DJ-1 is dispensable for human stem cell homeostasis

Safeguarding cellular redox homeostasis is crucial for maintaining organism health and preventing diseases. Oxidative stress, often characterized by decreased mitochondrial integrity and increased reactive oxygen species (ROS) production, disrupts proteostasis and genomic stability, which may eventually lead to cellular decomposition (Oh et al., 2014). Stem cells (such as mesenchymal stem cells, MSCs, and neural stem cells, NSCs) are susceptible to various external and internal stresses, and their dysfunction contributes to aging and aging-related diseases. Thus, it is of importance to elucidate the complex signaling networks regulated by oxidative stress in stem cells. Although redox signaling has been implicated in multiple cellular processes, how the redox system functions in various human stem cells is unclear. DJ-1 is considered to be a key regulator of cellular redox homeostasis. It was first identified as an oncogene and later a causative gene for autosomal recessive early-onset Parkinson’s disease (Bonifati et al., 2003; Nagakubo et al., 1997). Later, DJ-1 was reported to sense and protect against oxidative stress in neuronal cells (Biosa et al., 2017; Taira et al., 2004; Zhang et al., 2018). As an antioxidant protein, DJ-1 not only eliminates peroxide under oxidative stress by auto-oxidation but also regulates the transcription of NRF2 and its target genes (Biosa et al., 2017; Clements et al., 2006). Recently, DJ-1 was reported to function as a deglycase against lipid peroxidation, DNA oxidation and glucose oxidation implicated in aging and neurodegenerative disorders (Richarme et al., 2017; Sharma et al., 2019). While a number of studies have revealed the critical roles of DJ-1 in regulating the cellular oxidative state in multiple somatic cell lines and animal models (neuronal cells, human tumor cell lines, drosophila and mice) (Biosa et al., 2017; Kahle et al., 2009; Taira et al., 2004), the biological function of DJ-1 in human stem cells remains largely unknown. To investigate the role of DJ-1 in various human diploid cells, especially in human stem cells, we first generated DJ-1 knockout human embryonic stem cells (hESCs) using the CRISPR/Cas9 technique (Figs. 1A and S1A). Loss of DJ-1 protein was confirmed by Western blotting and immunofluorescence staining (Fig. 1B and 1C). The use of both N-terminal and C-terminal antibodies demonstrated that DJ-1 was completely ablated in DJ-1 hESCs. Immunofluorescence staining revealed that DJ-1 was localized mainly in the cytoplasm but also in the nuclei of wild type (WT, DJ-1) hESCs, and was absent in DJ-1 hESCs (Fig. 1C). The DJ-1 hESCs expressed pluripotency markers (Figs. 1D and S1B), were able to differentiate into all three germ layer lineages (Fig. S1C), and maintained normal karyotype (Fig. S1D). In addition, no remarkable difference in cell proliferation ability, cell cycle kinetics, or ROS levels was observed between DJ-1 and WT hESCs (Figs. 1E, 1F, S1E and S1F). These observations indicate that DJ-1 is dispensable for the maintenance of hESC self-renewal and pluripotency. To investigate the role of DJ-1 in human neural stem cells (hNSCs), we directly differentiated WT and DJ-1 hESCs into hNSCs. Ablation of DJ-1 was verified by Western blotting in DJ-1 hNSCs (Fig. 1G). Immunofluorescence staining revealed that DJ-1 was mainly localized in the cytoplasm and partially in the nuclei and mitochondria in WT hNSCs (Fig. 1H). Both WT and DJ-1 hNSCs expressed typical hNSC markers, including PAX6, Nestin and SOX2 (Fig. 1I). Importantly, DJ-1 hNSCs were able to efficiently differentiate into human neurons (hNeurons) as did WT hNSCs (Figs. 1J and S1G), indicating that DJ-1 deficiency had no adverse effect on neuronal differentiation ability of hNSCs. In addition, Ki67 immunofluorescence staining, clonal formation, and cell cycle analysis showed comparable proliferation abilities between WT and DJ-1 hNSCs (Figs. 1K,1L and S1H). By comparison, enhanced cell migration was observed in DJ-1 hNSCs (Fig. 1M). The expression of DNA damage response markers 53BP1 and γH2AX showed no difference between WT and DJ-1 hNSCs (Fig. 1N). We next sought to explore whether the absence of DJ-1 resulted in oxidative stress and mitochondrial dysfunction in hNSCs. No difference was detected in cellular ROS levels, lipid peroxidation, or mitochondrial mass between WT and DJ-1 hNSCs (Figs. 1O, 1P and S1I). Because DJ-1 has been implicated in cellular response to oxidative and mitochondrial stress, cell viabilities of WT and DJ-1 hNSCs were examined upon treatment with various oxidative and mitochondrial stress inducers (PX-12, paraquat, carbonyl cyanide 3-chlorophenylhydrazone (CCCP),

Safeguarding cellular redox homeostasis is crucial for maintaining organism health and preventing diseases. Oxidative stress, often characterized by decreased mitochondrial integrity and increased reactive oxygen species (ROS) production, disrupts proteostasis and genomic stability, which may eventually lead to cellular decomposition (Oh et al., 2014). Stem cells (such as mesenchymal stem cells, MSCs, and neural stem cells, NSCs) are susceptible to various external and internal stresses, and their dysfunction contributes to aging and aging-related diseases. Thus, it is of importance to elucidate the complex signaling networks regulated by oxidative stress in stem cells. Although redox signaling has been implicated in multiple cellular processes, how the redox system functions in various human stem cells is unclear.
DJ-1 is considered to be a key regulator of cellular redox homeostasis. It was first identified as an oncogene and later a causative gene for autosomal recessive early-onset Parkinson's disease (Bonifati et al., 2003;Nagakubo et al., 1997). Later, DJ-1 was reported to sense and protect against oxidative stress in neuronal cells (Biosa et al., 2017;Taira et al., 2004;Zhang et al., 2018). As an antioxidant protein, DJ-1 not only eliminates peroxide under oxidative stress by auto-oxidation but also regulates the transcription of NRF2 and its target genes (Biosa et al., 2017;Clements et al., 2006). Recently, DJ-1 was reported to function as a deglycase against lipid peroxidation, DNA oxidation and glucose oxidation implicated in aging and neurodegenerative disorders (Richarme et al., 2017;Sharma et al., 2019). While a number of studies have revealed the critical roles of DJ-1 in regulating the cellular oxidative state in multiple somatic cell lines and animal models (neuronal cells, human tumor cell lines, drosophila and mice) (Biosa et al., 2017;Kahle et al., 2009;Taira et al., 2004), the biological function of DJ-1 in human stem cells remains largely unknown.
To investigate the role of DJ-1 in various human diploid cells, especially in human stem cells, we first generated DJ-1 knockout human embryonic stem cells (hESCs) using the CRISPR/Cas9 technique (Figs. 1A and S1A). Loss of DJ-1 protein was confirmed by Western blotting and immunofluorescence staining ( Fig. 1B and 1C). The use of both N-terminal and C-terminal antibodies demonstrated that DJ-1 was completely ablated in DJ-1 −/− hESCs. Immunofluorescence staining revealed that DJ-1 was localized mainly in the cytoplasm but also in the nuclei of wild type (WT, DJ-1 +/+ ) hESCs, and was absent in DJ-1 −/− hESCs (Fig. 1C). The DJ-1 −/− hESCs expressed pluripotency markers (Figs. 1D and S1B), were able to differentiate into all three germ layer lineages (Fig. S1C), and maintained normal karyotype (Fig. S1D). In addition, no remarkable difference in cell proliferation ability, cell cycle kinetics, or ROS levels was observed between DJ-1 −/− and WT hESCs (Figs. 1E, 1F, S1E and S1F). These observations indicate that DJ-1 is dispensable for the maintenance of hESC self-renewal and pluripotency.

DJ-1 -/-
Notably, we observed minimal global changes in the gene expression profile of DJ-1 −/− hNSCs and hMSCs compared with those of WT hNSCs and hMSCs, respectively ( Fig. 2O and 2P). While DJ-1 has been reported to regulate the transcription of NRF2 and its target genes (Biosa et al., 2017), we did not detect any marked change in the expression levels of NRF2 target genes between WT and DJ-1 −/− hNSCs and hMSCs ( Fig. S3G and S3H). Only 12 upregulated genes and 18 downregulated genes were identified in DJ-1 −/− hNSCs, and 37 upregulated genes and 33 downregulated genes were found in DJ-1 −/− hMSCs relative to their WT counterparts ( Fig. S3I and S3J). Venn diagram analysis revealed 3 commonly upregulated genes and 3 commonly downregulated genes upon DJ-1 deficiency in hMSCs and hNSCs (Fig. S3K). Among those genes, we found that CHCHD2, a mitochondrial-localized antioxidant  gene, was upregulated in both DJ-1 −/− hNSCs and hMSCs ( Fig. S3L) (Aras et al., 2015). The expression levels of CHCHD2 were upregulated in DJ-1 −/− hNSCs, hMSCs, hNeurons and hVECs, but not in DJ-1 −/− hESCs, revealed by RT-qPCR, immunofluorescence and Western blotting (Figs. 2Q, 2R, S3M and S3N). Other mitochondrial-localized genes were not differentially regulated by DJ-1 deficiency in hNSCs and hMSCs (Figs. 2S, 2T, S3O and S3P), suggesting the possible transcriptional regulation of CHCHD2 by DJ-1. We next cloned the CHCHD2 promoter upstream of a luciferase reporter and detected increased CHCHD2 promoter activity in DJ-1 −/− hMSCs compared with that of WT controls (Fig. 2U). We further observed that DJ-1 bound to CHCHD2 promoter in WT hNSCs and hMSCs, but not in WT hESCs (Fig. 2V), suggesting that DJ-1 may transrepress CHCHD2 transcription in non-pluripotent cells by binding to the promoter of CHCHD2. Additionally, increased levels of active histone mark H3K4me3 and decreased levels of repressive histone mark H3K27me3 at CHCHD2 promoter were detected in DJ-1 −/− hNSCs ( Fig. S3Q and S3R). These results are in line with the notion that DJ-1 functions as a transcriptional repressor of CHCHD2 in hNSCs and hMSCs.
In this study, we generated DJ-1-deficient hESCs and differentiated them into human adult stem cells, including hNSCs and hMSCs, as well as hVECs, providing valuable experimental models for studying the biological roles of DJ-1 in various human cell types. Consistent with previous studies (Kahle et al., 2009), DJ-1 was mainly localized in the cytoplasm and nucleus and partially in the mitochondria of all cell types tested. Although DJ-1 has been implicated in antioxidative pathways and deglycation (Richarme et al., 2017;Taira et al., 2004), we found that DJ-1 deletion had no adverse effect on hESCs, hNSCs, hMSCs and hVECs at baseline and even upon treatment with various stress inducers. In particular, the absence of oxidative stress hallmarks in those DJ-1 −/− cells suggests that DJ-1 may be dispensable in the maintenance of redox homeostasis of human stem cells, at least those we tested. In addition, the absence of DJ-1 exhibited minimal impact on global gene expression. We did not observe any difference in the expression of genes previously reported to be regulated by DJ-1 in some human tumor cell lines and experimental animal models (rat, mouse, fly), such as NRF2 and its target genes (Biosa et al., 2017;Kahle et al., 2009). It is therefore possible that the antioxidative and transcription-regulatory functions of DJ-1 were species-or cell-type-specific. Interestingly, CHCHD2 transcription was remarkably upregulated in DJ-1-deficient hNSCs, hMSCs, hNeurons and hVECs, but not in DJ-1-deficient hESCs. We showed, for the first time, that DJ-1 could act as a transcriptional repressor, as binding of DJ-1 to the CHCHD2 promoter is associated with the silencing of CHCHD2 in human adult stem cells, but not in hESCs that often exhibit strong buffering capability to various cellular defects (Zhang et al., 2013). Given that DJ-1 has been reported to act as a transcriptional regulator for NRF2 and P53 (Biosa et al., 2017;Clements et al., 2006), DJ-1 might transrepress CHCHD2 expression by regulating certain client transcription factors, which awaits further investigation. As CHCHD2 is a mitochondrial-localized antioxidative protein whose deficiency disrupts mitochondrial integrity (Aras et al., 2015), the increased CHCHD2 expression we observed might reflect a compensatory antioxidative response to DJ-1 deficiency that contributed to the maintenance of mitochondrial integrity and cellular redox homeostasis. Interestingly, increased CHCHD2 expression has been linked to enhanced migration of human neurons (Shimojima et al., 2015), which is consistent with the enhanced cell migration in DJ-1-deficient hNSCs and provides a plausible clue for understanding the early pathogenesis of Parkinson's disease. However, no difference was detected in the migration abilities between WT and DJ-1 −/− hMSCs (data not shown) and hVECs, suggesting that the regulation of DJ-1 on cell migration may be cell lineagespecific. Recent studies causatively link CHCHD2 mutations with autosomal dominant and sporadic Parkinson's disease (Funayama et al., 2015), whereas loss-of-function mutations of DJ-1 have been associated with autosomal recessive Parkinson's disease with unknown mechanisms (Bonifati et al., 2003). Therefore, the normal interplay between DJ-1 and CHCHD2 may be required for the maintenance of functional homeostasis of human dopaminergic neurons, whose deregulation contributes to the onset of Parkinson's disease (Damier et al., 1999). Based on our results that CHCHD2 was upregulated in DJ-1-deficient hNSCs, panneurons, hMSCs and hVECs, we speculated that the expression of CHCHD2 may be deregulated in DJ-1-deficient dopaminergic neurons. Yet, the interplay between DJ-1 and CHCHD2 in dopaminergic neurons is still unclear, which warrants further investigation.
In summary, our data showed that the absence of DJ-1 had no adverse effect on proliferation, differentiation, and oxidative stress responses of human stem cells, such as hNSCs and hMSCs as well as hVECs, suggesting that loss of DJ-1 function alone is insufficient to disrupt the homeostasis of these human cells. In addition, we found that CHCHD2 was upregulated upon DJ-1 deficiency, which may account for the absence of severe phenotypes in various types of DJ-1-deficient cells. For the first time, our study revealed the 'see-saw' expression pattern of two Parkinson's disease-associated genes, providing potential clues for understanding the mechanisms of DJ-1and CHCHD2-associated Parkinson's disease.

FOOTNOTES
We are grateful to Lei Bai, Qun Chu, Ruijun Bai, Jing Lu, Ying Yang and Shikun Ma for administrative assistance, to Junying Jia (IBP, CAS) and Shuang Sun (IBP, CAS) for their help in the flow cytometry experiments, to Yihui Xu (Key Laboratory of Infection and Immunity, IBP, CAS) for her help in optical in vivo imaging, and to Yi Yang (East China University of Science and Technology) for providing iNap1 and iNapc probes. This work was supported by the National Key