Melatonin suppresses senescence‐derived mitochondrial dysfunction in mesenchymal stem cells via the HSPA1L–mitophagy pathway

Abstract Mesenchymal stem cells (MSCs) are a popular cell source for stem cell‐based therapy. However, continuous ex vivo expansion to acquire large amounts of MSCs for clinical study induces replicative senescence, causing decreased therapeutic efficacy in MSCs. To address this issue, we investigated the effect of melatonin on replicative senescence in MSCs. In senescent MSCs (late passage), replicative senescence decreased mitophagy by inhibiting mitofission, resulting in the augmentation of mitochondrial dysfunction. Treatment with melatonin rescued replicative senescence by enhancing mitophagy and mitochondrial function through upregulation of heat shock 70 kDa protein 1L (HSPA1L). More specifically, we found that melatonin‐induced HSPA1L binds to cellular prion protein (PrPC), resulting in the recruitment of PrPC into the mitochondria. The HSPA1L‐PrPC complex then binds to COX4IA, which is a mitochondrial complex IV protein, leading to an increase in mitochondrial membrane potential and anti‐oxidant enzyme activity. These protective effects were blocked by knockdown of HSPA1L. In a murine hindlimb ischemia model, melatonin‐treated senescent MSCs enhanced functional recovery by increasing blood flow perfusion, limb salvage, and neovascularization. This study, for the first time, suggests that melatonin protects MSCs against replicative senescence during ex vivo expansion for clinical application via mitochondrial quality control.

Although inter-individual variability in MSCs isolated from donors is excluded, serial cell expansion induces replicative senescence, which can affect the clinical-grade MSC safety and the clinical therapeutic efficacy of MSCs (Hayflick, 1965;Loisel et al., 2017;Tarte et al., 2010). Therefore, the investigation on the replicative senescence of human MSCs for clinical application and its molecular mechanism is important for enhancing the clinical efficacy of MSCs.
Replicative senescence is associated with the inhibition of the regenerative potential of stem and progenitor cells (Correia-Melo & Passos, 2015). In particular, senescent cells increase the generation of reactive oxygen species (ROS), resulting in mitochondrial dysfunction (Korolchuk, Miwa, Carroll, & von Zglinicki, 2017). In healthy cells, mitochondrial homeostasis is well-regulated by mitochondrial quality control mechanisms, mitochondrial fission/fusion, and mitophagy. Replicative senescence impairs mitochondrial quality control, resulting in decreased mitochondrial respiratory coupling, increased ROS production, and, ultimately, dysfunctional mitochondria (Korolchuk et al., 2017;Ni, Williams, & Ding, 2015). However, it is currently unclear how replicative senescence dysregulates mitofission/fusion and mitophagy processes in mitochondrial quality control.
Melatonin is associated with several physiological functions, including sleep, circadian rhythms, and neuroendocrine actions (Jan, Reiter, Wasdell, & Bax, 2009). Accumulating evidence has also shown that melatonin regulates apoptosis, autophagy, endoplasmic reticulum stress, and anti-oxidant effects (Fernandez, Ordonez, Reiter, Gonzalez-Gallego, & Mauriz, 2015;Garcia et al., 2014;Reiter et al., 2016). Furthermore, preclinical studies suggest that melatonin enhances the therapeutic potential of MSCs in myocardial infarction, chronic kidney disease (CKD), and hindlimb ischemia Lee, Jung, Oh, Yun, & Han, 2014). Our previous studies indicate that melatonin increases the regenerative potential of MSCs in ischemic disease and CKD through upregulation of cellular prion protein (PrP C ) which is involved in self-renewal, differentiation, and angiogenesis in stem and/or progenitor cells (Doeppner et al., 2015;Han et al., 2019;. However, the mechanism by which melatonin-induced PrP C regulates the bioactivity of MSCs and protects them from stress and pathophysiological condition is still unclear. This study focused on the effect of melatonin on replicative senescence in MSCs through mitophagy. We also investigated the mechanisms by which melatonin-induced mitophagy rescues replicative senescence-associated dysfunction of mitochondria through regulation of heat shock protein, especially heat shock 70 kDa protein 1L (HSPA1L) by promoting the recruitment of PrP C into mitochondria. Finally, we assessed the functional recovery in a murine hindlimb ischemia model by engrafting melatonin-treated senescent MSCs which were cultured at late passage.

| Replicative senescence increases damaged mitochondria in MSCs via impaired mitochondrial quality control
To investigate whether replicative senescence induces abnormal mitochondria in MSCs, we assessed the morphological change of mitochondria in MSCs at early (P2) and late passage (P9) by transmission electron microscopy (TEM) analysis (Figure 1a).
Morphological analysis showed that replicative senescence in MSCs significantly increases abnormal mitochondria (Figure 1b,c).
To examine the effect of melatonin on mitochondria in senescent MSCs, we analyzed the morphology, quality control, and function of mitochondria in MSCs at P9. Melatonin significantly inhibited replicative senescence-induced abnormal mitochondrial area and size ( Figure S1, Figure 1a-c). Western blot analysis for mitofusion-associated proteins, such as phosphor-dynamin-1-like protein at Ser 637 (p-DRP (Ser 637)) as the inactive state of a mitochondrial fission-associated protein (Kashatus et al., 2011;Wang et al., 2012), mitofusion-1 (MFN1), and optic atrophy 1 (OPA1), suggested that replicative senescence significantly increased the expression of mitofusion-associated proteins, whereas melatonin significantly decreased the level of mitofusion-associated proteins in senescent MSCs (Figure 1d-f). Furthermore, mitofissionassociated protein, total DRP1 was significantly augmented in senescent MSCs treated with melatonin ( Figure S2). In addition, replicative senescence significantly increased the generation of ROS and decreased mitochondrial membrane potential in MSCs, whereas melatonin significantly inhibited the production of ROS and increased mitochondrial membrane potential in senescent MSCs (Figure 1g-j). These results indicate that melatonin protects mitochondrial dysfunction in MSCs from replicative senescence through inhibition of mitofusion.

| Melatonin facilitates the recruitment of PrP C into mitochondria through upregulation of HSPA1L
Previous studies have shown that PrP C binds to PTEN-induced kinase 1 (PINK1) which regulates mitochondrial quality control through the mitophagy process Yoon et al., 2019). To understand whether melatonin stabilizes the expression of PrP C for recruitment into mitochondria in senescent MSCs via chaperone protein, we focused on HSPA1L, which is a member of the heat shock protein 70 (HSP70) family and contributes to stabilization of protein, signal transduction, and protein homeostasis (Mayer & Bukau, 2005). The expression of HSPA1L was significantly decreased in senescent MSCs (P9), compared to that in healthy MSCs (P2; Figure 2a,b). However, melatonin significantly increased the level of HSPA1L in senescent MSCs (Figure 2a,b). Interestingly, the PrP C level in senescent MSCs was also significantly decreased, compared to that in healthy MSCs, whereas melatonin significantly augmented the expression of HSPA1L and PrP C in senescent MSCs (Figure 2a,b). These findings suggested that the expression of PrP C might be regulated by melatonininduced HSPA1L in senescent MSCs. To reveal the relationship between HSPA1L and PrP C in MSCs, co-immunoprecipitation of HSPA1L with PrP C was performed in MSCs ( Figure 2c). HSPA1L bound to PrP C in MSCs, and this binding was significantly decreased by replicative senescence (Figure 2c  whereas silencing of HSPA1L decreased the expression of PrP C ( Figure 2g). These data indicate that melatonin increases the expression of HSPA1L and the binding of HSPA1L with PrP C , resulting in the stabilization of PrP C and the recruitment of PrP C into mitochondria.

| Melatonin increases mitochondrial function via the HSPA1L-PrP C -COX4I1 complex
A previous study has revealed that the cytochrome c oxidase subunit 4 isoform 1 (COX4I1)-HSP70 complex plays a pivotal
Mitofission-associated protein, total DRP1, was also significantly inhibited in senescent MSCs treated with melatonin by knockdown of HSPA1L ( Figure S4). Silencing of DRP1 showed the decrease in melatonin-induced mitophagy in melatonin-treated senescent MSCs ( Figure S5). In addition, knockdown of MFN1 delayed replicative senescence in MSCs ( Figure S6). Furthermore, melatonin significantly increased the levels of Parkin and PINK1 in mitochondria of senescent MSCs through upregulation of HSPA1L, indicating that this F I G U R E 2 Melatonin-induced HSPA1L facilitates the recruitment of PrP C in mitochondria in senescent MSCs through binding of PrP C with COX4I1. (a) Expression of HSPA1L and PrP C in healthy MSCs (P2), senescent MSCs (P9), and senescent MSCs treated with melatonin (P9 + Mel). (b) The levels of HSPA1L and PrP C were determined by densitometry relative of β-actin, respectively. Values represent the mean ± SEM (n = 3 mitophagy-mediated effect is distinguished from general autophagy ( Figure S7). These results show that melatonin increases mitophagy in damaged mitochondria in senescent MSCs by regulating mitofusion through the expression of HSPA1L.

| Melatonin rescues replicative cellular senescence of MSCs
To examine whether melatonin protects MSCs against replicative senescence, senescence markers were assessed in healthy and senescent MSCs (Figure 4a). The expression of anti-senescence marker senescence marker protein-30 (SMP30) was significantly decreased in senescent MSCs (P9), compared to that in healthy MSCs (P2), and the levels of pro-senescence markers, p21 and p16, were significantly increased in senescent MSCs (Figure 4a,b). In senescent MSCs, melatonin markedly increased the expression of SMP30 and decreased the levels of p21 and p16, but these effects were blocked by silencing of HSPA1L (Figure 4a,b). A senescence-associated β-galactosidase assay showed that melatonin significantly inhibited the cellular se-

| D ISCUSS I ON
A variety of studies have shown that melatonin protects against senescence associated with oxidative stress (Zhou et al., 2015), neurodegeneration (Caballero et al., 2008), and CKD .
Melatonin rescues oxidative stress-induced premature senescence in MSCs through attenuation of p-p38, inhibition of p16 INK4α , and augmentation of SIRT1 (Zhou et al., 2015). It also prevents aberrant differentiation and senescence of MSCs from iron imbalance by inhibiting ROS accumulation and membrane potential depolarization through down-regulation of p53, ERK, and p38 (Yang et al., 2017). Melatonin also suppresses CKD-associated senescence in MSCs through upregulation of PrP C . We found that melatonin protects MSCs against replicative senescence-mediated mitochondrial dysfunction through activation of mitophagy. Under physiological and pathophysiological conditions, the mechanisms for mitochondrial quality control are regulated by mitochondrial fission and fusion (Ni et al., 2015). Mitofission and fusion play important roles in cell growth, division, cellular distribution, and turnover of mitochondria (Mishra & Chan, 2014). To prevent the accumulation of damaged mitochondria by several stresses, dysfunctional mitochondria are divided into two heterogeneous sets of daughter mitochondria which have increased or decreased mitochondrial membrane potential. Daughter mitochondria with higher mitochondrial membrane potential, which is better quality mitochondria, proceed to mitofusion, whereas other depolarized mitochondria with lower mitochondrial membrane potential, which is characteristic of lower quality mitochondria, are degraded by mitophagy, resulting in a turnover of mitochondria (Twig et al., 2008).

HNA DAPI
Lee, Han, Yoon, et al., 2017;Lee et al., 2018;Mayer & Bukau, 2005;Wang et al., 2013). In damaged mitochondria of HeLa cells, HSPA1L and BAG family molecular chaperone regulator 4 induces the translocation of parkin to mitochondria, leading to mitochondrial quality control (Hasson et al., 2013). In colorectal cancer cells, HIF-1α is stabilized by binding to HSPA1L (Lee, Han, Yoon, et al., 2017). Additionally, HSPA1L binds to OCT4 in cancer stem cells, resulting in maintenance of stemness (Lee et al., 2018). PrP C , the normal cellular prion protein, is a pivotal molecule with fundamental roles in self-renewal, proliferation, and angiogenesis of stem/ progenitor cells ( sulting in improvement of neovascularization in ischemic tissues . In CKD, melatonin-induced PrP C enhances mitochondrial function by binding to PINK1, leading to an increase in mitochondrial metabolism . Our results show that melatonin-induced PrP C in senescent MSCs is regulated by binding to HSPA1L and that this complex facilitates the recruitment of PrP C into mitochondria, resulting in reduced mitochondrial ROS production and increased oxidative phosphorylation. Furthermore, knockdown of HSPA1L inhibits the beneficial effect   To understand the precise mechanism by which the HSPA1L-PrP C complex regulates mitochondrial quality control in senescent MSCs, we revealed the interaction between PrP C and COX4I1. COX4I1 is the principal isoform for cytochrome c oxidase (COX). COX is an enzyme of the mitochondrial respiratory chain which plays a key role in mitochondrial oxidative phosphorylation through a transfer of electrons from cytochrome c to oxygen, leading to the induction of the proton electrochemical gradient and mitochondrial membrane potential (Li, Park, Deng, & Bai, 2006). A clinical study has shown that the defective production of COX causes myopathy and Leigh syndrome (Pillai et al., 2019). Mutation in the human COX4I1 gene induces short stature, poor weight gain, and chromosomal breaks (Abu-Libdeh et al., 2017). Furthermore, low COX4I1 expression is associated with impaired ATP production and elevated ROS (Abu-Libdeh et al., 2017).
A recent study has indicated that a COX4I1-HSP70 complex contributes to the formation of COX, leading to the maintenance of the mitochondrial membrane potential (Bottinger et al., 2013). Like this study, our results have shown that PrP C binds to COX4IA and that this binding is inhibited by knockdown of HSPA1L, suggesting that a HSPA1L-PrP C -COX4I1 complex might be a key component for regulating mitochondrial quality control in senescent MSCs.
Although replicative senescence and stress-induced premature senescence (SIPS), which is induced by various oxidative stresses including H 2 O 2 and other chemicals inducing oxidative stress, share many cellular features including altered cell morphology, DNA damage, and inhibition of cell cycle (Ho et al., 2011;Moussavi-Harami, Duwayri, Martin, Moussavi-Harami, & Buckwalter, 2004;Yu et al., 2018), several studies found different phenotypes including the phase of cell cycle arrest, global DNA methylation, telomere length, protein profiles, and gene expression between replicative senescence and SIPS (Aan, Hairi, Makpol, Rahman, & Karsani, 2013;Bielak-Zmijewska et al., 2014;Kural, Tandon, Skoblov, Kel-Margoulis, & Baranova, 2016;Pascal et al., 2005). These findings suggest that it is important to select proper model for the study on senescence due to differences in the signal pathways and molecular mechanisms between replicative senescence and SIPS.
Pathophysiological condition in several diseases reduces the therapeutic effect of MSC-based therapy. Recent reviews have described preclinical studies in which melatonin improves MSC bioactivities, including the migration to injured sites, the increase in the anti-oxidant effect, the augmentation of survival of transplanted MSCs, and alleviation of inflammation, in myocardial infarction, acute kidney disease, and limb ischemia (Hu & Li, 2019). Like these preclinical studies, we have shown that melatonin-treated senescent MSCs enhance functional recovery in a murine hindlimb ischemia model by inhibiting apoptosis, increasing proliferation, and augmenting neovascularization via HSPA1L expression. These findings suggest that in order to expand human MSCs for preclinical and/or clinical applications, treatment with melatonin may be a powerful strategy for preventing replicative senescence. Taken together, this study reveals that melatonin protects human MSCs, which are expanded at the late passage, against replicative senescence through the recruitment of the HSPA1L-PrP C -COX41A complex into mitochondria, and the removal of dysfunctional mitochondria by mitophagy.
We also found that melatonin-induced HSPA1L is an important molecule for rescuing defective mitochondrial function. These findings suggest that the regulation of HSPA1L in MSCs might provide an important clue for preventing replicative senescence and regulating mitochondrial quality control.

| E XPERIMENTAL PROCEDURE S
All detailed experimental protocols and materials are presented in the Supporting Information 1.

| Human MSCs cultures
This study was approved by the local ethic committee, and informed consent was obtained from all the study subjects. Human

| Western blot analysis
The lysates from MSCs (passages 2 and 9) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).
The protein-transferred membranes were incubated with the appropriate primary antibodies, followed by detection using HRPconjugated secondary antibodies (Cell Signaling Technology).

| Flow cytometry analysis
The mitochondrial superoxide was measured using flow cytometry analysis for MitoSOX TM (Thermo Fisher Scientific) and tetramethylrhodamine, ethyl ester (Abcam) staining.

| Immunoprecipitation
The lysates were incubated with anti-PrP C antibody and then mixed with the Protein A/G PLUS-Agarose Immunoprecipitation Reagent (Santa Cruz Biotechnology). The immunocomplexes were separated by SDS-PAGE and assessed by Western blotting.

| Superoxide dismutase activity
MnSOD activity was determined in MSCs using a SOD activity kit (Enzo Life Sciences).

| Mitochondrial complex I and IV activity
Mitochondrial complex I and IV activities in MSCs were measured using a Complex I and IV Enzyme Activity Microplate Assay Kit (Abcam).

| Murine hindlimb ischemia model
Experiments using a murine hindlimb ischemia model were performed as previously reported with minor modifications (Limbourg et al., 2009). Blood perfusion was assessed using laser Doppler perfusion imaging (LDPI; Moor Instruments, Wilmington, DE, USA).

| TUNEL assay
The TUNEL assay was performed using a TdT fluorescein in situ apoptosis detection kit (Trevigen Inc).

| Immunofluorescence staining
The ischemic areas were isolated and embedded in paraffin.
Immunofluorescence staining was performed using the appropriate primary antibodies followed by incubation with the secondary antibodies.

| Statistical analysis
Data were expressed as the mean ± standard error of the mean (SEM). All experiments were evaluated using the one-way analysis of variance (ANOVA). Comparisons of three or more groups were made using Tukey's post hoc test. A p value < .05 was considered statistically significant.

ACK N OWLED G M ENTS
The authors thank Mi Ra Yu in Dr. Noh's laboratory for isolation of MSCs. This study was supported by a National Research Foundation grant funded by the Korean government (NRF-2017M3A9B4032528, NRF-2016R1D1A3B04933480). The funders had no role in the study design, data collection or analysis, the decision to publish, or preparation of the manuscript.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. This study investigated mesenchymal stem cells which were isolated from adipose tissues of human individual (IRB: SCHUH 2017-10-016).