Sevoflurane Postconditioning Ameliorates Neuronal Migration Disorder Through Reelin/Dab1 and Improves Long-term Cognition in Neonatal Rats After Hypoxic-Ischemic Injury

Sevoflurane postconditioning (SPC) has been widely reported to attenuate brain injury after hypoxia–ischemia encephalopathy (HIE) by inhibiting neural necrosis and autophagy. Moreover, recent reports revealed that sevoflurane facilitated hippocampal reconstruction via regulating migration. Yet, it remains unclear whether the promotion of neural migration by SPC repairs the hippocampal injury after HIE. Here, we hypothesize that SPC exerts a neuroprotective effect by ameliorating neuronal migration disorder after HIE and regulating Reelin expression. Furthermore, the downstream Reelin/Dab1 pathway may be involved. The classical Rice–Vannucci model of hypoxia–ischemia was performed on postnatal day 7 rat pups, which was followed by SPC at 1 minimum alveolar concentration (MAC 2.5%) for 30 min. Piceatannol, causing Reelin aggregation in vivo, was used to detect whether Reelin/Dab1 was involved in the neuroprotection effect of SPC. Hippocampal-dependent learning ability tests were conducted to assess the long-term effects on locomotor activity and spatial learning ability. Our findings suggest that hypoxia–ischemia injury inhibited neurons migrated outward from the basal zone of dentate gyrus, disrupted cytoarchitecture of the dentate gyrus (DG), and led to long-term cognition deficits. However, SPC could relieve the restricted hippocampal neurons and repair the hippocampal-dependent memory function damaged after HIE by attenuating the overactivation of the Reelin/Dab1 pathway. These results demonstrate that SPC plays a pivotal role in ameliorating neuronal migration disorder and maintaining normal cytoarchitecture of the DG via inhibiting overactivated Reelin expression. This process may involve overactivated Reelin/Dab1 signaling pathway and spatial learning ability by regulating the Reelin expression which may associate with its neuroprotection.


Introduction
Neonatal hypoxic-ischemic encephalopathy (HIE) is one of the prevalent causes of severe neurological morbidity in neonates, which may lead to devasting neurological, cognitive, and behavioral disorders. The incidence of HIE is approximately 1/1000 in live births, and the survivors may still suffer from severe sequelae 1,2 . To date, in addition to hypothermia, very few clinically promising therapies have been reported. However, the long-term efficacy of these therapies in terms of alleviating brain damage after HIE is still very low.
The development of hippocampal neurons is a sensitive, multistep process involving progenitor proliferation from the neurogenesis niche dentate gyrus (DG ) zone at a low rate, migration to the final location, and differentiation and synaptogenesis until integration with the existing neuronal circuit 3 4 . However, although the central nervous system initiates neural self-repairment after injury, including reactive neurogenesis, it is still unable to repair severe injuries 5 . Previous evidence has also confirmed that precise neuronal migration is vital for the establishment of hippocampal functional neuronal circuits. There are more than 9,000 immature neurons being sent daily from the subgranular zone (SGZ) of DG, and this process continues throughout the life, with the migration of neurons from the subgranular layer outward to the granular and multiple molecular layers. Disruption of this process can lead to neurological disorders such as schizophrenia 6 . Therefore, a medicine that can promote effective neural migration from DG after hypoxic-ischemic injury may be a promising therapy.
Sevoflurane is a widely used inhalational anesthetic in pediatric clinics. Mounting evidence support the idea that sevoflurane postconditioning (SPC) may alleviate neurological deficits by inhibiting neural apoptosis, suppressing overactive autophagy process, and inhibiting inflammation 6,7 . Recent evidence proved that sevoflurane promoted the migration of microglia and astrocytes toward ischemic foci to repair the injury after transient ischemic injury 8 ; this indicated that sevoflurane may exert its neuroprotective effect by aligning hippocampal neural cell orientation 9,10 . Yet, it remains unclear whether SPC affects the hippocampal DG reconstruction process after HIE and the 4 underlying mechanism of its action.
Reelin is a large extracellular molecular protein secreted by Cajal-Retzius cells, which plays a key role in the correct positioning of cortical and hippocampal neurons 11 . It consists of eight Reelin repeats, and its N-terminal region is required for signal transduction 12,13 . The research on Reelin mutant mice, called "reeler" phenotype, has contributed to our knowledge of the Reelin effect on orchestrating the arrangement of neurons in the cortical layer, hippocampal layer, and other regions such as cerebellum 14 .
Studies on canonical and noncanonical Reelin signaling pathways have shown that Reelin can influence the neuron migration process through multiple ways, including binding to the apolipoprotein E receptor 2 or the very low-density lipoprotein receptor, which subsequently activates intracellular Dab1 or N-cadherin to participate in neuron migration, synaptic plasticity modulation, and neurotransmitter release process 15  Neonatal HIE model and drug administration Neonatal HIE model protocol was based on a previously described protocol 18 . Briefly, sevoflurane-anesthetized pups (12-20 g body weight) were operated for permanent ligation of the left common carotid artery. After recovering for 2 h, the pups were ventilated with 30% O2 and 70% N at the flow rate of 2 L/min for 2 h in the Sham group and 8% O2 and 92% N in the HIE groups. SPC at 1 minimum alveolar concentration (MAC 2.4%) was conducted immediately for 30 min after asphyxia by using a gas monitoring device (Datex-Ohmeda Inc., Tewksbury, MA, USA).  Pulse labeling study by bromodeoxyuridine injection 5-Bromo-2-deoxyuridine (BrdU) pulse birth-labeling method for detecting neuron migration has been considered as a potential and accepted method 16,19,20 . To determine the migration of newborn neurons, BrdU (B5002, Sigma-Aldrich) was intraperitoneally injected into dams at the dose of 150 mg/kg per day for 3 days consecutively from PND6 to PND9.

Drug administration in vivo
The pups were sacrificed, and their brains were harvested on PND14 and PND21.
Tissue processing and immunofluorescence staining For the immunofluorescence assay, the pups were deeply anesthetized with pentobarbital (80 mg/kg), and their brains were harvested. Coronal sections of paraffin-fixed brain were cut consecutively at the thickness of 3-3.5 µm. Similar to that for BrdU staining, the sections were treated with 2N HCl at 37 °C for 30 min for denaturing DNA, followed by neutralization with 0.1M borate buffer (pH 8.5). The sections were then blocked with 5% fetal bovine serum for 30 min and then incubated overnight with primary antibodies. On the next day, the slices were incubated with one or two types of secondary antibodies for

Western blotting analysis
Hippocampal DG tissue was carefully dissected from the brains of the pups by using an anatomical microscope (Nikon SMZ445) at PND8 (n = 5 per group). Frozen hippocampi DG was cut and then lysed on ice for 30 min. The lysate was centrifuged, and the total protein concentration was measured using the BCA Protein Assay Kit (P0010; Beyotime). The proteins were separated by electrophoresis on 8% or 10% SDS-PAGE gel and then electrotransferred to polyvinylidene fluoride membranes (IPVH0010; Millipore, Germany).
The target proteins were incubated with the specific primary and second antibodies. The

Ethological Tests
Three behavioral tests were conducted on days PND28-P34 as previously described.

Open field test
To evaluate anxiety and locomotor behavior, the rats were placed in a 100 cm (W) × 100 cm (D) × 40 cm (H) arena with opaque sidewalls equipped with infrared detectors for 10 min (Noldus Ethovision XT, Netherland), and the movement distance, time spent in the center and along the wall, movement velocity, and fecal pellets were recorded for anxiety mood evaluation.

Morris water maze test
Spatial learning and memory ability of rats were tested by the Morris water maze (MWM) test from PND29 to PND34. Briefly, during the first consecutive 5 days, the rats were placed into a circular, opaque pool (1.6 m diameter, 60 cm height) for searching an invisible platform within 90 s. Once failed, they were forced to stay on it for another 20 s.
During the spatial probe test, the rats were released from the opposite quadrant with the platform removed away and permitted to swim for 90 s. The paths of each animal were recorded by infrared detectors, and the data were analyzed by an image analysis software (Noldus Ethovision XT)

8-Arm radial maze test
The 8-arm radial maze (RAM) test was conducted as described previously 21 . During the acclimatization phase, the rats were allowed to explore freely for 10 min with all arms baited after controlling their weight until 85-90% of free-feeding body weight was achieved. During the next 5 days of the training phase, four of the eight arms were baited randomly, and the rats were trained twice a day to find the food reward with an interval of 9 1 h between the trials. Each trial was terminated within 10 min or before 10 min if the rats found all food rewards. During the choice phase, all arms were unbaited, and the rats were allowed to find the food reward within 10 min. The total travelled distance, reference memory error (RME), and working memory error (WME) were recorded and analyzed using an image analysis software (Noldus Ethovision XT).

Statistical analysis
Data are presented as mean ± standard deviation (SD). All data were analyzed with SPSS 20.0 or GraphPad Prism 7.0. The data were analyzed using Bartlett's test for equal variances and Shapiro-Wilk test for normality. One-way analysis of variance (ANOVA) followed by Newman-Keuls post hoc test was used. Data of escape latency in MWM and the ratio of neuronal migration were analyzed using two-way ANOVA for repeated measurements. Differences with p < 0.05 were considered to be statistically significant.

Results
Neuron migration process was perturbed and DG neuron distribution was disordered after HIE, and SPC alleviated these injuries To better visualize neuron migration from the SGZ to the superficial GCL, the BrdU pulse labeling protocol was used (Supplementary Figure S1). Initially, the migration of the pulselabeled cells to the destination was detected and the ratio of BrdU/neuron cells to the total BrdU-positive cells in each layer was calculated (Fig. 1A, PND14; 1B, PND21). The ratio of BrdU+/neuron + cells in the SGZ layer to BrdU + cells in the whole layer was enhanced in the HI group as compared to that in the Sham group on PND14 and PND21.
This change in the ratio indicated that a large number of labeled neurons were restricted to the SGZ rather than migrating outward (Fig. 1C, D, HI versus Sham, p < 0.05 and p < 0.05 on PND14 and PND21, respectively). However, the ratio of BrdU+/neuron + cells in the GCL to the total BrdU + cells in the whole layer was decreased as compared to that in the Sham group (Fig. 1C,D, HI versus Sham, p < 0.01 and p < 0.01 on PND14 and PND21, respectively). To exclude the interference of reactive neurogenesis in BrdU + cells after HIE, the total BrdU + cells were counted, and the data showed no significant difference among all groups ( Supplementary Fig. 2). SPC decreased the elevated number of neuronal cells restricted in the SGZ zone and increased the ratio in the GCL (Fig. 1C, D, HI + Sev versus HI, p < 0.05, p < 0.05; Fig. 1C, D, HI + Sev versus HI, p < 0.05, p < 0.01 on PND14 and PND21, respectively).
Nissl staining was conducted on PND14 to study the distribution of hippocampal DG neurons (Fig. 1E). As compared to the Sham group, the HI group showed decreased neuron ratio with rare cytoplasm and few Nissl bodies and disorganized neurons without polarity in DG (Fig. 1F, HI versus Sham, p < 0.001). SPC alleviated neuron loss and promoted reorganization of the neurons with proper orientation (Fig. 1F, HI + Sev versus HI, p < 0.01).

Reelin expression was enhanced after HIE
Here, we investigate the link between Reelin expression and SPC protective effect. Our data showed that rats in the HI group persistently expressed more Reelin-positive cells than those in the Sham group on PND8, PND14, and PND21 ( Fig. 2A-F  was conducted. The co-expression of Reelin and Dab1 was significantly enhanced after HIE, and SPC significantly inhibited the elevated expression ( Fig. 3A-F, HI versus Sham, p < 0.01, p < 0.001, and p < 0.001; HI + Sev versus HI, p < 0.01, p < 0.05, and p < 0.01 on PND8, PND14, and PND21, respectively). The beneficial effects of SPC were blocked by Piceatannol (Fig. 3A-F HI + Sev + G versus HI + Sev, p < 0.01, p < 0.01, p < 0.05 on PND8, PND14, and PND21, respectively) Consistent with the above findings, there were more BrdU/Neun-labeled cells in the SGZ and GCL zones of rats of the HI group than those in rats of the Sham group (Fig. 4C, D, HI versus Sham, p < 0.01, p < 0.05 in SGZ; p < 0.05, p < 0.01 in GCL on PND14 and PND21, respectively). This finding indicated that a large amount of neurons were restricted to the basal zone of DG, and this high ratio was decreased by SPC (Fig. 4C, D, HI versus Sham, p < 0.05, p < 0.05 in SGZ; p < 0.01, p < 0.01 in GCL on PND14 and PND21, respectively).
Nissl staining was performed on PND14 to study the role of Reelin in the distribution of hippocampal DG neurons (Fig. 5A). The finding was consistent with the previous result that the HI group showed decreased neuron ratio with rare cytoplasm and few Nissl bodies, as well as disorganized neurons without polarity in DG as compared to that in the Sham group (Fig. 5B, HI versus Sham, p < 0.0001). SPC alleviated neuron loss and promoted the reorganization of neurons with proper orientation (Fig. 5B, HI + Sev versus HI, p < 0.01). The beneficial effect of SPC was, however, blocked by Piceatannol (Fig. 5B, HI + Sev + G versus HI + Sev, p < 0.05)

SPC promotes hippocampal spatial learning and memory in HIE rats
To exclude anxiety and locomotor ability interference produced by several treatments, we conducted an open field test experiment on PND28. However, there were no difference in the test results among the four groups (Supplementary Figure S3).
In the MWM test, no significant difference in swimming velocities was observed among the four groups (Fig. 7A, p > 0.05 among all groups). Rats in the HI group showed significantly increased escape latency as compared to rats in the Sham group (Fig. 7B, HI versus Sham, p < 0.001, p < 0.001, p < 0.001, and p < 0.01 on the 2nd, 3rd, 4th, and 5th day, respectively). SPC attenuated this phenomenon (Fig. 7B, HI + Sev versus HI, p < 0.01, p < 0.001, and p < 0.01 on the 3rd, 4th, and 5th day, respectively). Rats in the HI + Sev + G group treated with Piceatannol needed more time to find the platform than rats who were administered SPC alone (Fig. 7B, HI + Sev + G versus HI + Sev, p < 0.01 and p < 0.05 on the 4th and 5th day, respectively). In the spatial probe test, the HI group rats showed 13 crossed the platform less number of times than the Sham group rats (Fig. 7C, HI versus Sham, p < 0.01). Rats in the HI + Sev group crossed the platform a larger number of times than rats in the HI group (Fig. 7C, HI + Sev versus HI, p < 0.05). Piceatannol notably blocked the protective effect of SPC on spatial learning ability (Fig. 7C, HI + Sev + G versus HI + Sev, p < 0.05).

SPC facilitates DG-dependent spatial learning and memory in HIE rats
The 8-arm radial maze test was conducted in rats of the four groups from PND28 to PND34, and the path parameters were presented (Fig. 8A). The distance traveled during the reward seeking process was increased in the HI group, but significantly decreased in the HI + Sev group. Rats in the HI + Sev + G group traveled more distance to find the rewards than rats in the HI + Sev group (Fig. 8B, HI versus Sham, p < 0.001, p < 0.0001; HI + Sev versus HI, p < 0.05, p < 0.05; HI + Sev + G versus HI + Sev, p < 0.05, p < 0.05).
SPC abolished the increased RME of the HI group as compared to that of the Sham group ( Fig. 8C, HI versus Sham, p < 0.001; HI + Sev versus HI, p < 0.05), and Piceatannol blocked this beneficial effect (Fig. 8C, HI + Sev + G versus HI + Sev, p < 0.05). Interestingly, WME, which represents short-term memory, showed no difference among the four groups ( Fig. 8D, p > 0.05 compared with all groups). This indicated that the specific hippocampal region that controlled short-term memory production might not be influenced by HIE or SPC.

Discussion
Our study showed that HIE perturbed the neuron migration process and disordered DG neurons, leading to long-lasting neurological deficits. SPC can abolish these impairments and finally facilitate hippocampal-dependent long-term spatial learning and cognition. The relative mechanism of SPC action may involve regulation of Reelin expression and suppression of the Reelin/Dab1/GSK-3β/Tau signaling pathway to modulate the dynamics 14 of cytoskeleton assembly in DG neurons.
HIE is a devastating consequence of perinatal asphyxia that leads to severe brain structure aberration and long-term neurocognitive sequela such as mental retardation 22− 25 . The lack of an effective therapy for brain injury has therefore aroused intense interest in investigating appropriate drugs for administration. To our knowledge, sevoflurane is a popular general inhaled anesthetic that is particularly used in pediatric operation. In addition to its traditionally accepted protective effect by decreasing neuronal death and promoting long-term memory consolidation 26 , an appropriate dose of SPC may facilitate endogenous hippocampal neural network reconstruction through neurogenesis and promotion of migration of astrocytes and microglia to the infarcted region for neurocognitive repairment 27,28 . Moreover, Chai reported that multiple use and overdose of sevoflurane in the pregnant period may directly inhibit neural migration in the cortex of offspring 29 . In our present study, we first investigated DG neuron migration and neuron organization after HIE. Consistent with the findings of previous studies, we found that the normal DG neuron migration process was significantly perturbed and neural orientation was disordered after HIE. SPC administration functionally ameliorated these impairments.
DG of the hippocampus is special trilaminar, C-shaped structure where neurogenesis occurs postnatally and has a pivotal role in learning and memory 30 . Progenitors are generated in DG and migrate short distance from the inner SGZ to the superficial molecular layer; this process is mediated by other intercellular factors. Reelin, for example, integrates with the pre-existing hippocampal circuit and forms functional synaptic attachments almost in 1 week 31,32 . Here, we chose 7-day-old rats as experimental candidates. This programmed process is, however, vulnerbable to hypoxia and changes in the extracelluar matrix, which result in hippocampal neuron migration deficits, leading to aberrations in the orchestration and dysfunction of the hippocampus 24 . The indicated mechanisms that modulate the migration of newborn neurons to their destination are more clear in human health studies, for example, disruptions of neuronal migration for cortex development may induce lissencephaly and seizure 33 . Consistent with other studies, our study data showed that Reelin expression was significantly enhanced after HIE. SPC normalized Reelin expression, which indicates that its neuroprotective effect on reconstructing DG structure may involve modulation of Reelin expression.
Reelin is a 388-kDa extracelluar matrix glycoprotein with several fragments and plays important roles in cortical and hippocampal lamination, synaptogenesis, neurite formation, and neurotransmitter release 34,35 . Furthermore, accumulating evidence showed that disrupted Reelin expression may be a potential reason for various neurodegenerative disorders such as seizure and mental retardation 36,37 . When the canonical Reelindependent Dab1 is phosphorylated, it activates downstream signaling cascades and acts as a stop signal in neuron migration 38 . A previous study also revealed that Reelin can stabilize actin cytoskeleton and microtubule cytoskeleton by binding to ApoER2 or VLDLR receptors to promote nuclear translocation acting on actin and microtubule cytoskeleton 39 . Evidence based on prenatal hypoxia-induced impaired neural migration emphasized the relationship between Reelin-associated migration deficits and hypoxia stimuli. 40 In addition, the reduced expression of Reelin during the prenatal period might lead to hippocampal-dependent cognitive impairments in adults 41 . It is reasonable to hypothesize that sevoflurane could ameliorate the effects of HIE by regulating DG neuron migration and neurological behaviors. Our study data revealed that SPC may facilitate the perturbed migration process of DG neurons in HIE by markedly disengaging perturbed prelabeled BrdU+/Neun + cells from the SGZ zone and promoting their percentage in the GCL from PND14 to PND21. Piceatannol, which directly increases Reelin level, blocked SPC function.
This phenomenon implied the effect of Reelin on regulating DG lamination and Reelin/Dab1 interaction played the vital role in SPC-induced alleviation of DG developmental deficits after HIE.
The role of canonical Reelin-mediated signaling pathway in stabilizing the cytoskeleton by binding with receptors to activate Dab1-associated cascade has been highly debated 42 .
This cytoskeleton rearrangement ability of modifying Tau and MAP1B might involve proline-directed kinase GSk3 and CDK5 43 . A study on the relationship between Reelinmediated neuron migration process and microtubules showed that Reelin dynamically promoted the polymerization of microtubules, leading to the process of perturbed nuclear translocation 44 . In the present study, HIE obviously enrich Reelin expression and activated phopho-Dab1 at Tyr198, phosphor-GSK-3β at Ser 9 and phosphorylated Tau at Ser396 level. SPC can suppress the activated Reelin/Dab1/GSK-3β/Tau signaling pathway.
Notably, an injection of Piceatannol abolished this phenomenon, suggesting that Reelin/Dab1 pathway which influence microtubules polymerization did involve in the protection of SPC after HIE. This indicates that HIE may cause the spread of DG neurons without polarity, and this malformation may be associated with abnormal cleavage by Reelin and overactivation of the Reelin-Dab1 cascade. This damage may be reduced to some extent by SPC.
To our knowledge, endogenous hippocampal events, including neurogenesis and neuron migration contribute to hippocampal memory and learning 45− 47 . To verify whether SPC promotes long-term hippocampal memory after HIE, the MWM test was conducted. The results demonstrated that in the absence of motor disability and anxiety motion, hippocampal spatial and learning memory were extremely damaged by HIE. Gilber t48 showed that colchicine-induced DG lesion damaged DG-dependent learning ability. We conducted the 8-arm radial maze test to investigate DG-dependent pattern separation by referring to the capacity of transforming similar memories or events into nonoverlapping representations 28 . Our study data showed that RME, which represents long-term memory, was increased in the HI group. The phenomenon that rats in the HI group entered the working arm of the 8-arm maze without food rewards after training was abolished significantly by SPC. Piceatannol injection showed a similar harmful effect as that observed for the HI group. Interestingly, WME, which represents short-term memory retrieval, showed no difference among the four groups. This finding implied that HIE may particularly damage DG-dependent long-term memory storage instead of short-term memory, which suggests that HIE-triggered neuron migration disorder influenced the integration of short-term memory into long-term memory instead of inhibiting its production. These results are consistent with a previous research on traumatic brain injury 49 .
Our studied have several limitations. First, we mostly focused on clarifying the neuron migration disorder resulting from HIE and the neuroprotective effect of SPC on modulating this impairment, instead of neurogenesis and relative molecular mechanisms. Second, we used BrdU pulse-labeled technique, which may add difficulties in absolutely qualifying and tracing single neuron migration destination. We are currently working on our transgenic mice and a new method and may present the dynamics of the neuron migration pathway in a better way in further investigations.    Nissl staining was performed on PND14. Representative photographs of Nissl stained-Dentate gyrus(E). Neuronal density relative ratio(F) Values are mean ± SD(n=5) One-way ANOVA with Tukey post hoc multiple comparison tests was used for data analysis. Scale bar = 50 μm.****p <0.0001 vs Sham group, ^^p<0.01vs HI group, &p<0.05vs HI+Sev group.