Fig. 4; day 1: p = 0.015 (ISO vs. CON); p = 0.015 (SEV vs. CON); MANOVA (F = 4.266, df = 2); day 8: p = 0.001 (ISO vs. CON); p = 0.048 (SEV vs. CON); MANOVA (F = 6.671, df = 2)]. The percentage of correct alternations on days 7 and 10 was significantly lower for the ISO group compared to the CON group [Fig. 4; day 7: p = 0.005; MANOVA (F = 4.664, df = 2); day 10: p = 0.044; MANOVA (F = 2.147, df = 2)]. And the percentage of correct alternations on days 7 was significantly lower for the ISO group compared to the SEV group [Fig. 4; day 7: p = 0.043; MANOVA (F = 4.664, df = 2)]. Moreover, the percentage of correct alternations was lower for the SEV group of rats compared to the CON group rats on days 7 and 10, but the differences were statistically insignificant (Fig. 4). These results showed that the working memory of rats was impaired 2 weeks after isoflurane or sevoflurane exposure. However, the percentage of correct alternations 6 weeks after exposure to the anesthetics was similar for all three groups (Fig. 5). This suggested that working memory was restored 6 weeks after exposure to anesthetics.

The levels of IL-6, TNF-α, and CD11b (microglial marker) were significantly higher in the hippocampus of ISO (Fig. 6A,B; p = 0.000 for IL-6; p = 0.000 for TNF-α; p = 0.000 for CD11b) and SEV (Fig. 6A,B; p = 0.000 for IL-6; p = 0.000 for TNF-α; p = 0.000 for CD11b) in groups of rats. This suggested increased neuroinflammation due to microglial activation after exposure to anesthetics. Furthermore, in comparison with the CON group, the expression levels of Nestin, Sox2, and p-VEGFR2 as well as the ratio of p-VEGFR2/VEGFR2 were significantly reduced in the hippocampus of rats belonging to the ISO group (Fig. 6A–F; p = 0.000 for Nestin; p = 0.000 for Sox2; p = 0.000 for p-VEGFR2; p = 0.000 for p-VEGFR2/VEGFR2) and SEV group (Fig. 6A–F, p = 0.000 for Nestin; p = 0.000 for Sox2; p = 0.000 for p-VEGFR2; p = 0.000 for p-VEGFR2/VEGFR2). These results suggested reduced activity of VEGFR2 in the hippocampus of rats after exposure to the anesthetics. Furthermore, expression levels of Nestin and Sox2 mRNAs were significantly lower in the hippocampus of rats belonging to the ISO group compared to those belonging to the SEV group (Fig. 6A,B; p = 0.040 for Nestin; p = 0.010 for Sox2). This suggested inhibition of the downstream VEGF/VEGFR2 signaling pathway.

We then performed Immunohistochemical (IHC) analysis to assess the effects of isoflurane and sevoflurane on the proliferation and differentiation of NSCs. IHC results demonstrated that Nestin⁺/BrdU⁺ co-labeling was reduced in the ISO and SEV groups on day 5 after exposure to the anesthetics compared to the CON group, but these differences were not statistically significant. However, significant reduction in Sox2⁺/BrdU⁺ co-labeling was observed in the ISO and the SEV groups compared to the CON group [Fig. 7A,B; p = 0.001 (ISO vs. CON) and p = 0.004 (SEV vs. CON)]. This suggested that the anesthetics decreased the proliferation of NSCs. Furthermore, we observed decreased Sox2⁺/GFAP⁺ co-labeling in the ISO and SEV groups compared to the CON group [Fig. 7A,B; p = 0.001 (ISO vs. CON); p = 0.013 (SEV vs. CON)]. This showed that the anesthetics decreased the number of NSCs in the hippocampus of rats. We also observed reduced DCX⁺/BrdU⁺, CR⁺/BrdU⁺, and NeuN⁺/BrdU⁺ co-labeling in the ISO (Fig. 7A,B; p = 0.000 for DCX⁺/BrdU⁺; p = 0.000 for CR⁺/BrdU⁺; p = 0.000 for NeuN⁺/BrdU⁺) and SEV (Fig. 7A,B; p = 0.000 for DCX⁺/BrdU⁺; p = 0.023 for CR⁺/BrdU⁺; p = 0.002 for NeuN⁺/BrdU⁺) groups compared to the CON group. We also observed decreased CR⁺/BrdU⁺ co-labeling in the ISO group compared to the SEV group (Fig. 7A,B; p = 0.032). These results showed that the anesthetics inhibited differentiation of NSCs. We also observed significantly higher numbers of IBA-1⁺cells in the ISO and SEV groups compared to the CON group [Fig. 7A,B; p = 0.010 (ISO vs. CON); p = 0.013 (SEV vs. CON)]. IBA-1 is a marker of microglial activation (Norden et al. 2016). This showed that the anesthetics activated the microglia in the hippocampus of rats.

At P28, 24 h after the MWM, we observed reduced Nestin⁺/BrdU⁺, Sox2⁺/BrdU⁺, Sox2⁺/GFAP⁺, DCX⁺/BrdU⁺, CR⁺/BrdU⁺ and NeuN⁺/BrdU⁺ co-labeling in the ISO group (Fig. 8A,B, p = 0.006 for Nestin⁺/BrdU⁺; p = 0.000 for Sox2⁺/BrdU⁺; p = 0.000 for Sox2⁺/GFAP⁺; p = 0.000 for DCX⁺/BrdU⁺; p = 0.000 for CR⁺/BrdU⁺; p = 0.000 for NeuN⁺/BrdU⁺) and SEV group (Fig. 8A,B; p = 0.006 for Nestin⁺/BrdU⁺; p = 0.001 for Sox2⁺/BrdU⁺; p = 0.000 for Sox2⁺/GFAP⁺; p = 0.000 for DCX⁺/BrdU⁺; p = 0.016 for CR⁺/BrdU⁺; p = 0.000 for NeuN⁺/BrdU⁺) compared to the CON group. Furthermore, the number of IBA-1⁺ cells were significantly higher in the ISO group (Fig. 8A,B; p = 0.000) and SEV group (Fig. 8A,B; p = 0.005) compared to the CON group. Moreover, Sox2⁺/GFAP⁺ co-labeling was significantly lower in the ISO group compared to the SEV group (Fig. 8A,B, p = 0.000). The number of IBA-1⁺ cells were significantly higher in the ISO group compared to the SEV group (Fig. 8A,B; p = 0.000).. However, at P56, 24 h after the second MWM, we did not observe any differences in the co-labeling markers for all three groups (Fig. 9A,B).

Our study showed that exposure to 1.1% isoflurane or 2% sevoflurane for 4 h impaired spatial memory in neonatal rats. The MWM escape latency was significantly longer in the ISO and SEV group rats compared to those in the CON group. Moreover, rats in the ISO group showed more persistent and prolonged escape latency than those in the SEV group. T maze task results showed that the percentage of correct selections (alternations) was significantly lower in the ISO and SEV group rats compared to the CON group rats on days 1 and 8. On days 7 and 10, the percentage of correct selections was significantly lower in the ISO group rats compared to the CON group rats, whereas, the differences between the SEV and CON group rats were not statistically significant. Overall, the results of the T maze tests showed that isoflurane and sevoflurane impaired working memory of the rats at 2 weeks. Furthermore, isoflurane induced more severe and prolonged neurotoxicity compared to sevoflurane at 2 weeks. The control rats from day 2 till 5 showed decreased percentage of correct alteration which is surprised but similar to that in a previous study (Mogensen et al. 2008; Schaefers and Winter 2011). The possible explanation could be that there were indications for behavioral asymmetries, since most rats showed a rightward direction of turning within the start arm (Castellano et al. 1987; LaHoste et al. 1988). Such biases may affect the outcome of experimental manipulations, be they behavioral or physiological (Andrade et al. 2001). It has been shown that the likelihood of asymmetry and its degree can be affected by time of testing (Schwarting and Borta 2005). Therefore, the performance could improve with time as seen after day 5.

The results of MWM and T maze tests of rats at 6 weeks after exposure to anesthetics was similar to the control group rats. This finding was consistent with previous reports (Stratmann et al. 2009b). This showed that cognitive function was completely restored in rats at 6 weeks after exposure to isoflurane or sevoflurane. A previous study reported that exposure of P7 rats to 1 MAC isoflurane for 4 h caused persistent deficits in spatial reference memory (Stratmann et al. 2009a, b). Another study reported that exposure to 1.5% isoflurane for 4 h caused cognitive impairment in P7 rats (Schaefer et al. 2019). Furthermore, cognitive function of P6 mice was significantly impaired after daily exposure to 3% isoflurane for 2 h over 3 consecutive days (Shen et al. 2013; Yu et al. 2020). Our results were consistent with the results from these studies. However, our results were contrary to findings from another study, which reported that cognitive impairment was induced at 6 weeks after isoflurane or sevoflurane exposure in neonatal animals (Zhu et al. 2010). We speculate that these differences might be due to variations in the age of rats, protocol for exposure to anesthetics, concentration of anesthetics, exposure time, or use of different animal strains.

Isoflurane anesthesia in rats altered postnatal hippocampal neurogenesis in an age-dependent manner (Erasso et al. 2013). The dentate gyrus restored normal numbers of GFP⁺-expressing granule cells in the Gli1-CreER::GFP bitransgenic mice on day 60 after a single, developmental exposure to 1.5% isoflurane for 6 h, thereby suggesting increased proliferation of GFP⁺ granule cells (Jiang et al. 2016). The proliferation of neuronal progenitor cells was significantly reduced for 5 days after administration of P7 rats with one minimum alveolar concentration (MAC) isoflurane for 4 h (Stratmann et al. 2009b). P14 mice and rats anesthetized with 1.7% isoflurane for 35 min daily over 4 successive days showed reduced hippocampal stem cell pool including radial glia-like stem cells and neurogenesis until 4 weeks after anesthesia (Zhu et al. 2010). Moreover, maternal exposure to 3.5% sevoflurane during the mid-gestational period inhibited fetal NSC proliferation via the Wnt/β-catenin pathway and impaired postnatal learning and memory function in rats (Wang et al. 2018). Exposure of P7 rats to 2.5% sevoflurane for 9 h inhibited proliferation of neural progenitor cells until 2 weeks after anesthesia (Liu et al. 2014).

It has been shown that drugs that act by enhance GABA_A receptors (isoflurane or sevoflurane) or block NMDA receptors (ketamine or nitrous oxide) induce neurotoxicity in immature rodent brain when administered during synaptogenesis. Neonatal exposure to ketamine in rats inhibits the proliferation and astrocytic differentiation of NSCs in the hippocampal dentate gyrus and impairs neurocognitive function in adulthood (Huang et al. 2021). Repeated neonatal ketamine exposure inhibits the proliferation and astrocytic differentiation of NSCs in the SVZ and induces olfactory cognitive dysfunction in the adulthood (Sha et al. 2021). Furthermore, ketamine exposure inhibits proliferation and neuronal differentiation of NSCs in the SVZ and SGZ of the hippocampus and leads to adult cognitive deficits (Li et al. 2019). Nitrous oxide is similar in nature to ketamine. Long-term impairments of neuronal development and synaptic communication could be caused when P7 rats were exposed to a sedative dose of midazolam followed by combined nitrous oxide and isoflurane anesthesia for 6 h (Dalla Massara et al. 2016). Recent research has raised concerns about possible neurotoxicity of nitrous oxide, particularly in the developing brain (Savage and Ma 2014; Shu et al. 2010). Neonatal exposure to nitrous oxide (75% N₂O and 25% O₂) in mice inhibits cell proliferation in developing brain (Rodier et al. 1986). Neurogenesis can be down-regulated after exposure to nitrous oxide (Covacu et al. 2006). However, the neurotoxic effects of nitrous oxide are not observed for the P5-6 rhesus monkeys (Zou et al. 2011).

Our results showed that isoflurane and sevoflurane inhibited the proliferation and differentiation of NSCs at P12 or P28. Moreover, the effects of isoflurane-induced neurodevelopmental toxicity were stronger than those of sevoflurane. These results were consistent with previous reports. Moreover, proliferation and differentiation of the NSCs in the isoflurane and sevoflurane groups were similar to the control group when analyzed at P56. Our study showed that the neurodevelopmental toxicity observed after exposure of P7 rats to the anesthetics for 4 h persisted for at least 3 weeks after anesthesia.

Hippocampal neurogenesis was inhibited and the numbers of microglia in the dentate gyrus were significantly increased after peripheral LPS injections in rodents (Belarbi et al. 2012; Dinel et al. 2014; Valero et al. 2014). However, LPS-induced inhibition of neurogenesis was abrogated by anti-inflammatory drugs (Monje et al. 2003). Systemic administration of minocycline in adult rats restored hippocampal neurogenesis, which was significantly reduced after intracortical injection of LPS for 28 days; moreover, a negative correlation was observed between the number of new neurons and the number of activated microglia (Ekdahl et al. 2003). Microglial activation was observed in multiple brain regions of male newborn piglets that were exposed to 2.0% isoflurane for 6 h (Broad et al. 2016). These findings suggested that anesthesia-induced neurodevelopmental toxicity was related to the release of inflammatory cytokines by activated microglia.

TNF-α secretion from the microglia aggravates neurotoxicity of hippocampal NSCs in co-culture experiments (Cacci et al. 2005; Iosif et al. 2008). Moreover, cytokines and chemokines released during neuroinflammation inhibited neurogenesis (Monje et al. 2003). However, microglia also release factors that promote proliferation and maintenance of the NSCs (Deierborg et al. 2010). IL-6 levels are increased when primary murine microglia are exposed to 2% isoflurane or 4.1% sevoflurane for 6 h (Wang et al. 2014; Zhang et al. 2013). Furthermore, pretreatment of primary microglia with 2% or 4% sevoflurane inhibited IL-4-induced M2 microglial activation and decreased the levels of IL-10, Arg1, and Ym1 (Pei et al. 2017). Previous studies also showed that inhaled anesthetics increase the levels of inflammatory cytokines, such as TNF-α, IL-6, and IL-1β in the brains of experimental mice (Shen et al. 2013) and induce neurotoxicity (Wu et al. 2012). However, sevoflurane post-conditioning can also exert anti-inflammatory and neuroprotective effects. For example, sevoflurane post-conditioning decreased infarct size, improved neurological deficit score, and reduced the levels of pro-inflammatory cytokines, such as TNF-α and IL-6 (Ye et al. 2015).

Our results showed increased number of IBA-1⁺ cells in the brain after isoflurane and sevoflurane exposure. Moreover, the levels of CD11b, IL-6, and TNF-α were increased in the hippocampus of rats exposed to isoflurane and sevoflurane. This demonstrated that isoflurane and sevoflurane induced neurodevelopmental toxicity by activating microglia and up-regulating the release of inflammatory cytokines.

VEGFR2 signaling pathway promotes proliferation, differentiation, and migration of neural stem cells (Jin et al. 2002; Mani et al. 2010). Our results showed that VEGFR2 activity was significantly decreased (reduced levels of p-VEGFR2 and low ratio of p-VEGFR2/VEGFR2) in the hippocampus of rats exposed to isoflurane and sevoflurane. This suggested that anesthesia-induced neurodevelopmental toxicity was also associated with reduced VEGFR2 activity in addition to increased microglial activation.

There is a limitation of our study. Isoflurane or sevoflurane induced “brain damage” through the interaction between microglia and NSCs and the level of VEGFR2 phosphorylation may be just an association rather the “real” mechanism. Therefore, pharmacological intervention (such as PLX3397 for microglia elimination) or conditional knockout rodents should be considered in future study to verify the mechanism of anesthetics-induced neurodevelopment toxicity.