Maternal sevoflurane exposure induces temporary defects in interkinetic nuclear migration of radial glial progenitors in the fetal cerebral cortex through the Notch signalling pathway

Abstract Objectives The effects of general anaesthetics on fetal brain development remain elusive. Radial glial progenitors (RGPs) generate the majority of neurons in developing brains. Here, we evaluated the acute alterations in RGPs after maternal sevoflurane exposure. Methods Pregnant mice were exposed to 2.5% sevoflurane for 6 hours on gestational day 14.5. Interkinetic nuclear migration (INM) of RGPs in the ventricular zone (VZ) of the fetal brain was evaluated by thymidine analogues labelling. Cell fate of RGP progeny was determined by immunostaining using various neural markers. The Morris water maze (MWM) was used to assess the neurocognitive behaviours of the offspring. RNA sequencing (RNA‐Seq) was performed for the potential mechanism, and the potential mechanism validated by quantitative real‐time PCR (qPCR), Western blot and rescue experiments. Furthermore, INM was examined in human embryonic stem cell (hESC)‐derived 3D cerebral organoids. Results Maternal sevoflurane exposure induced temporary abnormities in INM, and disturbed the cell cycle progression of RGPs in both rodents and cerebral organoids without cell fate alternation. RNA‐Seq analysis, qPCR and Western blot showed that the Notch signalling pathway was a potential downstream target. Reactivation of Notch by Jag1 and NICD overexpression rescued the defects in INM. Young adult offspring showed no obvious cognitive impairments in MWM. Conclusions Maternal sevoflurane exposure during neurogenic period temporarily induced abnormal INM of RGPs by targeting the Notch signalling pathway without inducing long‐term effects on RGP progeny cell fate or offspring cognitive behaviours. More importantly, the defects of INM in hESC‐derived cerebral organoids provide a novel insight into the effects of general anaesthesia on human brain development.


| INTRODUC TI ON
Based on numerous studies in rodents and non-human primates, 1 the US Food and Drug Administration issued a warning that 'repeated and lengthy use of general anaesthetics in children younger than three years or pregnant women during their third trimester may affect the development of children's brains' (http://www. fda.gov/Drugs/ DrugS afety/ ucm53 2356.htm). Large-scale clinical studies (GAS, 2 PANDA, 3 MASK, 4 Canadian population-based studies, 5,6 ALSPAC 7 and the prospective clinical trial GAS) 8 provided no evidence of clinically and statistically significant neurological defects in young children exposed to general anaesthetics.
Controversy still exists regarding whether or how general anaesthetics affect neurodevelopment. 9,10 More importantly, these concerns might delay the necessary procedure and result in an adverse consequence.
It has been reported that 0.75%-2% of pregnant women undergo non-obstetric surgeries, and most of these surgeries are performed during the first two trimesters under general anaesthetics, [11][12][13] which is a crucial period for neurogenesis. 14 In mice, neurogenesis begins around embryonic day 10.5 (E10.5). 15 Radial glial progenitors (RGPs), derived from neuroepithelial cells to form the ventricular zone (VZ) in the embryonic brain, give rise to most, if not all, pyramidal neurons. Unlike other progenitors, RGPs are bipolar cells with their apical processes anchored at the apical of the ventricle and basal processes to contact the basal lamina. 16 The principal character of RGPs is interkinetic nuclear migration (INM), which describes the movement of nuclei along the apical-basal axis synchronized with the cell cycle. In detail, nuclei move away from the apical surface during the G1 phase and stay at the basal side of the VZ during S phase. During the G2 phase, they return to the apical surface and undergo mitosis at the apical surface of the VZ. 17 INM is vital for the efficient and continued production of neurons. 18 Newborn pyramidal neurons migrate along the basal process of RGPs to form the sixlayered structure of the neocortex in an 'inside-out' manner. 14 The sequential generation of early-born deep-layer neurons followed by late-born superficial-layer neurons ensures the laminar organization of the mature neocortex during embryogenesis. 19 Disruptions in neural progenitor maintenance and the balance between proliferation and differentiation have been shown to contribute to many neurodevelopmental disorders. 20 Recently, INM was also proposed to underlie the pathogenesis of Huntington's disease and to be responsible for the abnormalities that occur in the developing cortex in Huntington's disease patients, including changes in mitosis and cell cycle progression. 21 Clinically, sevoflurane is the most commonly used anaesthetic in pregnant women undergoing non-obstetric surgery. Our previous studies on fetal brains have demonstrated that maternal sevoflurane exposure induces the abnormal proliferation of neural progenitors. 22,23 However, the potential mechanism remains elusive.
In this study, we found that maternal sevoflurane exposure transiently induced defects in the INM of RGPs during the peak of neurogenesis in the fetal brains via the Notch signalling pathway without eliciting long-term effects. The same phenomena were observed in hESC-derived cerebral organoids. These findings provide a novel insight into the effects of general anaesthetics on human brain development.

| Human ESC culture and cerebral organoid culture
Human embryonic stem cells (hESCs) were obtained from WiCell, and cultured in a feeder-free condition. Cells were maintained with mTeSR medium (Stemcell Technologies) on the Matrigel-coated 6-well plates at 37°C supplied with 5% CO 2 . Cells were cultured and passaged using standard procedures according to the previous description. 24 Normal karyotype and contamination-free were confirmed.
Cerebral organoids were cultured as a previous publication 25 with slight modifications. Briefly, H9 hESCs were treated with 0.5 mmol/L EDTA and Accutase to obtain single-cell suspension.
Embryoid bodies (EBs) were generated with 9000 cells/well in the Ubottom, Ultra low-attachment 96-well plates (Corning) with 150 µL of mTeSR medium containing 1xRevitaCell supplement (Gibco) at day 0. Fresh mTeSR medium without RevitaCell supplement was fed to EBs at day 3. At day 5, EBs were transferred into Neural Induction (NI) medium, and medium was exchanged with fresh NI medium every second day for 6 days. EBs were then embedded into Matrigel droplets, and cultured in differentiation medium without vitamin A and shaking. Five days later, cerebral organoids were cultured in differentiation medium supplied with vitamin A on an orbital shaker.
Media were exchanged every 5 days until day 30, and used for further experiments.

| Drugs and antibodies
Drugs and antibodies used in this study can be found in Table S1.

| Animals anaesthesia
According to the previous protocol, 26 minimum alveolar concentration (MAC) of C57Bl/6 mice was tested and 2.5% sevoflurane (approximately 0.9 MAC) was adopted in this study. At E14.5, E15.5 and E16.5, which is corresponding to the first two trimesters in human, 14 the pregnant mice were randomly assigned into Control (Ctr) groups with 100% O 2 exposure or sevoflurane-treated groups (Sevo) with 2.5% sevoflurane exposure carried in 97.5% O 2 for 6 hours (hrs) as previously. 23 The mice in the Sevo group were anaesthetized in a box that was 20 × 30 × 20 cm 3 . A warm pad was used to avoid hypothermic. Arterial blood was sampled after 6 hrs of anaesthesia (data not shown) to guarantee the adequacy of ventilation and oxygenation.
Caesarean sections were performed to extract embryonic brains at the end of O 2 /Sevoflurane treatment or 24 hrs after treatment.

| Anaesthesia of hESC-derived 3D cerebral organoids
Cerebral organoids were randomly assigned to Ctr group and Sevo group. Because the water/gas partition coefficient of sevoflurane is half lower than the blood/gas partition coefficient, the concentration of 4.1% sevoflurane was used in vitro instead of 2.5% in vivo as in our previous study. 22,27 In brief, the Sevo group was exposed to 4.1% sevoflurane in a 5% CO 2 incubator, while the Ctr group was placed in another 5% CO 2 incubator without anaesthesia. Both groups were incubated at 37°C for 6 hrs and then analysed after the treatment.
Five 50 ms pulses of 33 V with 950 ms intervals were applied with a BTX electroporation system (ECM830). After electroporation, the uterine horns were placed back and the incision was sutured. The embryonic brains were used for further experiments at E15.5.

| Immunofluorescence
Timely embryonic brains were fixed with 4% paraformaldehyde in PBS overnight and transferred to 30% sucrose in PBS for 24 hrs.
Brains were embedded in tissue-Tek OCT Compound (Sakura) and cryosectioned into 14 μm thickness. Cryosections were permeated in 0.5% Triton X-100 in PBS for 30 minutes, and incubated with blocking solution (0.3% Triton X-100, 5% normal donkey serum in PBS) for an hour at room temperature. After incubation with the primary antibody at 4°C overnight, sections were incubated with fluorescence-conjugated secondary antibodies and DAPI (0.5 μg/mL in PBS, Sigma) for nuclei staining. Slices were mounted with aquapoly/mount (Polysciences). EdU staining was performed using the Click-iT EdU Alexa Fluor ® 647 kit (Thermo Fisher Scientific) according to the manufacturer's instructions. The proteins were transferred onto polyvinylidene difluoride membranes. The membranes were incubated with 5% non-fat dry milk for 2 hrs at room temperature and primary antibody at 4°C overnight.

| Western blot
After washed by TBST three times, the membranes were incubated with secondary antibody for 1 hr at room temperature, followed by washing with TBST three times. Blots were detected by ECL luminescence reagents (BBI) and imaged using ChemiDoc Imaging System (Bio-rad). The bands were quantified by densitometry (ImageJ) to determine the expression of the protein. The ratio of band density of NICD over GAPDH was calculated.

| Image acquisition and analysis
For statistics, at least three embryos from different pregnant mice were used in each group. Images were acquired by fluorescence microscopy (Nikon) and processed by NIS-Elements AR (Nikon) and ImageJ. To analyse the distribution of BrdU+ or EdU+ cells, the VZ or the cortex was divided into 10 bins or 5 bins as described previously. 28

| RNA-Seq and analysis
Total RNA was isolated from both Ctr group and Sevo group embryonic cerebral cortex at E14.5 using TRIzol reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. RNA-Seq was performed at Shanghai Majorbio Bio-pharm Technology Co., Ltd. The data were analysed on the free online platform of Majorbio Cloud Platform (www.major bio.com) and Metascape (metascape. org). Differential expression analysis was performed using DESeq2 with a cut-off of FDR <.05 and abs (log 2 FC) >1.0. The heatmap and volcano plots were generated using R programming language, and the results of GO enrichment analysis were presented using Metascapse.

| Quantitative real-time PCR (qPCR)
Total RNA was isolated, and the cDNA was synthesized using Hifair III First-Strand cDNA Synthesis Kit (Yeasen). QuantStudio 3 Real-Time PCR Systems was used for quantitative real-time PCR (Thermo Fisher Scientific) was used for qPCR with SYBR Green Master Mix (Yeasen). QuantStudio Design & Analysis Software (Thermo Fisher Scientific) was used for quantification with data normalized to the level of GAPDH mRNA. Each sample was measured in triplicate and the 2 −ΔΔCt method was used. Primers used can be found in Table S2.

| Morris water maze
Pregnant mice exposed to O 2 or sevoflurane for 6hrs at E14.5 and the male adult offspring from both Ctr group and Sevo group (n = 20 in each group) were tested in MWM at P60 according to the published protocol. 29 The young adult offspring were given training for five consecutive days (P60-P64) and probed trails on the sixth day (P65). Mice were placed under a heated lamp for 5 minutes after each trial and then were put back to the regular cages. All tracks were recorded and analysed by EthoVision XT 8.5 (Noldus).

| Statistical analysis
Data analysis and graphical presentation were performed using GraphPad Prism 8. Two-tailed unpaired Student's t test was performed for comparison of two groups of data in this study. Two-way ANOVA with repeated measurements with Bonferroni's post hoc was used to analyse the difference of escape latency in MWM and the distribution of BrdU-positive or EdU-positive cells in the VZ or in the cortex. At least three brain slices were analysed to obtain the mean number labelled by different neural markers per 100 μm surface length. Data are presented as mean ± SEM, and P < .05 was considered as statistically significant.

| Maternal sevoflurane exposure impairs the INM of the mouse RGPs during the neurogenic period
In rodents, neurogenesis peaks at E14.5 and ceases at E16. 5. 30 To investigate the effects of sevoflurane on the RGPs during the neurogenic period, BrdU (50mg/kg) was administered into pregnant mice to label S phase RGPs at E14.5, E15.5 or E16.5 before they were treated with 6 hrs O 2 or sevoflurane ( Figure S1A).

| Maternal sevoflurane exposure disturbs the cell cycle progression
INM is associated with the neurogenic process, which can influence RGP progeny. 31 To determine the effects of impaired INM on neurogenesis, we used the proliferation marker Ki67 to label proliferating RGPs 24 hrs after treatment as shown in Figure 2A Figure S2A,B), which suggested that maternal sevoflurane induces cell cycle arrest. Besides, the distribution of BrdU+ cells in the VZ still maintained abnormal 24 hrs after treatment ( Figure 2D). RGPs undergo either symmetric divisions to expand the progenitor pool or asymmetric divisions to generate neurons and intermediate progenitors (IPs). 15 The embryonic cortex was immunostained with the RGP marker Pax6 and the IP marker Tbr2 24 hrs after exposure ( Figure S2C). There was no significant difference in the number of RGPs or IPs between the two groups ( Figure S2D-G

| Sevoflurane exposure impairs the INM of RGPs in the human cerebral organoids
Organoids derived from pluripotent stem cells are able to selfassemble to mimic early developmental processes. 34 Previous studies showed that cerebral organoids display a similar organization to that of the developing human brain in the early stage. 25 Hence, to understand the potential relevance of the rodent data described above to humans, hESC-derived 3D cerebral organoids were used. On day 30, the cerebral organoids were pulsed with BrdU and treated with or without sevoflurane for 6hrs ( Figure 3A). Pax6 was used as a marker of RGPs in the VZ-like structure ( Figure 3B,C). Importantly, we found a similar altered distribution of the BrdU-labelling cells of Pax6+ RGPs ( Figure 3D) as in rodents ( Figure 1F) upon sevoflurane treatment. Therefore, our results indicate that sevoflurane exposure also impairs the INM of cerebral organoids derived from human embryonic stem cells.

| Maternal sevoflurane exposure alters the expression of genes related to neurogenic progress
RNA-Seq analysis was performed to screen out the potential mechanism underlying the effect of maternal sevoflurane exposure on the INM of RGPs. Both heatmap and volcano plots ( Figure 4A,B) showed that the gene expression profiles in the fetal cortex were indeed altered by sevoflurane. GO term analysis revealed that the upregulated genes ( Figure 4C) were mainly enriched in signalling transport (for eg, regulation of vesicle-mediated transport and regulation of cation transmembrane transport), while the downregulated genes ( Figure 4D) showed the significant enrichment in biological processes were related to neurogenesis (for example, cell division, mitotic cell cycle process, cell cycle phase transition and kinetochore organization). Taken together, the GO enrichment analysis findings indicate that sevoflurane exposure indeed affected neurogenic progress.

| The Notch signalling pathway is involved in the INM defects induced by sevoflurane
Based on the RNA-Seq analysis results, we performed qPCR on several differentially expressed genes (DEGs) that have been reported to be associated with embryonic neurodevelopment ( Figure 5A,B and Figure S3). Among these genes, Bcl6 exhibited a nearly 4-fold increase in expression after sevoflurane treatment, while the expression of Notch-related genes-RBPJ, MAML1 and Jag1 was all downregulated by at least 30%. A recent study found that Bcl6 is a single cell-intrinsic factor that ensures the robustness of neuronal fate transition 35 by repressing multiple extrinsic pathways that promote self-renewal, such as the Notch, Wnt and SHH signalling pathways. As a ligand of the Notch signalling pathway, 36 Jag1 binds to the Notch receptor and triggers activation. The Notch intracellular domain (NICD) translocates to the nucleus, forms a complex with RBPJ binding protein and recruits co-activators like MAML1 to promote the transcription of target genes. 36 Moreover, we found the protein level of NICD decreased in the fetal cortex after sevoflurane exposure at E14.5 ( Figure 5C,D). These results imply that the Notch signalling pathway is the potential mecha- Additionally, the total body weight (TBW) and the ratio of brain weight to TBW ( Figure 6H,I) had no difference between groups.
These results indicate that the learning and memory abilities of young adult offspring remain intact after maternal sevoflurane exposure, suggesting that the alteration of INM upon maternal sevoflurane exposure has been recovered during brain development. However, the underlying mechanism needs to be uncovered in the future.

| D ISCUSS I ON
We provide a new perspective on the mechanism underlying the potential neurotoxicity of sevoflurane during the embryonic stage of brain development. Our results showed that during the peak of neurogenesis, the interkinetic nuclear migration of radial glial progenitors is transiently impaired both in the mice and in the hPSC-derived 3D brain organoids after sevoflurane exposure. Furthermore, our data indicated that the Notch signalling pathway acts as a critical mechanism.  Figure 2C). Moreover, mitosis for securing self-renewal of neural progenitors in the M phase 37 was also decreased by sevoflurane exposure (Figure 1H,I). Interestingly, delayed mitosis has already been found in tissues outside the brain following anaesthesia exposure. 38 Considering the difference between rodents and human beings, hESC-derived 3D brain organoids were used in this study. Brain organoids have been widely used to explore the early development process 39,40 and are proved to be effective in the neurotoxicityrelated study in the developing brains. 41 In this study, we observed a similar INM defects pattern in VZ-like structure of 3D brain organoids after exposure to sevoflurane for 6hrs ( Figure 3D) as that in the embryonic mouse brain ( Figure 1F). This is a significant step for the study of general anaesthetics in human brain development.
In retinal neurogenesis, apical-basal notch gradient is critical in neurogenesis regulated by interkinetic nuclear migration. 42 When INM is perturbed, the nucleus is exposed to altered levels of Notch signalling, causing premature cell cycle exit and a temporal increase in neurogenesis which suggested the Notch acts as a key extrinsic path- INM is believed to be a hallmark of vertebrate neural progenitors, 16 which is vital for the efficient and continued generation of neurons. 18 In our study, although the higher ratio for progenitors was temporarily arrested in the cell cycle and abnormal INM 24 hrs after treatment ( Figure 2C,D), the proliferation and differentiation of neural progenitors (RGPs and IPs) were comparable after sevofluraneinduced INM impairment ( Figure 2E-H, Figure S2), suggesting that Additionally, the Mayo Anesthesia Safety in Kids Study 4 also showed no neuropsychological and behavioural defects in individuals aged 15-20 years exposed to anaesthesia during childhood. Our study, along with above two researches, indicated an inspiring possibility that a self-recovery mechanism exists. However, other studies in infant rhesus monkeys suggest that inhalation anaesthetics affect social behaviours, but do not impair specific cognitive domains, 45,46 which imply further investigations are needed to elucidate the potential neurotoxicity of inhalation anaesthetics on developing brains.
In conclusion, maternal sevoflurane exposure at the peak of neurogenesis transiently affects the INM of RGPs in the fetal VZ through the Notch signalling pathway but has no long-term effect neurocognitive outcome. Maternal sevoflurane exposure is possibly safe for the neurodevelopment of its offspring.

ACK N OWLED G EM ENT
We thank the members of the State Key Laboratory of Medical Neurobiology of Fudan University for technical assistance and discussion.

CO N FLI C T O F I NTE R E S T
The authors declare that there are no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
JC, FF and YLX contributed to study design. MJ, TXT, XYL, JCL, SB and LSW contributed to experimentation. MJ and YQ contributed to data analysis. JC, FF, YLX and MJ contributed to final manuscript preparation.