Genome-wide screening in the haploid system reveals Slc25a43 as a target gene of oxidative toxicity

Reactive oxygen species (ROS) are extensively assessed in physiological and pathological studies; however, the genes and mechanisms involved in antioxidant reactions are elusive. To address this knowledge gap, we used a forward genetic approach with mouse haploid embryonic stem cells (haESCs) to generate high-throughput mutant libraries, from which numerous oxidative stress-targeting genes were screened out. We performed proof-of-concept experiments to validate the potential inserted genes. Slc25a43 (one of the candidates) knockout (KO) ESCs presented reduced damage caused by ROS and higher cell viability when exposed to H2O2. Subsequently, ROS production and mitochondrial function analysis also confirmed that Slc25a43 was a main target gene of oxidative toxicity. In addition, we identified that KO of Slc25a43 activated mitochondria-related genes including Nlrx1 to protect ESCs from oxidative damage. Overall, our findings facilitated revealing target genes of oxidative stress and shed lights on the mechanism underlying oxidative death.


INTRODUCTION
Redox metabolism is a critical metabolic pathway in diverse biological processes, such as organism development, homeostasis and diseases. Redox metabolism includes the removal or production of reactive oxygen species (ROS) (e.g., hydrogen peroxide (H 2 O 2 ), superoxide (O 2 − ) and hydroxyl radical (HO·)) [1,2]. A moderate increase in ROS can promote cell proliferation, whereas excessive ROS production can activate oxidative stress. Oxidative stress occurs when there is an imbalance between the generation and scavenging of free radicals and induces oxidative damage to lipids, proteins, DNA and genomic stability [3][4][5]. Therefore, maintaining ROS homeostasis is crucial for normal cell growth and survival. Furthermore, ROSinduced oxidative stress is likely responsible for the onset and progression of several diseases [6]. Emerging evidence shows that cancer cells, especially those in advanced tumors, frequently exhibit high oxidative stress [7]. Pre-treatment of highly metastatic tumor cells with ROS scavengers suppresses their metastatic potential in mice [8]. Therefore, understanding ROS regulatory mechanisms might have potential value in clinical research [9].
Many efforts have been made to clarify the mechanism underlying ROS-induced oxidative stress in the context of cell death. Different types of ROS inducers have distinct diffusion capabilities and diverse effects on cells. H 2 O 2 easily penetrates the cell membrane, which makes it an ideal inducer of oxidative stress in cells [10]. Treatment with high concentrations of H 2 O 2 induces severe cell death through apoptosis [11], autophagy [12] and oxeiptosis [13]. Excessive H 2 O 2 results in cytotoxicity by directly attacking cells at the molecular level or indirectly by generating secondary reactive species such as HO· [14]. Although the intrinsic antioxidant mechanisms in cells have been widely studied recently, it is still unclear which genes are involved in oxidative stress.
Mammalian haploid cell lines are powerful tools for genetic screening to uncover unknown gene functions, owing to their single genome feature [15]. Not only mouse haESCs have advantages to discover Gpr107 as a targeting gene for ricin toxicity [16], but also haploid KBM7 cells are advanced to figure out crucial genes (including Acsl4) for ferroptosis [17].
Since haploid systems work well for metabolism-and diseaserelated genetic screening, the concept of discovering target genes of oxidative stress response mechanisms based on these systems is interesting. In this study, using the piggyBac (PB) transposonmediated method, we introduced high-throughput mutations in mouse haESCs and conducted genetic screening for oxidative toxicity, revealing numerous insertions related to the oxidative stress response. Furthermore, we determined whether the KO of candidate genes could protect ESCs from oxidative stress and discussed the potential underlying mechanism.

Purification of haploid cells by FACS and chromosome spread analysis
To purify haploid cells, dissociated cells were incubated with 5 μg/ml Hoechst 33342 (Thermo, H3570) in a water bath at 37°C for 30 min and sorted on a MoFlo XDP (Beckman, CA, USA) cell sorter. Diploid ES cells were used as a control.
For chromosome spread analysis, ESCs (haESCs and WT-diploid ESCs (WT-diESCs)) were incubated with 0.2 μg/ml nocodazole (MCE, HY-13520) for 12 h. Cells were trypsinized to generate single cells and resuspended in 75 mM KCl at 37°C for 30 min. Samples were fixed in methanol:acetic acid (3:1 in volume) for 30 min and dropped onto precooled slides. Then, the cells were stained with Giemsa (Sigma, GS500) solution for 7 min before observation.

Plasmid construction and cell transfection
The PB-trapping vector was constructed according to a previous report [19]. To construct the Slc25a43 knockout (KO) plasmids, the pSpCas9 (BB)-2A-GFP (PX458) vectors were digested and dephosphorylated by BbsI (Thermo, FD1014) and FastAP (Thermo, EF0654). Slc25a43-targeting sgRNAs were designed using the CRISPR design tool (http://crispr.mit.edu). Single-strand oligonucleotides were synthesized by a local company. Each pair of oligonucleotides was annealed and ligated into a lined PX458 vector with T4PNK (Takara, 2021A, Kusatsu, Japan). All plasmids were purchased from Addgene (MA, USA). All the primers used were listed in Table S1.
For the gene-trapping experiment, 2 × 10 6 haESCs were transfected with 3 μg of PBase plasmid and 9 μg of PB-trapping plasmid using the LTX kit (Thermo, 15338100) according to the manufacturer's instructions. Then, the cells were selected via incubation with 1 μg/ml puromycin (Thermo, A1113803) for 48 h and further screened for genes targeting oxidative toxicity. Splinkerrette PCR was utilized to identify the insertion sites after PB transfection as previously described [20]. Splinkerrette PCR products of ML1 and ML2 were sent to a local company (Novogene) for nextgeneration sequencing. The bioinformatics analysis of insertions was conducted according to a previous report [21].

Quantitative PCR and western blotting
Total RNA was extracted using the TRIzol (Thermo, 15596018) method. cDNA was obtained using Hifair II 1st Strand cDNA Synthesis SuperMix for qPCR (Yeasen, 11123ES60, Shanghai, China). Quantitative PCR was performed with SYBR Green reagents (Yeasen, 11202ES03). All the primers used were listed in Table S1.

Cell viability assays
Cell viability was assessed in 96-well plates using Cell Counting Kit-8 (Yeasen, 40203ES76). Cells were treated with 0 mM, 0.4 mM, 0.8 mM, and 1.2 mM H 2 O 2 for 0, 0.25, 4, and 8 h, respectively. Cell viability was reported as a percentage relative to the negative control.

In vivo study
Severe combined immunodeficient (SCID) mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All animal experiments were performed according to the ethical guidelines of the Nankai University Animal Centre.
Teratomas were established by subcutaneous injection of Slc25a43-KO ESCs and WT-diESCs (2 × 10 7 cells for each mouse) into SCID mice. Randomization and single blinding applied during data collection, and the volume of the teratomas was calculated using the formula (length × width 2 )/2. After the teratomas formed, the mice were intratumorally injected with H 2 O 2 (at 0, 0.25, 0.5 and 0.1% concentrations (w/w)). Each injection was performed every 2 days. After seven injections, the teratomas were imaged, removed and weighed.

RNA-seq analysis
All the RNA-seq data were sequenced by a local company (Novogene). The abundance of transcripts was counted by Kallisto with Gencode M18 and further summarized for each gene by the R package tximport. Filtered clean and sound genes were normalized using relative log-expression from DESeq2. The heatmap R function was utilized to describe the differentially expressed genes (DEGs), and ggplot2 (volcano plot) was utilized to visualize the DEGs.

RESULTS
Mutant mouse haESCs with resistance to oxidative toxicity are screened out To generate genome-wide mutant libraries, we used mouse haESCs to perform gene trapping with the designed PB-trapping vectors. The haESCs used in this study were derived from our group as reported previously, and the genetic background of the haESCs was 129 Sv/Jae [18]. These haESCs showed standard domed mouse ESC colonies (Fig. 1A). After several rounds of enrichment for haploid cells with FACS using Hoechst 33342, haESCs with a high proportion of haploid cells (Fig. S1A) were utilized to generate homozygous mutants on the whole genome scale. The chromosome spread analysis results further confirmed the haploidy in the cell cultures (Fig. S1B). Given the high transposition activity and low bias toward the genome of the PB transposon [22][23][24], we used a PB-trapping vector carrying a puromycin resistance (Puro r ) gene to allow for massive insertions (Fig. 1B). Approximately 2 × 10 6 haESCs were transfected with the PB-trapping vector and selected by 1 μg/ml puromycin. Two days later, some of the transfected haESCs survived, whereas most of the cells in the control group died (Fig. 1C). This result indicated that these Puro r haESCs were efficiently inserted with the PBtrapping vector. H 2 O 2 is a common type of oxidant, and a certain concentration of H 2 O 2 can induce severe oxidative toxicity and cause cell death in cell cultures [25]. In this regard, we chose H 2 O 2 as an oxidant to investigate genes related to oxidative stress. To detect the optimal lethal dosage of H 2 O 2 , we treated WT-haESCs with 0, 0.4, 0.8 and 1.2 mM H 2 O 2 for 4 h independently. The CCK-8 assay demonstrated that once H 2 O 2 was added, haESCs exhibited significantly reduced cell viability (Fig. 1D). The cells recovered when H 2 O 2 was withdrawn in the 0.4 mM group (data not shown). However, none of the cell cultures escaped cell death when the dose of H 2 O 2 was over 0.8 mM. Herein, we defined 0.8 mM as the optimal concentration of H 2 O 2 for the subsequent selection experiment. We treated haESCs with 0.8 mM H 2 O 2 for different times (0.25, 4 and 8 h) to find an optimal treatment time compared to the 0 h group. The viability of haESCs showed a significant reduction in the different treatment groups (Fig. 1E). None of the cells could survive when the H 2 O 2 treatment time was over 4 h, which was why we defined this duration as our treatment time. Thereafter, we designed a strategy to perform genetic screening of H 2 O 2resistant genes (Fig. 1F). Briefly, PB-trapped haESCs were treated with 0.8 mM H 2 O 2 for 4 h, and the surviving cells were collected for sequencing. The DRAQ7 staining analysis results indicated that the PB-trapped haESCs showed higher viability than the WT-haESCs after treatment with H 2 O 2 (Fig. 1G). Approximately 4 days later, some PB-trapped colonies survived, whereas the WT-haESCs hardly showed any live cells (Fig. 1H). The CCK-8 assay further confirmed this result (Fig. 1I). To assess the insertion efficiency, we randomly picked 17 subclones for the analysis of insertion sites by inverse PCR. The results demonstrated that all subclones carried insertions (Fig. S1C). The Sanger sequencing results suggested that some of the PB vectors could integrate into the gene body (Fig. S1D).
Inserted genes related to oxidative stress are revealed Next, we repeated the genetic screening experiments and screened two oxidative toxicity-targeting gene libraries independently. The results showed that multiple insertions were amplified in each library (mutant library 1 (ML1) and mutant library 2 (ML2)) by Splinkerrette PCR products (Fig. S2A). According to deep sequencing, there were~22 million independent insertions covering more than 20,000 genes in the two oxidative toxicitytargeting gene libraries. Both libraries had half of the insertions integrated into the sense orientation, while the other half integrated into the antisense orientation ( Fig. 2A). In addition, 24.10% of insertions were located in intergenic regions, and 75.90% of the insertions were integrated in intragenic regions (including coding regions, promoter regions, introns and 5′-UTRs and 3′-UTRs) in ML1. The insertions in ML2 also covered the whole genome (9.95% intergenic regions and 90.05% intragenic regions, respectively) without bias (Fig. 2B). There were 2679 overlapping inserted genes between ML1 and ML2 among the top 5000 genes (Fig. 2C). Kyoto Encyclopedia of Genes and Genomes analysis of the overlapping inserted genes of ML1 and ML2 illustrated that insertions enriched for calcium signaling and Ras signaling were the top 2 pathways (Fig. 2D). The calcium signaling pathway coordinates protein synthesis, mitochondrial activity and cell proliferation [26,27], which is involved in oxidative stress. In addition, Ras signaling is related to antioxidant capacity [28]. Gene ontology (GO) analysis data indicated that most of the inserted genes were correlated with membrane transport, mitochondria and redox (Fig. S2B). The inserted genes of both libraries related to mitochondrial function, Ras and redox metabolism were shown (Fig. 2E, F).
Knockout of Slc25a43 enhanced the oxidative stress resistance of ESCs Mitochondria play very important roles in oxidative stress [10,29]. Interestingly, most of our insertions were mitochondrial carriers. One of the mitochondrial carriers, solute carrier family 25 member 43 (Slc25a43), occurred in both ML1 and ML2 and was chosen for subsequent validation experiments. There were more independent insertions located at Slc25a43 in MLs (ML1 and ML2) than in the control group (non-H 2 O 2 screened) and were enriched in exons (Figs. 3A and S3A). Thereafter, we investigated whether Slc25a43-null could enable the antioxidant ability of the cells. We designed two Cas9-mediated guide RNA (gRNA) vectors to induce Slc25a43 KO in WT-diESCs (Fig. S3B). To enrich Cas9-GFP-positive ESCs,~11.3% of GFP-positive cells were sorted and further cultured for subclone selection (Fig. S3C). Next, 24 subclones were randomly picked and expanded for genotyping. The results showed that three subclones (#1 (−/−), #2 (+/−) and #3 (+/−)) carried different KO status in the Slc25a43 gene (Fig. S3D). The expression of Slc25a43 in the three subclones was reduced significantly, as indicated by qPCR (Fig. 3B). Western blotting further confirmed that SLC25A43 protein levels were decreased in all three subclones (Fig. S3E). To test whether KO of Slc25a43 might enhance the antioxidant capacity of mouse ESCs, we treated #1 (−/−), #2 (+/−), #3 (+/−) and WT-diESCs with 0.4 mM H 2 O 2 for 4 h independently. The CCK-8 assay results indicated that the cell viability of Slc25a43-KO lines was higher than that of WT-diESCs after treatment with H 2 O 2 (Fig. 3C). All the Slc25a43-KO lines had live colonies 2 days after H 2 O 2 treatment, whereas the control WT-diESCs hardly had any surviving cells (Fig. 3D). Oxidant stimuli usually have effects on ROS production [30], which results in tremendous harm to cells. Therefore, we detected the ROS levels in Slc25a43-null cells after H 2 O 2 treatment and compared them with those in WT-diESCs. We stained cells with C11-BODIPY and found that lipid ROS levels in Slc25a43-null ESCs were lower than those in WT-diESCs, suggesting that KO of Slc25a43 could prevent lipid peroxidation induced by H 2 O 2 ( Fig. 3E and Table S2). Similarly, cytosolic ROS increase was suppressed in the Slc25a43null ESCs compared with that in WT-diESCs according to H 2 DCFDA staining ( Fig. 3F and Table S2). To further verify that KO of Slc25a43 can protect ESCs against H 2 O 2 -induced oxidative damage, we assessed cytosolic ROS production with the optimized probe CellROX. Compared to WT-diESCs, Slc25a43-null ESCs displayed a significant reduction in H 2 O 2 -induced intracellular ROS production (Fig. 3G). To explore why the Slc25a43-null ESCs presented better cell viability than WT-diESCs treated with H 2 O 2 , we assessed the major antioxidants (reduced glutathione (GSH) and superoxide dismutase (SOD) [31]) in these cell cultures. It was reported that the absence of GSH or SOD increased oxidative damage to the cells [25,32,33]. We found that deletion of Slc25a43 maintained the GSH levels and SOD activities in cells treated with H 2 O 2 (Fig.  3H, I). Overall, the Slc25a43-KO cells had a stronger antioxidant capacity when exposed to H 2 O 2 than WT-diESCs.
We next investigated the mechanism by which Slc25a43-KO regulates oxidative stress resistance. GPX4 is a phospholipid hydroperoxidase that combines with GSH to eliminate lipid peroxides [34]. Inactivation of GPX4 with the inhibitor RSL3 triggers lipid peroxidation and results in cell death [35]. Based on these findings, we treated the cells with the inhibitor RSL3 and found that ROS production in Slc25a43-KO ESCs increased significantly when RSL3 was added (Fig. 3J). This result suggested that Slc25a43 participates in redox regulation as part of a potential mechanism of regulating lipid peroxidation associated with GPX4.

Slc25a43-null ESCs maintain mitochondrial function in the presence of H 2 O 2
Given that Slc25a43 KO induced H 2 O 2 resistance in ESCs, it was quite important to address the potential underlying mechanisms. Mitochondria play a key role in redox metabolism, and the resulting damage could promote ROS production and impair cellular function [36]. Next, we investigated whether KO of Slc25a43 could affect the function of mitochondria treated with H 2 O 2 . We found that Slc25a43-KO ESCs showed less positive MitoSox signals than WT-diESCs after H 2 O 2 treatment (Fig. 4A), which was further confirmed by flow cytometric analysis (Fig. 4B and Table S2). Previous reports showed that damaged mitochondria showed weaker fluorescence intensity because of collapsing of inner membrane potential [37]. The collapse of MMP tended to trigger cell death [38]. Slc25a43-null cells showed stronger fluorescence intensity than WT-diESCs treated with H 2 O 2 using MitoTracker Red CMSRos, a fluorescent dye dependent on the MMP (Fig. S4A). The results of rhodamine 123 staining indicated that Slc25a43-KO ESCs showed higher MMP than WT-ESCs after H 2 O 2 treatment (Fig. 4C). The prominent function of mitochondria was the production of ATP for cells via respiration. Thus, we tested the production of ATP in Slc25a43-KO ESCs and WT-diESCs treated with H 2 O 2 . The results showed that the concentrations of ATP produced in Slc25a43-KO ESCs were higher than those in WT-diESCs treated with H 2 O 2 (Fig. 4D). These data proved that H 2 O 2 treatment preserved mitochondrial function in Slc25a43-KO ESCs. To investigate whether Slc25a43 KO enhanced the antioxidant capacity of ESCs in vivo, we subcutaneously injected Slc25a43-KO ESCs and WT-diESCs separately into SCID mice. All 2-week-old teratomas were injected with 0%, 0.25%, 0.5% and 1% concentrations of H 2 O 2 separately for another 2 weeks. The results showed that there were significant differences in size, volume and weight between the Slc25a43-KO ESCs and WT-diESCs treated with 0.25% H 2 O 2 (Fig. 4E-G). However, there were no obvious differences between the two groups when treated with 0.5% and 1% H 2 O 2 , suggesting that Slc25a43 KO promoted the antioxidant capacity of ESCs at specific physiological concentrations in vivo.
Knockout of Slc25a43 activates some mitochondrial-related genes, including Nlrx1, to gain resistance to oxidative stress To investigate the mechanism by which Slc25a43 acted in the context of oxidative stress, we analyzed the global transcriptome levels of Slc25a43-KO ESCs and WT-diESCs treated with H 2 O 2 (0.4 mM, 4 h). There were 895 significant DEGs between Slc25a43-KO ESCs and WT-diESCs (Fig. 5A). According to the GO analysis, the upregulated genes were mainly enriched in defence response, cell proliferation and other functions (Fig. S5A), whereas the downregulated genes were enriched in R-SMAD binding, ROS biosynthetic process and other functions (Fig. S5B). The oxidative stress described previously was correlated with mitochondrial function, [39] the cell cycle [40] and peroxiredoxin [41]. Here, we assessed the relationships of DEGs with these three aspects. There were several oxidative stress DEGs related to mitochondrial function instead of the cell cycle and peroxiredoxin (Fig. 5B). To comprehensively understand the effects of Slc25a43-KO, we identified the top 10 upregulated genes (Fig. S5C) and the top 10 downregulated genes (Fig. S5D) among the DEGs. As shown in the list, the significantly upregulated genes (Rnf26, Rpl29 and Nlrx1) and the markedly downregulated genes (Rgcc and Casr) in Slc25a43-null ESCs were related to mitochondrial function. To validate whether the downstream genes were involved in oxidative stress, we chose one of the top 10 upregulated genes, Nlrx1, to perform validation experiments. NLRX1 is a mitochondrial nucleotide-binding, leucine-rich repeat (NLR)-containing protein that resides in the mitochondria and is encoded by Nlrx1 [42]. We introduced gene disruptions in Slc25a43 −/− ESCs by the CRISPR system with specific sgRNAs of Nlrx1 (Fig. S5E). We identified three Slc25a43-Nlrx1-double KO ESCs (DKO ESCs) subclones from 12 randomly picked subclones by genotype PCR and sequencing (Fig.  5C, D), which were treated with H 2 O 2 and analyzed for cell viability. Interestingly, only Slc25a43 −/− ESCs survived after H 2 O 2 treatment, whereas all WT-diESCs and DKO ESCs died (Fig. 5E). The CCK-8 analysis further confirmed this observation (Fig. 5F). The major antioxidant assessments showed that the GSH levels and SOD activities of DKO ESCs decreased significantly compared with those of Slc25a43 −/− ESCs after H 2 O 2 treatment (Fig. 5G, H). Taken together, these results showed that Nlrx1 plays a very important role in the antioxidative stress response of Slc25a43 −/− ESCs.
We were interested in whether Nlrx1 functions as a modulator of antioxidative stress itself. For this, we overexpressed Nlrx1 in WT-diESCs and performed subsequent assessments. Exogenous Nlrx1 was integrated into the genomes of WT-diESCs by a PB system with a GFP indicator (Fig. S6A). The expression levels of Nlrx1 in Nlrx1-overexpression (OE) ESCs were significantly higher than those in WT-diESCs and the empty vector group (Fig. 6A). Compared with the WT-diESCs and the empty vector group members, the Nlrx1-OE ESCs survived better after H 2 O 2 treatment according to observations and CCK-8 analysis (Fig. 6B, C). In addition, the Nlrx1-OE ESCs had higher GSH levels and SOD activities than did WT-diESCs and the empty vector group members when treated with H 2 O 2 (Fig. 6D, E). Overall, Nlrx1 promoted the resistance of ESCs to oxidative stress. To investigate the mechanism by which Nlrx1-OE ESCs suppresses oxidative stress, we treated Nlrx1-OE cells with RSL3. However, we did not observe any ROS production increase in Nlrx1-OE ESCs when RSL3 was added (Fig. 6F), demonstrating that the antioxidant capacity of the Nlrx1-OE cells had no effect on GPX4. To further investigate the potential mechanism underlying Nlrx1-OE ESCs, we performed RNA-seq to analyze the global transcriptome levels of Nlrx1-OE ESCs, Slc25a43-KO ESCs and WT-diESCs when treated with H 2 O 2 (0.4 mM, 4 h). There were 914 upregulated genes and 406 downregulated genes in the Nlrx1-OE ESCs compared to the Slc25a43-KO ESCs (Fig. 6G). According to GO analysis, the 914 upregulated genes were mainly enriched in nucleus-related, protein binding and other functions, whereas the 406 downregulated genes were enriched in cytoplasm-related, metal ion binding and other functions (Fig. 6H). These results suggested that a possible reason for the resistance of Nlrx1-OE ESCs to oxidative stress involved the regulation of nucleus-related, protein binding, cytoplasm-related and metal ion-binding factors.

DISCUSSION
In this study, we used haESCs to screen oxidative stress target genes because of their features of unlimited self-renewal ability and single genome phenotype. Abundant homozygous mutations were generated by combining haploid cells with PB transposons (Fig. 2B), which was consistent with previous reports [19]. We identified many genes related to oxidative toxicity from two mutant libraries (ML1 and ML2), which could be used for oxidative stress research. Most of our candidate genes were from the SLC25A family (Table S3). Previous reports showed that SLC25A family proteins were a family of carrier proteins located in the inner membranes of mitochondria that catalyze the translocation of solutes across the membrane [43,44]. It has been demonstrated that the mitochondrial membrane protein SLC25A43 contributed to cell cycle progression [45,46], but other functions of Slc25a43 were unknown. Here, we focused on the effect of Slc25a43-null on antioxidative toxicity. Our results showed that KO of Slc25a43 increased ESC resistance to H 2 O 2 -induced ROS (Fig.  3E-G). The Slc25a43-null cells showed higher GSH levels and SOD activities than the control when treated with H 2 O 2 (Fig. 3H, I), which suggested that KO of Slc25a43 enhanced the antioxidant capacity. Emerging evidence has shown that various pathological conditions caused by oxidative stress usually caused mitochondrial damage [47]. Our results showed that MMP and ATP production in SLC25A43-null ESCs was higher than that in WT-diESCs treated with oxidants (Fig.  4C, D). Therefore, KO of Slc25a43 might preserve mitochondrial function. A previous report suggested that Nlrx1 played a crucial role in ROS-induced oxidative stress [48]. The results of our study demonstrated that Slc25a43 interacted with Nlrx1 to exert protective effects on H 2 O 2 -induced cell death. In addition, we found that the expression of Slc25a43 was decreased in ESCs with increasing passaging, and the cells showed higher antioxidant capacity (data not shown), which was consistent with our results.
In summary, using haESCs, we conducted a forward highthroughput genetic screening of H 2 O 2 toxicity and identified numerous candidate genes related to the oxidative toxicity response. Furthermore, on the basis of a series of validation experiments, our findings suggested that KO of Slc25a43 could reduce harm to ESCs caused by oxidants. Moreover, our results indicated that KO of Slc25a43 protected ESCs from H 2 O 2 -induced cell death through regulation of GPX4 and Nlrx1. Furthermore, our data suggested that Nlrx1 OE protected against oxidative stress. These findings provide new insights into the mechanisms underlying the protection against many pathological processes caused by oxidative stress, including cellular senescence and neurodegenerative diseases [49][50][51].

DATA AVAILABILITY
The detailed procedures of methods, five figures and the Supplementary Information are attached.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.