HIF1α‐mediated AIMP3 suppression delays stem cell aging via the induction of autophagy

Abstract Senescence in stem cells, which occurs as a consequence of chronic responses to the environment, defines the capacity of stem cells for proliferation and differentiation as well as their potential for tissue regeneration and homeostasis maintenance. Although stem cells reside under low oxygen pressure and the availability of oxygen is known to be a crucial determinant in their fate, the key modulators in stem cell aging and the underlying mechanism have yet to be unraveled. Human placenta‐derived mesenchymal stem cells (hpMSCs) were cultured under hypoxia (3% O2) or normoxia (21% O2) to investigate the key factors that regulate stem cell senescence under hypoxic conditions. RNA sequencing results suggested that the expression of aminoacyl‐tRNA synthetase‐interacting multifunctional protein 3 (AIMP3, EEF1E1), an aging inducer, in the hpMSCs was dramatically repressed under hypoxia with concurrent suppression of the aging marker p16INK4a. The hpMSCs that overexpressed AIMP3 under hypoxic conditions displayed significantly decreased proliferation and fewer stem cell characteristics, whereas the downregulation of AIMP3 ameliorated the age‐related senescence of MSCs. Consistent with the results of the hpMSCs, MSCs isolated from the adipose tissue of AIMP3‐overexpressing mice exhibited decreased stem cell functions. Interestingly, AIMP3‐induced senescence is negatively regulated by hypoxia‐inducible factor 1α (HIF1α) and positively regulated by Notch3. Furthermore, we showed that AIMP3 enhanced mitochondrial respiration and suppressed autophagic activity, indicating that the AIMP3‐associated modulation of metabolism and autophagy is a key mechanism in the senescence of stem cells and further suggesting a novel target for interventions against aging.

p53. The aged cells then affect inflammation and the aging of an individual by secreting specific combinations of growth factors, cytokines, and chemokines called the senescence-associated secretion phenotype (SASP; Lopez-Otin, Blasco, Partridge, Serrano, & Kroemer, 2013;Rayess, Wang, & Srivatsan, 2012;Rodier & Campisi, 2011). Given that the disruption of homeostasis induced by stem cell senescence is one of the fundamental causes of increased cellular senescence and that stem cell aging is a major risk factor for human diseases at various ages, such as cancer, neurodegeneration, weakness and metabolic diseases (Oh, Lee, & Wagers, 2014;Signer & Morrison, 2013), understanding the mechanism of stem cell aging is a prerequisite for preventing cell senescence and further clinical approaches.
The microenvironment around stem cells under low oxygen pressure, called a niche, is suitable for promoting self-renewal and proliferation via the induction of pluripotent genes, including Oct-4 and Klf-4 (Park et al., 2013). Stem cells exposed to hypoxia display increased viability, prolonged proliferation, and differentiation capacities along with delayed aging processes, including the inhibition of telomere shortening during replication and the suppression of HIF1α-TWIST-mediated E2A-p21 (Buravkova, Andreeva, Gogvadze, & Zhivotovsky, 2014;Tsai et al., 2011). Hypoxia-inducible factor (HIF), a nuclear heterodimeric transcription factor, is known as a key regulator in responses to hypoxia and consists of one of three α subunits (1α, 2α, and 3α) and one of two β subunits (ARNT and ARNT2).
The HIF complex regulates the expression of genes containing the hypoxia response element (HRE: ACGTG) that controls stress adaptation, angiogenesis, glucose metabolism, proliferation, survival, and senescence in stem cells as well as in somatic cells (Brahimi-Horn & Pouyssegur, 2009;Nakayama, 2009).
Considering the fact that hypoxic conditions are able to impede stem cell senescence, we investigated the mechanism by which HIF1α inhibits stem cell senescence and explored a novel regulatory pathway in stem cell aging using human placenta amnion-derived mesenchymal stem cells (hpMSCs). Compared with the cells cultured under normal oxygen conditions, 21% O 2 , the hpMSCs under hypoxic conditions, 3% O 2 , remained proliferative and survived for longer periods with the inhibition of p16 INK4a and p53. RNA sequence analysis revealed that aminoacyl-tRNA synthetase-interacting multifunctional protein 3 (AIMP3), a key aging-inducing factor, was suppressed under hypoxia. This result was confirmed by in vitro assays: both the hpMSCs expressing AIMP3 and the adipose-derived stem cells from AIMP3overexpressing mice, AIMP3 TG, exhibited accelerated stem cell aging and dysfunction. We found that the AIMP3 level in hpMSCs was negatively regulated by HIF1α and positively regulated by Notch3 and that the combination of HIF and Notch signals determined the cellular behaviors of hpMSCs. In addition, HIF1α mediated the repression of AIMP3-induced autophagy and suppressed mitochondrial respiration, whereas the overexpression of AIMP3 compromised autophagy function in the adipose stem cells from AIMP3 TG mice. In this study, for the first time, we found that AIMP3 is a key molecule in the hypoxiaautophagy-associated antiaging pathway and that the AIMP3-autophagy axis is a new therapeutic target for aging.

| The hpMSCs cultured under hypoxia displayed a prolonged proliferation capacity and maintained their characteristics with passages
To investigate the responses of hpMSCs to low oxygen pressure, cells were separately maintained either under hypoxia, 3% O 2 , or normoxia, 21% O 2 , immediately after isolation, which was considered passage 0 (p0). No significant difference was observed in the cellular morphology between the two groups up to p5. However, while large and flattened heterogeneous hpMSCs, a senescent phenotype, were frequently found at under normoxia at later passages, p10, the hpMSCs under hypoxia maintained their homogenous morphology at  Figure S1C).

| The hypoxic culture condition suppresses cell death and senescence-inducing pathways in hpMSCs at late passage
Given that the reduced potential for self-renewal represents stem cell aging (Signer & Morrison, 2013), the enhanced proliferative character under hypoxia indicates that low oxygen pressure suppresses stem cell aging. We performed RNA sequencing (R-seq) analysis of transcripts from the hpMSCs at p10 under normoxia and hypoxia to investigate which factors lead to the antiaging effect under hypoxia.
Gene ontological analysis showed that 57 genes induced under hypoxia were strongly associated with the positive regulation of transcription and proliferation, aging, the MAPK cascade, the regulation of PI3K signals, and responses to estradiol, drug, and insulin, indicating that the hpMSCs respond to hypoxia via responses to external signals and the active modulation of gene expression and protein synthesis (Supporting Information Table S1). However, 18 genes significantly repressed under hypoxia were associated with apoptosis, the response to cellular stress such as UV, DNA damage, and unfolded protein response, and receptor tyrosine signaling, including Erb-B2 receptor tyrosine kinase (ERBB2 ; Table 1), indicating that the prolonged capacity for self-renewal under hypoxia is through the induction of proliferation and macromolecular synthesis and the inhibition of cellular stresses and cell death. Among the highly suppressed apoptosis-related genes, the ARS interacting multifunctional protein 3 (AIMP3/EEF1E1) was notable because AIMP3overexpressing mice (AIMP3 TG), a model of progeria, manifest reduced lifespan, premature aging, reduced bone density, and wrinkled skin with reduced subcutaneous adipocytes (Oh et al., 2010), although it is unknown what role AIMP3 plays in stem cell senescence in response to differential oxygen pressures.

| The stem cells derived from AIMP3overexpressing mice exhibited compromised stem cell properties
It is possible that the weakened skin elasticity in AIMP3 TGs results from the accelerated senescence-induced differentiation deficits of the adipose-derived MSCs (AD-MSCs). To test this hypothesis, AD-F I G U R E 1 Hypoxic conditions prolonged the proliferative capacity and the maintenance of the stemness of hpMSCs. (a) The isolated hpMSCs were cultured long-term under normoxia (21% O 2 ) or hypoxia (3% O 2 ), and their morphological changes were observed. The cells under normoxia became larger, whereas the hpMSCs under hypoxic conditions maintained their shape up to the late passage. Early: passage 1 (p1), Middle: p5, and Late: p10. The scale bars represent 100 μm. (b) Population doubling level (PDL) of hpMSCs under normoxia declined from p5, and the cells stalled growth at approximately p15-18. However, the hpMSCs under hypoxia maintained their ability to proliferate up to p15. The PDL of the hpMSCs under normoxia was also significantly reduced at a given passage and trend over passages compared to that under hypoxia. **p ≤ 0.01; ***p ≤ 0.001: ## p ≤ 0.01. TG female mice and were differentiated into adipocytes under normoxia because hypoxia is reported to suppress adipogenic differentiation in our system (Yun, Maecker, Johnson, & Giaccia, 2002). The reduced expression of stem cell factors such as Sox2, c-Myc, and KLF4 was observed with increased levels of p57, a p53-dependent senescence marker, in AIMP3 AD-MSCs even from 3 M AIMP3 TG mice (Supporting Information Figure S2). Compared to that of the

AIMP3-associated stem cell senescence
Here, we suggest that AIMP3 transcription declines under hypoxia and that its forced expression accelerates stem cell aging. To investigate how AIMP3 expression is regulated and how it influences stem cell aging at low oxygen levels, hpMSCs were cultured under either normoxia or hypoxia and then were compared at different passages.
The induction of HIF1α was observed under hypoxia, whereas the level of AIMP3 under hypoxia was significantly reduced compared to normoxia ( Figure 3a). Along with AIMP3 reduction, the hypoxic condition suppressed the senescence factors p16 INK4a and p53 and induced the anti-senescent sirtuins SIRT1 and SIRT6 in the hpMSCs across passages. In addition, the reduced TIGAR (TP53-induced glycolysis and apoptosis regulator), an inhibitor of glycolysis and ROSmediated cell death, indicates that the hpMSCs under hypoxia rely less on mitochondrial respiration and are exposed to lower levels of stress (Bensaad et al., 2006;Wanka, Steinbach, & Rieger, 2012). The negative correlation between HIF1α and AIMP3 under hypoxia suggested negative regulation of AIMP3 by HIF1α. This hypothesis was supported by experiments using HIF1α RNA interference (siHIF1α) and overexpression: HIF1α suppression increased AIMP3 protein levels in the hpMSCs under hypoxia, whereas HIF1α overexpression under normoxia significantly reduced the AIMP3 protein levels (Figure 3b). As expected, the level of p16 INK4a was dependent on AIMP3 ( Figure 3c). Because HIF1α regulates the expression of its target genes via binding to hypoxia response elements (HREs: G/ACGTG), we examined the HREs within 1,000 base pair from the AIMP3 transcriptional start site (Majmundar, Wong, & Simon, 2010). We found an incomplete putative HRE sequence at the −604 position from the start codon (Supporting Information Figure S3). Additionally, under hypoxia, the interaction between endogenous HIF1α and the putative HRE within the AIMP3 promoter region was confirmed by chromatin immunoprecipitation (ChIP) assay (Figure 3d), indicating that HIF1α is directly able to regulate AIMP3. Taken together, our results suggest the possibility that HIF1α negatively regulates aging, inducing AIMP3 via its binding to the AIMP3 promoter region and inhibiting stem cell aging under hypoxia.
Among Notch signal components, Hey1 is reported to support stem cell maintenance under hypoxia, suggesting that both HIF1 and Hey1 are able to regulate stem cell aging (Gustafsson et al., 2005). To test whether HIF1 and Hey1 collaborate on the regulation of AIMP3 under hypoxia, we modulated the level of either Hey1 or HIF1α in the hpMSCs under hypoxia. Consistent with its roles in stem cells, suppression of Hey-1 under hypoxia induced AIMP3 expression without altering the HIF1α level ( Figure 4a). AIMP3 was induced in both si-HIF1α-and si-Hey1-treated hpMSCs, and the synergistic accumulation of AIMP3 was observed in both si-HIF1α-and si-Hey1-treated cells, indicating that both HIF1α and Hey1 negatively regulate AIMP3 in an additive manner (Figure 4b). Because Notch3, one of the Notch receptors, suppresses the proliferation of trophoblasts in the placenta and functions as a tumor suppressor by inducing cellular senescence (Cui, Kong, Xu, & Zhang, 2013;Zhao, Zhuang, Huang, Feng, & Lin, 2015), the differential expression of Notch3 was first examined under different oxygen levels. Under hypoxic conditions, the levels of

| AIMP3 inhibited the autophagy-associated antiaging mechanism
The suppression of AIMP3 was concomitant with alterations in apoptotic and metabolic regulators such as p53, TIGAR, SIRT1, and

| DISCUSSION
Although the underlying mechanism is still elusive, many lines of evidence suggest that adverse conditions such as increased oxidative stress cause the aging-associated stem cell dysfunction and depletion that lead to the reduction in tissue homeostasis and regeneration F I G U R E 5 AIMP3 modulates mitochondrial biogenesis and autophagy-associated stem cell aging. (a) Under normoxia, the hpMSCs treated with siAIMP3 displayed a significant increase in the LC3BII/I ratio compared to that of the control cells, indicating autophagy induction. In addition, siAIMP3-treated hpMSCs showed a remarkable significant reduction in the NDUFS3/VDAC ratio as well as p53, suggesting the repression of mitochondrial respiration in AIMP3-suppressed cells. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. (b) AD-MSCs from 3-month-old AIMP3 TG mice exhibited compromised responses to autophagy-inducing rapamycin. The rapamycin-treated wild-type AD-MSCs showed significant induction of the LC3BII/I ratio, whereas AIMP3 TG AD-MSCs were not responsive, indicating compromised autophagy activation. In addition, AIMP3 AD-MSCs maintained a significantly increased level of Bcl-xL. The level of AIMP3 was not altered in either WT or AIMP3 TG AD-MSCs. *p ≤ 0.05; ***p ≤ 0.001 growth arrest of cytotrophoblasts in the placenta under hypoxic conditions (Genbacev, Zhou, Ludlow, & Fisher, 1997), the hpMSCs cul- and death (Kim, Hur, Kim, Yoo, & Lee, 2011;Kwon et al., 2011).
Human AIMP3 is known as a potent tumor suppressor via the induction of ATM/ATR-p53-mediated cell cycle arrest in response to DNA damage, and the systemic depletion or haploidy of AIMP3 was reported to cause massive DNA damage in human cancers. In response to DNA damage, AIMP3 can dissociate from the MSC and translocate to the nucleus for DNA repair and induce ATM/ATR-p53mediated cell cycle arrest (Park et al., 2006). Recently, the induced depletion of AIMP3 was reported to cause severe DNA damage and to show a phenocopy of acute radiation syndrome in adult mice , leading to embryonic stem cell death with increased DNA damage (Kim, Jeon, Kim, & Jang, 2018). All of these findings support the functional significance of AIMP3 for DNA integrity. On the other hand, increased levels of AIMP3 are observed in aged human tissues and cells, and mice overexpressing AIMP3 (AIMP3
Both Drosophila and human AIMP3 contain a putative glutathione transferase domain that is capable of making protein-protein interactions as well as modulating cellular metabolism and metabolism-induced cellular fates (flybase.org; Kim et al., 2008). These results suggest that AIMP3 is a key determinant for controlling aging, tumorigenesis, and stemness; thus, its optimal level in the cell should be tightly regulated to prevent aberrant cell fate determination.
Here, we describe a novel mechanism regulating AIMP3 in stem cells in response to oxygen availability: in hypoxic conditions, HIF1α and Hey1 suppress AIMP3 expression and stem cell aging, whereas Notch3 show opposite effects (Figures 5 and 6). The direct transcriptional suppression of AIMP3 mediated by HIF1 is plausible upon our results and analysis: (a) The AIMP3 promoter contains an incomplete HRE sequence at −604 bp from a start codon, and a ChIP assay showed the existence of a HIF1α binding site (Figure 3b,d); (b) a putative binding site for a complex of aryl hydrocarbon receptor nuclear transporter (ARNT: HIF1β) and an aryl hydrocarbon receptor (AHR) exists at the −120 bp position, but the region was not detected by a HIF1α antibody, indicating that the precipitated fragment is specific for the HIF1α antibody against a HIF complex; and (c) both si-HIF1α treatment and a HIF1α suppressor FIH1 induced AIMP3 expression. Although HIF1α-Hey1 complex-mediated regulation is still plausible, the failure to detect their interaction in our system suggests that their synergistic suppression of AIMP3 comes from different regulatory pathways in AIMP3 expression (Figure 4b).
Contrary to the role of HIF1α and Hey1, Notch3 was first reported to enhance AIMP3 expression in this study (Figures 4 and 5). The Notch-associated regulatory mechanisms in stem cells are strongly dependent on cellular contexts, resulting in a large spectrum of outcomes ranging from stem cell expansion and survival to differentiation, senescence, and cell death. Notch3 inhibits tumorigenesis by inducing p53-p21-associated cellular senescence of many human cells and suppresses the proliferation of placental trophoblast cells, whereas it enhances the tumor progression of human prostate cancers (Cui et al., 2013;Danza et al., 2013;Liu, Sato, Cerletti, & Wagers, 2010). Considering that the interaction of Notch3 and the AIMP3 promoter is undetectable, it is assumed that the Notch3-mediated regulation is not direct. Interestingly, Raf kinase inhibitory protein (RKIP), an endogenous inhibitor of ERK, was recently reported to be negatively regulated by miR543, an AIMP3 suppressor (Du et al., 2017;Huttlin et al., 2017). Because RKIP binds to the Notch receptor and blocks its cleavage into the intracellular domain (NICD), inhibiting transcriptional activity, it is possible that miR543 modulates stem cell aging through RKIP-associated Notch regulation and direct AIMP3 suppression.
In addition to AIMP3-LMNA-mediated cellular aging, we discovered that autophagy plays a key role in AIMP3-associated cellular senescence: HIF1α-Hey1-mediated AIMP3 suppression induces autophagy and restricts mitochondrial respiration, which consequently inhibits stem cell aging ( Figure 6). Contrary to the siAIMP3- HIF1α-associated hypoxia pathways, energy metabolism, including mitochondrial respiration, and autophagy-involved cellular homeostasis: HIF1α induces pluripotency-and glycolysis-related genes and represses mitochondrial biogenesis, which reduces cellular damage and enhances the potential for self-renewal and pluripotency.
Once cellular damage accumulates, stem cells activate autophagy to remove cellular detriments and maintain their stemness (Wanet, Arnould, Najimi, & Renard, 2015). The endogenous AD-MSCs from AIMP3 TGs were less responsive to the autophagy-inducing rapamycin; however, while autophagy in AD-MSCs from WT mice was increased by approximately twofold by rapamycin, the cells from AIMP3 TGs did not exhibit rapamycin-mediated autophagy induction ( Figure 5b). Because only healthy HSCs are able to activate autophagy (Ho et al., 2017), AIMP3 overexpression may compromise stem cell functions and inhibit autophagy in AD-MSCs even from 3-month-old mice. The mechanism of the AIMP3-induced disturbance of autophagy is still under investigation, but our results clearly show that AIMP3 is a crucial regulator of autophagy-associated antiaging mechanisms in stem cells. Whether rapamycin is able to relieve the progeroid deficits manifested in AIMP3 TG mice is an interesting question because rapamycin-mediated autophagy induction rescued premature aging of a rodent model of progeria, including Hutchinson-Gilford progeria syndrome (HGPS), through the degradation of mutated lamin A protein (Blagosklonny, 2011). Our results suggest that rapamycin may not be efficient without the modulation of AIMP3 levels and that the novel HIF1α-Notch3-mediated AIMP3 regulation is a key pathway for developing antiaging interventions.
In conclusion, we present a novel regulatory mechanism of AIMP3 under hypoxia and an AIMP3-associated autophagy pathway in stem cell aging. Because multiple molecules, including HIF1α, Notch3, and autophagic molecules, are associated with AIMP3-implicated stem cell aging, the results here provide us with several possible targets for developing interventions against aging as well as expanding our knowledge about aging processes.

| Isolation and culture of hpMSCs
Human full-term placentas (≥37 gestational weeks) were obtained by Caesarean section as previously described . All

| Growth curve and population doubling time and flow cytometry analysis
Described in detail in the Supporting Information.

| Quantitative reverse transcription PCR and immunoblot analysis
Described in detail in the Supporting Information.

| Adipogenic, osteogenic, and chondrogenic differentiation of hpMSCs
Described in detail in the Supporting Information. F I G U R E 6 AIMP3 is a key modulator in autophagy-associated antiaging mechanisms in stem cells. In stem cells under hypoxia, HIF1 is able to bind to a promoter region and to suppress the expression of AIMP3 in an additive manner with Hey-1. The stem cells with repressed AIMP3 are able to activate autophagy and to reduce mitochondrial OXPHOS activity. As a result, less ROS are generated, and the aging process is delayed. However, this antiaging mechanism in stem cells was inhibited by Notch3-and FIH1mediated AIMP3 induction with hypoxia. The small RNA interference assays conducted under normoxia strongly support that AIMP3 is a key modulator in the autophagy-associated antiaging pathway as well as mitochondrial metabolism

| Chromatin immunoprecipitation assay
Described in detail in the Supporting Information.
Sunghoon Kim at the Seoul National University. All experimental animals were housed in specific pathogen-free conditions (CHA Laboratory Animal Research Center) and handled in accordance with an animal protocol approved by the CHA University Institutional Animal Care and Use Committee (IACUC number: 180091). Other procedures are described in detail in the Supporting Information.

| Isolation and culture of primary adiposederived mesenchymal stem cells from AIMP3 TG mice
Three-and 10-month-old AIMP3 TG and littermate female control mice were sacrificed for the isolation of adipose-derived mesenchymal stem cells (AD-MSCs). For autophagy induction, the AD-MSCs from 3month-old mice were treated with rapamycin (300 nmol; Sigma) for 24 hr, and the cell lysates were collected for Western blot analysis.
Other procedures are described in detail in the Supporting Information.

| Western blot analysis
Described in detail in the Supporting Information.

| mRNA sequence analysis (R-seq) and functional annotation of hpMSCs cultured under normoxia and hypoxia
The messenger RNAs were isolated from hpMSCs at p10 cultured either under normoxia or hypoxia, and differentially expressed genes (DEG) were analyzed. After identifying DEG between the two conditions, the DEGs were sorted based on 305 human aging genes from the "Aging gene" database: genes of +1.5 log2 or −1.5 log2 fold change under hypoxia compared to those under normoxia were designated as upregulated or downregulated genes, respectively. The analysis is described in detail in the Supporting Information.

| Statistics
Statistical analyses were conducted with a CHA University mainframe computer using the Statistical Package for the Social Sciences (IBM SPSS Statistics, version 22.0; IBM Korea, Inc., Seoul, Korea) and described in detail in the Supporting Information.

ACKNOWLEDGMENTS
The authors truly appreciate the invaluable support of Yun-Hwa Jeong and Sang-Hun Bae regarding statistical analysis and R-seq analysis, respectively. This work was supported by the Ministry of Trade, Industry and Energy (MOTIE, Korea, Grant number. P0004793).

CONFLI CT OF INTEREST
The authors declare that no conflicts of interest exist.