Gene Expression Profiles of Human Cerebral Organoids Identify PPAR Pathway and PKM2 as Key Markers for Oxygen-Glucose Deprivation and Reoxygenation

Ischemic stroke is one of the most common neurological diseases. However, the impact of ischemic stroke on human cerebral tissue remains largely unknown due to a lack of ischemic human brain samples. In this study, we applied cerebral organoids derived from human induced pluripotent stem cells to evaluate the effect of oxygen-glucose deprivation/reoxygenation (OGD/R). Pathway analysis showed the relationships between vitamin digestion and absorption, fat digestion and absorption, peroxisome proliferator-activated receptor (PPAR) signaling pathway, and complement and coagulation cascades. Combinational verification with transcriptome and gene expression analysis of different cell types revealed fatty acids-related PPAR signaling pathway and pyruvate kinase isoform M2 (PKM2) as key markers of neuronal cells in response to OGD/R. These findings suggest that, although there remain some limitations to be improved, our ischemic stroke model using human cerebral organoids would be a potentially useful tool when combined with other conventional two-dimensional (2D) mono-culture systems.


INTRODUCTION
Stroke, a cerebrovascular disease, is a common neurological disorder, and ischemic strokes are among the major causes of permanent morbidity and disability (Spescha et al., 2013;Knowland et al., 2014;Benjamin et al., 2019). Several therapeutic agents are available for acute ischemic stroke, including thrombolytics, anticoagulants, and antiplatelets. However, their efficacy is restricted and their use is limited (Catanese et al., 2017). Mechanical thrombolytics and recombinant tissue plasminogen activator (rt-PA) have therapeutic time windows for hyperacute ischemic stroke, because the target of these therapies is to rescue the penumbra around the ischemic core (Fisher and Bastan, 2012;Dorado et al., 2014). Thus, elucidating the mechanism of ischemic stroke in human cerebral tissue is expected to be important for the development of appropriate therapies.
Thus far, two-dimensional (2D) neuron cultures have been utilized in ischemic models. However, there are limitations to these models as actual brain tissues are composed of multiple cell types and responses to ischemia take place from cell to cell. Recent progress in the development of organs in-a-dish (organoids) provides potential for the modeling of various diseases (Clevers, 2016). Organoids resemble the architecture of organs that are composed of multiple cell types, and three-dimensionally (3D) cultured cerebral organoids are expected to represent physiological environment (Pasca et al., 2015). In this study, we analyzed the gene expression profiles of human cerebral organoids after oxygen-glucose deprivation/reoxygenation (OGD/R).

Human Cerebral Organoids Generation and Culture
Human induced pluripotent stem cells (iPSCs) which were Cellartis Human iPS Cell Line 12 (ChiPSC12, Y00285, Takara Bio, Kusatsu, Shiga, Japan) were cultivated in mTeSRTM1 medium (05851, Stemcell Technologies, Vancouver, British Columbia, Canada) and maintained in feeder-free condition with mTeSR-TM1 media. Feeder-free human iPSC line (XY) from Takara was obtained on six well plates (3506, Corning, New York, USA) coated with growth factors reduced Matrigel (356230; BD Biosciences, San Jose, CA, USA). At the time of passage, we added Rho-associated protein kinase (ROCK) inhibitor (final concentration 10 µM; Selleck Chemicals, Houston, Texas, USA), and maintained these cells with daily medium change without ROCK inhibitor until they reached about 70% confluency. Then, they were detached by versene solution (15040-066; Thermo Fisher Scientific, Waltham, MA, USA) and seeded by a 1:20 dilution ratio.
Cerebral organoids were differentiated according to previously published protocol (Lancaster et al., 2013). Human iPSCs were detached and subjected to embryoid body (EB) induction using the protocol. After 4 days, half of the media was replaced by human EB medium without ROCK inhibitor and basic fibroblast growth factor (bFGF). Two days later, the EBs were transferred into a neural induction media which consisted of DMEM-F12 with 1× N2 supplement (17502048; Thermo Fisher Scientific, Waltham, MA, USA), 1× Glutamax, 1× non-essential amino acids, and heparin (1 µg/ml) and then embedded in Matrigel after 5 days. Subsequently, the organoids were induced in organoid medium by using an orbital shaker. The total process took 6 weeks, from EB formation.

Immunohistochemical Analysis of Human Cerebral Organoids
Human cerebral organoids were fixed in 4% paraformaldehyde in Phosphate-Buffered Saline (PBS) overnight at 4 • C, dehydrated with 30% sucrose in PBS and embedded in O.C.T. Compound (Thermo Fisher Scientific, Waltham, MA, USA). Cryostat sections (14 µm) were cut and mounted on slides (Thermo Fisher Scientific, Waltham, MA, USA). Mounted sections were incubated at room temperature for 1 h with blocking solution [3% normal goat serum + 0.3% Triton X-100 in tris-buffered saline (TBS)] and subsequently incubated with primary antibodies diluted in blocking solution overnight at 4 • C. Antibodies specific for TUJ1 (1:400, T8660, Sigma-Aldrich, St. Louis, Missouri, USA) were used for immunostaining. After three washes with TBS, corresponding fluorescent dye Alexa Fluor 488-conjugated anti-mouse IgG (715-545-151, Jackson Immunoresearch, West Grove, PA, USA) secondary antibodies diluted in the blocking solution were added and samples were incubated at room temperature for 2 h and followed by 4 ′ ,6-diamidino-

Differentially Expressed Genes Extraction of Human Cerebral Organoids
By comparing organoids under OGD/R condition to non-treated organoids and using R package edgeR (Robinson et al., 2010) in R software (version 3.5.3), differentially expressed genes (DEGs) were extracted. The MA plot in R is a visual tool for showing the total gene expression levels of DEGs. The X-axis represents log CPM-log counts per million-and are measures of gene expression level. The Y-axis indicates log FC. Log FC is the log fold-change, which in this case, is the log difference between cerebral organoids after OGD/R.

Gene Ontology and Pathway Analysis of Human Cerebral Organoids
Using the Database for Annotation, Visualization, and Integrated Discovery [DAVID, version 6.8 database (https://david.ncifcrf. gov/)], selected DEGs were analyzed. The DAVID shows the molecular function, biological process, and cellular component expressed in the gene profile. In this study, DAVID was applied to evaluate gene ontological (GO) annotation and Kyoto encyclopedia of genes and genomes (KEGG) pathways of DEGs. P-value < 0.05 was chosen as the threshold of KEGG pathway. Upregulation and downregulation of genes were analyzed separately. United States) as a housekeeping gene and analyzed using the CT method. Statistical analysis was performed using the GraphPad Prism7 (GraphPad Software). A two-tailed unpaired Student's t-test was applied to the q-PCR data. P-value < 0.05 was defined as the threshold.

Tissue-Specific Gene Expression Analysis
Gene expression data for normal human tissues were analyzed from the genotype tissue expression project (GTEx) portal (http://www.gtexportal.org/home/) to compare the similarity of expression pattern of some genes.
Frontiers in Cellular Neuroscience | www.frontiersin.org The expression of each gene was normalized to thegeometric mean of β-actin as a housekeeping gene and analyzed using the CT method. Statistical analysis was performed using the GraphPad Prism7 (GraphPad Software). A two-tailed unpaired Student's t-test was applied to the q-PCR data. P-value < 0.05 was defined as the threshold.

Primary Cortical Neuron Culture of Mice
Cortical neurons were prepared from C57BL/6 mice on E16-17. Briefly, the mouse cerebral cortex was digested with 0.25 % trypsin and DNase for 15 min. Cells were passed through a 100 m nylon mesh. The resultant cell suspension was diluted with DMEM/F12 (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (HyClone, SH30396.03, GE Healthcare Life Sciences, Logan, UT, USA) and 1% penicillin-streptomycin (26253-84; NACALAI TESQUE, Kyoto, Kyoto, Japan). Subsequently, the cells were plated on poly-D-lysine coated dishes and maintained at 37 • C in 5% CO 2 . The culture medium was replaced with a serum-free DMEM/F12 supplemented with B27 (Thermo Fisher Scientific, Waltham, MA, USA) 24 h after plating.
Cell culture and hypoxic condition were performed three times, and reactions were run in triplicate. The expression of each gene was normalized to the geometric mean of βactin as a housekeeping gene and analyzed using the CT method. Averaged relative quantification (RQ) was calculated from the average of three RQ values per three samples, every q-PCR. Therefore, one q-PCR demonstrated two averaged RQ values (Ctrl vs. OGD/R), and finally the data from three q-PCR demonstrated six averaged RQ values: three averaged RQ values of controlled samples vs. three averaged RQ after OGD/R. Statistical analysis was performed using the GraphPad Prism7 (GraphPad Software). A two-tailed unpaired Student's t-test was applied to the q-PCR data. P-value < 0.05 was defined as the threshold.

Transcriptome Analysis of Human Brain Ischemic Stroke Model in a Dish
We used six cerebral organoids from one batch and treated three organoids with OGD/R, which is a well-established method of mimicking the pathological processes of ischemia (Hossmann, 1998), to generate a model of cerebral ischemia/reperfusion in vitro, and three for normoxic condition. We first cultured the human cerebral organoids for 42 days. Subsequently, these organoids were placed under OGD condition for 1 h and reoxygenation for 1 h. After OGD/R treatment or normoxic condition, RNAs were extracted from each human cerebral organoid prior to RNA isolation and RNA sequencing, of which samples were from every organoid sample, whereupon gene ontology analysis was performed ( Figure 1A). Immunohistochemical staining of neuronal cells' marker (TUJ1) and DAPI showed the existence of neuronal cells in our organoids model (Figure 1B). Significant DEGs are colored in red ( Figure 1C). Using the threshold of p < 0.05 and |log FC| > 0.1, a total of 52 DEGs were determined between intact cerebral organoids and cerebral organoids under OGD/R, including 14 upregulated and 38 downregulated genes ( Table 1). When false discovery rate (FDR) adjusted p-value ≤ 0.05 was used as a threshold, 15 genes of AFP, TTR, APOA2, ALB, APOA1, RNA28SN4, APOC3, FTL, AHSG, FGG, FABP1, MIR3615, FGB, RNA45SN4, and FGA were detected, and all genes were found to be downregulated ( Table 1). The top five significant downregulated genes were AFP, TTR, APOA2, ALB, and APOA1, and upregulated genes were RN7SL2, YWHAE, PTN, NNAT, RN7SL1, POSTN, DNAJB1, MIR6132, LOC105379506, and PEA15 (Table 1). In the brain, YWHAE is involved in directing the movement of nerve cells by binding to other proteins (Toyo-oka et al., 2003). Pleiotrophin (PTN), which is mainly expressed in the cerebral cortex (Kido et al., 2014;Shen et al., 2017), plays important roles in cell growth and survival, cell migration, and the expression of inflammatory cytokines (Shen et al., 2017), and POSTN leads to decreased apoptosis during hypoxia (Aukkarasongsup et al., 2013). These results indicate that upregulated genes might be associated with neuroprotection.

Co-expression Network Construction
To further investigate pathways involved in ischemic response of human brain, several network analyses were applied. Gene ontological (GO) analysis in the biological process category showed that OGD/R on human cerebral organoids induced lipid metabolism and blood coagulation ( Table 2). The pathway-based analysis, KEGG pathway, further highlighted the vitamin digestion and absorption, fat digestion and absorption, peroxisome proliferator-activated receptor (PPAR) signaling pathway, and complement and coagulation cascades ( Table 3). Using the STRING online database, 52 DEGs were filtered by the PPI network of DEGs including nodes and edges with parameters of minimum required interaction score >0.4 (medium confidence) (Figure 2). These analyses revealed potential pathways involved in cellular response to ischemic stroke.

AHSG, TTR, and FGG Are Downregulated After OGD/R in the Human Cerebral Organoids
Next, we searched for important genes involved in neurological diseases. Among the affected genes after OGD/R in human cerebral organoids, transthyretin (TTR), alpha-2-HSglycoprotein (AHSG), and fibrinogen gamma chain (FGG) were found to be related to neurological disease. AHSG is associated with alopecia-mental retardation syndrome 1 (Reza Sailani et al., 2017), patients of familial amyloid polyneuropathy (FAP) carrying TTR Met30 mutation (Tanaka et al., 1994), and fibrinogen gamma chain (FGG) appear to be protected after ischemic stroke (Cheung et al., 2008). By quantitative PCR (q-PCR), we confirmed downregulation of these three genes; TTR, AHSG, and FGG were significantly downregulated ( Figure 3A). Our results suggest that these three genes have potential to contribute to pathogenesis of ischemic stroke in the human brain.
To find the genes that play more important roles in the brain, we then verified these three genes with gene expression data of normal human tissues obtained from GTEx database. Though a vast majority of AHSG is secreted by the liver in adults (Figure 3B), AHSG is synthesized by various tissues at fetal stage (Stefan et al., 2006;Dabrowska et al., 2015). We demonstrated AHSG was downregulated after OGD/R in the human cerebral organoids ( Figure 3A). Therefore, to validate AHSG expression after OGD/R in non-neural cells, human hepatocellular carcinoma (HepG2), human fibroblasts (hTERT-BJ), and human glioblastoma (U251MG), and murine microglia (BV2 cells) as neural cells were cultured and treated under OGD/R. Ahsg or AHSG expression of murine microglia, fibroblasts, and glioblastoma was unaltered after OGD/R (Figure 3C). TTR and FGG were not detected in BV2 cells and TTR was not detected in hTERT-BJ ( Figure 3C). These results suggest that downregulated AHSG of human cerebral organoids might be specific to human fetal brain or human cerebral organoids. AHSG expression of hepatocellular carcinoma showed downward trend (1.43 fold; P = 0.09) after OGD/R ( Figure 3C). This finding might indicate the existence of a cell type with similar gene expression to that of hepatic cells in human cerebral organoids.

Expression of FTL in Human Cerebral Organoids, Murine Cells, and Non-neural Cells
Ferritin light chain (FTL) is gene responsible for neurodegeneration with brain iron accumulation (Muhoberac and Vidal, 2019), and FTL was found to be abundantly expressed in the brain (Figure 4A). Although FTL was significantly downregulated after OGD/R by using RNA sequencing (Table 1), q-PCR analysis indicated a down-regulated trend of FTL (3.40 fold; P = 0.18) and no change to the ferritin heavy chain (FTH) in the cerebral organoids after OGD/R ( Figure 4B). Immunohistochemical staining of neuronal cells' marker (TUJ1) showed the existence of neuronal cells in primary cultured cerebral cortical neurons of mice ( Figure 4C). To examine the expression level of FTL in mice, the expression level of Ftl was tested and unaltered, using primary cultured murine cortical neurons and murine microglia (BV2 cells) treated with OGD/R ( Figure 4D). FTL expression of OGD/R in non-neural cells, hepatocellular carcinoma (HepG2), telomerase reverse transcriptase immortalized fibroblasts (hTERT-BJ), and glioblastoma (U251MG), were 2D cultured and challenged by OGD/R. While the expression level of FTL was unaltered in hTERT-BJ and U251MG, it was downregulated by OGD/R treatment in HepG2 ( Figure 4D). Additionally, to analyze the influence of different hypoxic conditions in non-neural cells (hTERT-BJ and U251MG), hTERT-BJ and U251MG cells were 2D cultured and treated by mild hypoxia (i.e., approximately 10% O 2 for 24 h). The expression level of FTL was upregulated by mild hypoxia treatment in both cell lines (Supplementary Figure 1B). In murine primary cultured neurons, fibroblast, and glioblastoma cells, FTL expression was not affected by OGD/R as severe hypoxia and might be a reaction specific to human cerebral organoids.

PKM2 Is Upregulated by OGD/R in Human Cerebral Organoids and Mouse Cortical Neurons
We validated some genes which could be indicators of hypoxia and ischemic markers for cerebral organoids and other cell types. Pyruvate kinase M2 (PKM2), which is associated with glycolytic pathway, converting glucose to lactate and generating ATP (Yang and Lu, 2015), is upregulated after OGD in murine dendritic cells (Jiang et al., 2018). Thus, we searched for important gene associated with hypoxia, and PKM2 were significantly upregulated after OGD/R in the human cerebral organoids ( Figure 5A). Vascular endothelial growth factor (VEGF) was relatively upregulated (1.83 fold; P = 0.15) and hypoxia inducible factor 1A (HIF1A) were not clearly changed after OGD/R in human cerebral organoids ( Figure 5A). BV2 cells, hTERT-BJ, U251MG, and HepG2 were cultured and treated under OGD/R. Pkm2 and PKM2 did not show significant change after OGD/R (Figure 5B). These results suggest that upregulation of PKM2 of human cerebral organoids might be specific to humans. VEGFA upregulation was significantly verified in hTERT-BJ and U251MG, Vegfa was upregulated in BV2 cells, and Vegfa and VEGFA did not change in murine primary cultured neurons   and HepG2 (Figure 5C). We found that the change of hypoxic marker genes independently showed various patterns in human cerebral organoids. We focused on some genes associated with transcriptional factors. Because prolyl hydroxylase 3 (PHD3, encoded by EGLN3 gene) works as a co-activator of PKM2 and makes the connectivity of HIF1A and PKM2 strong (Luo et al., 2011), we validated expression of PHD3/Egln3. PHD3 was relatively upregulated in organoids (1.74 fold; P = 0.05) (Figure 6A). According to pathway analysis of PPAR, we validated relatively downregulated apolipoprotein A1 (APOA1) (3.86 fold; P = 0.35), apolipoprotein A2 (APOA2) (33.28 fold; P = 0.06), apolipoprotein C3 (APOC3) (20.12 fold; P = 0.05), and fatty acid binding protein 1 (FABP1) (25.20 fold; P = 0.08) (Figure 6B). The expression level of PPARA as a nuclear transcription factor was upregulated in BV2 cells and was not altered in other cell types ( Figure 6C). Because the RE1-silencing transcription factor (REST) as another nuclear transcription factor works against the oxidative stress of aging brain and Alzheimer's Disease using immunostaining of human brain (Lu et al., 2014), we tested the expression level of REST and it was not clearly changed ( Figure 6D).

DISCUSSION
In vivo models reflect many factors including central nervous system, vascular system, and immune systems, and the interaction and physiological systems make interpretations of the  results complex. On the contrary, in vitro 2D models consist of homotypic cell populations without cellular diversity and are unlikely to represent tissue architecture of actual tissues. Therefore, development and application of in vitro 3D model organoids are required to interpret the actual organ systems and cellular reactions. Additionally, in vitro maturation of human iPSC-derived 3D organoids could be a useful model for studying the pathophysiology of disease. Some genes of 3D cultured chondrocytes are highly increased and other genes are decreased compared with 2D models (Caron et al., 2012), and kidney organoids are more sensitive to nephrotoxicity than 2D cultured cell lines (Astashkina et al., 2012). Although human cerebral organoids lack fully matured neuronal tissue, our model still provides a biologically plausible means to study ischemic injury response in the human brain tissue. By using human cerebral organoids combined with series of transcriptome analysis, we revealed that TTR, AHSG, and FGG were downregulated after OGD/R. Although we were able to detect a downregulation of FTL in human cerebral organoids after OGD/R by using RNAseq, we found no significant change in q-PCR. Simultaneously we validated unaltered FTL in non-neural cells (fibroblasts and glioblastoma), murine microglia, and primary cultured murine cortical neurons after OGD/R. A larger sample size of cerebral organoids is preferable for validating FTL expression level because there are differences in maturity among individual organoids. Although FTL was more expressed than AHSG, TTR, and FGG using gene expression data for normal human tissues (Figures 3B, 4A), it remains debatable whether iron itself and iron accumulation were beneficial or harmful after experimental stroke. Some groups have shown that brain damage after ischemia is correlated with overloaded iron (Castellanos et al., 2002;Mehta et al., 2004), while others suggest that iron storage is not evil (Christensen et al., 2002;Millerot et al., 2005). Therefore, further investigations are required to clarify the role of iron and iron storage after ischemic stroke. We revealed that FTL, TTR, AHSG, and FGG are downregulated after OGD/R, but the distribution of FTL, TTR, AHSG, and FGG in human tissue is independent (Figures 3B, 4A). While FTL, TTR, AHSG, and FGG were abundantly expressed in liver, TTR, AHSG, and FGG were not sufficient in brain (Figures 3B, 4A). To validate the reaction of hepatocytes, we used HepG2 cells and FTL and AHSG tended to downregulate after OGD/R (Figures 3C, 4D). It could be possible that hepatocytes or hepatocyte-like cells co-exist in the human cerebral organoids. Although cell types of the human cerebral organoids are derived from ectodermal origin (Quadrato et al., 2017), hepatic cells are derived from endoderm origin. There might be co-existing hepatic cells or hepatocyte-like cells in the human cerebral organoids we used in this study, and this might be potential limitations of the current study.
We revealed the expression level of PKM2 of the human cerebral organoids after OGD/R was predominantly elevated. On the contrary the expression level of PKM2 of other cells after OGD/R was not evidently changed. Pkm2, both mRNA and protein, is upregulated after OGD in murine dendritic cells, which are immune-modulatory cells that do not exist in our human cerebral organoids (Jiang et al., 2018). These results suggest that the upregulation of PKM2 might be useful as a hypoxic marker in human cerebral organoids, while an expression of PKM2 might be controversial in other cell types. Although PKM2 plays protective role in acute phase of focal ischemic stroke model of mice , loss of Pkm2 using Pkm2 knockout mice also protects ischemic tissues by maintaining mitochondrial biogenesis (Hauck et al., 2020). Thus, the role of PKM2 upon OGD/R insults is still controversial. PKM2 is a target gene of HIF-1α, regulates gene expression of proteins associated with the glycolytic pathway and can accelerate metabolic reprogramming (Luo et al., 2011). Furthermore, PHD3/Egln3 induced by hypoxia works as a coactivator of PKM2 by HIF pathway (Pescador et al., 2005). PHD3/Egln3 is relatively upregulated after OGD/R. This might accelerate metabolic switch and lead to decreased high-density lipoproteins (HDLs) including APOA1, APOA2, and APOC3, all of which showed a trend of downregulation upon OGD/R ( Figure 6B). Therefore, PKM2 may act synergistically with PHD/Hif-1α in our OGD/R model. On the other hand, although some genes about fatty acid were relatively downregulated (Figure 6B), PPARA downregulation was validated only in murine microglia ( Figure 6C). The reason why PPARA was unchanged in cerebral organoids might be due to a lack of microglia derived from mesoderm. Novel system of cerebral organoids including microglia will resolve this limitation. REST is a transcriptional factor as a silencer, concerning neurogenesis, neuronal differentiation, and growth (Paquette et al., 2000). We validated relatively downregulated REST in cerebral organoids, not murine primary cultured neurons ( Figure 6D). Because some genes about neuronal movement, growth, and survival were upregulated after OGD/R using RNA-seq data set, REST downregulation might lead to neurogenesis and neuronal differentiation. These results could be specific for 3D embryonic culture system. PKM2 is inhibited by PPARγ in two breast cancer cell lines, and this inhibition might decrease ATP levels and avoid apoptosis (Shashni et al., 2013). We showed PPAR signaling pathway is associated (Figure 2) and PKM2 was upregulated after OGD/R in human cerebral organoids ( Figure 5A). This could suggest that PPARγ is downregulated and PKM2 is upregulated subsequently after ischemia. The PPARγ agonist conducts neuroprotection against focal ischemia in the rat brain (Zhao et al., 2005). Dual PPARα/γ agonist improves stroke outcome after transient cerebral ischemia in mice (Boujon et al., 2019). Therefore, therapeutic agents related lipid metabolism, including PPAR agonist, may be effective for ischemia.
Our study did not clarify either the change of common hypoxic marker genes (VEGF and HIF1A) of human cerebral organoids occurs after OGD/R (Figures 5A,C) or a change in other hypoxic marker genes (LDHA, CA9, and PGK1) (Supplementary Figure 1C). Lacking vascular system, brain organoids necrotize in their core as they grow and exist around the core healthy tissues in vitro (Lancaster, 2018). To resolve the problem of central core necrosis of brain organoids, they are transplanted into the mouse brain as vascularized model of brain organoids in vivo (Mansour et al., 2018). In vitro organotypic slice culture improved long-term survival (Giandomenico et al., 2021), and air-liquid interface culture for cerebral organoids improved survival, axon outgrowth, and decreased apoptosis rate (Giandomenico et al., 2019). It might be the reason why VEGFA and HIF1A did not significantly change, and the lack of vascular system was one of the limitation of cerebral organoids. Although VEGF and HIF1A were not markedly upregulated, we validated upregulated HSPA8, stress-induced protein: heat-shock cognate protein 70 (HSC70) coding gene, using RNA-seq (Table 1). Because HSC70 assembles after cerebral ischemia of rats (Hu et al., 1998) and HSC70 involves diverse proteostasis mechanisms including chaperone mediated autophagy which contributes to oxidative stress response (Kiffin et al., 2004;Loeffler et al., 2016), upregulated HSPA8 coding HSC70 indicated the oxidative stress of our model was existing.
There are a number of groups trying to apply organoid system to study isolated organ model in a dish. For example, James Hudson and colleagues showed that mild-hypoxia and/or metabolic changes are critical environmental triggers for maturation of cardiac cells using human cardiac organoids (Mills et al., 2019). In their experiments, they treated the human cardiac organoids by changing the serum free media, lowering glucose concentration, and increasing the concentration of fatty acid. Palmitate was used as a fatty acid substrate, as it is one of the most abundant fatty acids circulating during the neonatal period (Bougnères et al., 1982). Importantly, Hesham Sadek and colleagues discovered mild-hypoxia to be a stimulus for the metabolic switch from carbohydrates to fatty acids in a neonatal mouse cardiac tissue (Puente et al., 2014) and adult mouse cardiac tissue (Nakada et al., 2017), and therefore human cardiac organoids could model the postnatal switch in metabolic substrates from carbohydrates to fatty acids. In our experiment, we removed glucose and oxygen without substituting fatty acid. Thus, the human brain organoids do not get substrates even if the gene expression may alter metabolic switch from carbohydrates to fatty acids. The difference of FTL expression between mild-hypoxia and severehypoxia Figure 4D, Supplementary Figure 1B might suggest mild-hypoxia tends to metabolize fatty acid, compared with severe-hypoxia. Taken together, our experimental condition is not comparable to environmental changes that induce postnatal metabolic shift; instead, it mimics ischemic tissue. More detailed time course about hypoxia will be needed for further investigation in the future. It is possible that neural cells in the brain and cardiac muscle cells in the heart may respond differentially to severe hypoxia. Therefore, in order to adapt their findings from the cardiac organoids (Mills et al., 2017) to our brain organoids, neural cells would need to have similar capability to respond to same environmental stimuli. Further investigations are required to understand how neural and non-neural cells in the brain respond to pathogenic severe hypoxia.
Our goal here is to extract the direct effects on neural tissues out of spatio-temporally complex system organized by various cell types. Our cerebral organoids are free from the effect of blood vessels and blood-derived inflammatory cells invading through blood vessels. Therefore, our experimental model enabled us to evaluate the direct effect of deprivation of both oxygen and glucose followed by reoxygenation on brain tissue. Although we are aware of the limitations of applying the human cerebral organoid as a model for human ischemic stroke, there are expectations for the potential development and advancement for further applications of our model.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ddbj.nig. ac.jp/, DRA010225.

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Care Committee of Nara Medical University.