Adult mesenchymal stem cell ageing interplays with depressed mitochondrial Ndufs6

Mesenchymal stem cell (MSC)-based therapy has emerged as a novel strategy to treat many degenerative diseases. Accumulating evidence shows that the function of MSCs declines with age, thus limiting their regenerative capacity. Nonetheless, the underlying mechanisms that control MSC ageing are not well understood. We show that compared with bone marrow-MSCs (BM-MSCs) isolated from young and aged samples, NADH dehydrogenase (ubiquinone) iron-sulfur protein 6 (Ndufs6) is depressed in aged MSCs. Similar to that of Ndufs6 knockout (Ndufs6−/−) mice, MSCs exhibited a reduced self-renewal and differentiation capacity with a tendency to senescence in the presence of an increased p53/p21 level. Downregulation of Ndufs6 by siRNA also accelerated progression of wild-type BM-MSCs to an aged state. In contrast, replenishment of Ndufs6 in Ndufs6−/−-BM-MSCs significantly rejuvenated senescent cells and restored their proliferative ability. Compared with BM-MSCs, Ndufs6−/−-BM-MSCs displayed increased intracellular and mitochondrial reactive oxygen species (ROS), and decreased mitochondrial membrane potential. Treatment of Ndufs6−/−-BM-MSCs with mitochondrial ROS inhibitor Mito-TEMPO notably reversed the cellular senescence and reduced the increased p53/p21 level. We provide direct evidence that impairment of mitochondrial Ndufs6 is a putative accelerator of adult stem cell ageing that is associated with excessive ROS accumulation and upregulation of p53/p21. It also indicates that manipulation of mitochondrial function is critical and can effectively protect adult stem cells against senescence.


Introduction
Preclinical and clinical trials have revealed mesenchymal stem cell (MSC)-based therapy to be a promising therapeutic strategy for many diseases [1][2][3][4][5] . Compared with other types of stem cells currently under investigation, MSCs have several appreciable advantages, including easy isolation and being highly expandable with multilineage differentiation potential and low immunogenicity. Nonetheless an increasing body of evidence has demonstrated that their benefits decline during the aging process. Prolonged ex vivo cell culture of MSCs isolated from ageing donors show a senescent state, and reduced proliferation and impaired differentiation capacity, thus limiting their clinical application 6,7 although the molecular network governing this senescence largely remains elusive. Exploration of the mechanisms that underlie cellular senescence is urgently needed.
To date, several potential mechanisms, including telomere shortening 8 , impaired autophagy 9 , and increased reactive oxygen species (ROS) have been reported to mediate the senescence of MSCs. Importantly, in addition to these factors stem cell senescence is strongly associated with mitochondrial dysfunction. One theory is that dysfunctional mitochondria accumulate with age 10 , while another proposes that mitochondrial dysfunction can directly induce stem cell ageing and impair autophagic function with a consequent decline in their regenerative function 11,12 . Mitochondria can no longer be viewed as simple bioenergy factories, but rather as platforms for intracellular signaling, regulators of innate immunity, and modulators of stem cell activity. In turn, each of these properties provides clues as to how mitochondria might regulate aging and age-related diseases 13 . Compared with young MSCs, senescent MSCs exhibit increased ROS, largely as a result of mitochondrial structural remodeling 14 . Inflammation-induced ROS leads to MSC senescence by upregulating the expression of miR-155 that in turn suppresses the expression of redox genes including Nfe2l2, Sod1, and Hmox1 15 . Moreover, cholesterol reduces senescence in bone marrow-MSCs (BM-MSCs) by inhibiting the ROS/p53/p21 Cip1/Waf1 pathway, indicating that ROS plays a very important role in the senescence of MSCs 16 . To the best of our knowledge, mitochondria are the major source of ROS and structural alterations to mitochondria result in their excessive level 17 .
Mitochondrial dysfunction with consequent increased ROS level is closely related to cellular senescence 18,19 . Among the mitochondrial complexes in the electron transport chains, mitochondrial complex I is most commonly impaired in aged MSCs 20,21 . It is the largest component of the mitochondrial respiratory chain and the major entry point for electrons into the process of oxidative phosphorylation (OXPHOS). The OXPHOS machinery in mitochondria has five complexes, including complex I, complex II (succinate dehydrogenase), complex III (cytochrome bc1 complex), complex IV (cytochrome c oxidase, COX), and complex V (ATP synthase) 22 . These complexes are localized within the inner mitochondrial membrane. Complex I mediates ROS generation, and complex I defect leads to increased ROS production and decreased antioxidant defenses, thus impairing mitochondrial function [23][24][25] . Although complex I consists of at least 45 subunits, little is known about which subunit contributes to its stability and activity in MSC ageing.
NADH dehydrogenase (ubiquinone) iron-sulfur protein 6 (Ndufs6) is one of the major accessory subunits of complex I. Loss of this subunit caused by mutations has been reported to lead to fetal disease via mitochondrial complex I deficiency 26 . A previous study also demonstrated that Ndufs6 knockdown resulted in complex I instability and functional deficiency, leading to renal impairment due to increased ROS generation 23 . Given that Ndufs6 has been linked to complex I function, we hypothesized that Ndufs6 plays putative roles in MSC ageing and that Ndufs6 mediates cell senescence by regulating complex I function. Whether and how Ndufs6 mediates MSC senescence has nevertheless not been determined. In this study, we reveal that Ndufs6 plays a critical role in the regulation of BM-MSC senescence, and downregulation of Ndufs6 accelerates BM-MSC senescence via complex I deficiency with consequent increased ROS generation and activation of the p53/p21 signaling pathway.

mRNA expression analysis
Five public mRNA expression microarray datasets were utilized to identify mRNA expression changes to mRNA of Ndufs1-8 genes during the aging process in humans and mice. They included (1) GSE9593 (GEO accession number) representing early and senescent passages of human BM-derived MSCs. (2) GSE35959 representing human MSC aging and primary osteoporosis. (3) GSE70376 comparing muscle stem cells (satellite cells) in homeostatic conditions and after cardiotoxin injury. (4) GSE47177 comparing quiescent satellite cells from mouse hindlimb muscle mouse uninjured young and old ones. (5) GSE56560 comparing young and senescent human MSCs from two donors. The processed data of these studies were extracted from GEO database. Expression data of Ndufs1-8 genes were selected for the linear regression analysis.

RT-PCR
Genes related to adipogenesis (LPL and Pparg), chondrogensis (Acan and Col2a1), and osteogenesis (Bglap and Alpl) were detected by real-time PCR (RT-PCR), as previously reported 29 . Total RNA was extracted using RNeasy kit (Qiagen, 74104). A total of 1 µg mRNA was reversetranscribed using PrimeScript RT reagent kit (Takara, RR047A). The program for PCR was as follows: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min. The primers for mRNA were as follows: LPL (F: TAACTGC SA-β-gal staining SA-β-gal staining was performed according to the manufacturer's instructions (Cell Signaling Technology). Briefly, after washing with PBS, MSCs were fixed with fixative solution for 15 min at room temperature and then stained with SA-β-gal staining solution at 37°C overnight. Senescent MSCs were stained blue and were photographed. The percentage was calculated from five different view fields of each sample in three independent experiments.

Transmission electron microscope
The mitochondrial morphology and autophagosomes of BM-MSCs and Ndufs6 −/− -BM-MSCs were evaluated by a transmission electron microscope (TEM), as previously described 30 . The samples were photographed randomly under a TEM (Hitachi, H-7650). Finally, the mitochondrial length and number of autophagosomes were calculated for five different view fields of each group in three independent experiments.

Immunofluorescence staining
Different groups of MSCs were cultured in 24-well plates with glass coverslips and fixed with 4% PFA, permeabilized with 0.1% Triton-100, and then blocked by 5% bovine serum albumin. Subsequently, cells were incubated with rabbit anti-Ki-67 (Abcam, ab15580) at 4°C overnight followed by fluorescent secondary antibody. Finally, the sample was mounted with DAPI, and ten random fields photographed. The percentage of Ki-67-positive cells was calculated.

siRNA intervention
To knockdown of Ndufs6 in MSCs, Ndufs6-siRNA (Santa Cruz, sc-149888) and control siRNA (Santa Cruz, sc-37007) were used to transfect MSCs with a Lipofectamine RNAiMAX Reagent Kit (Invitrogen, 13778-075) at a standardized MOI (multiplicity of infection) of 5 according to the protocol. After 72 h, MSCs were harvested and the silencing efficiency was evaluated by western blotting.

Western blotting
The proteins of the different groups of MSCs were extracted and the concentration measured by Bradford. A total amount of 25 μg protein of each sample was loaded, separated by SDS/PAGE, and then transferred to a PVDF membrane. The membrane was blocked with 5% fat-free milk in TBST and then incubated with the following antibodies: OXPHOS (Abcam, ab110413), Ndufs6 (Gen-eTex, GTX88043), p53 (Santa Cruz, SC-98), p21 (Santa Cruz, SC-271532), and GAPDH (Santa Cruz, SC-137179) at 4°C overnight. After washing with TBST, the membrane was incubated with horseradish peroxidaseconjugated secondary antibodies (1:10,000; Santa Cruz) at room temperature for 1 h and then allowed to develop.

Measurement of ROS
Intracellular and mitochondrial ROS were measured by 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) and Mito-Sox staining, respectively. Briefly, MSCs from different experimental groups were cultured in 24-well plates with glass coverslips. Then MSCs were incubated with 10 μM DCFH-DA (Invitrogen, C369) or 5 μM Mito-Sox (Invitrogen, M36008) for 10 min at 37°C in the dark. Finally, the sample was photographed randomly and fluorescence intensity calculated for five different view fields of each group in three independent experiments, using Image J software. Moreover, the ROS level was also examined by flow cytometry. Briefly, MSCs were grown to 80% confluence in 6-cm culture dish, and incubated in 10 μM DCFH-DA or 5 μM Mito-Sox with fluorescenceactivated cell-sorting staining buffer at 37°C for 10 min. After incubation, MSCs were washed twice with PBS and analyzed by flow cytometry (FC500; Beckman Coulter). A shift to the right indicates increased ROS levels. BM-MSCs were treated with 10 mM H 2 O 2 for 6 h for DCFH-DA staining or 100 mM antimycin A for 30 min for Mito-SOX staining as positive control.

TMRM staining
Mitochondrial membrane potential (MMP) was detected by tetramethylrhodamine ethyl ester perchlorate (TMRM, Invitrogen, T668) according to the protocol. Briefly, MSCs from different experimental groups were cultured in 24-well plates with glass coverslips and then stained with 50 nM TMRM for 10 min. Finally, the sample was photographed randomly and fluorescence intensity calculated for five view fields of each group in three independent experiments, using Image J software. Moreover, we also analyzed the MMP with TMRM staining, using flow cytometry. BM-MSCs were treated with 10 μM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) for 1.5 h as positive control.

AAV transfection
The adeno-associated virus-Ndufs6 (AAV2-Ndufs6) plasmid was constructed as previously reported 31 . Briefly, the open reading frame of human Ndufs6 (NM_004553.4) was amplified and cloned into an AAV2 core vector (Fig.  S1). After the Sanger sequencing, the AAV2-Ndufs plasmids were packaged for viral products (BioWit Technologies, China). Ndufs6 −/− -BM-MSCs were infected using AAV2-Ndufs6 virus with an MOI of 20,000. The empty AAV2-infected Ndufs6 −/− -BM-MSCs cells were used as negative control. The infection efficiency was confirmed by western blotting for Ndufs6 at 72 h post infection.

Statistical analysis
All values are expressed as mean ± SD. Statistical tests were performed using Prism version 5.0c (GraphPad Software). Statistical significance was determined by independent-samples T test between two groups or analysis of variance followed by Bonferroni test between more than two groups. A p < 0.05 was considered significant.

Accessory subunit of mitochondrial complex I Ndufs6 is downregulated in aged MSCs
To examine the relationship between accessory subunits of mitochondrial complex I and aging, we first extracted gene expression of Ndufs1-8 from five publicly available microarray datasets representing young versus old/senescent cells and tissues. The datasets included (1) GSE9593 (GEO accession number) representing early and senescent passages of human BM-derived MSCs. (2) GSE35959 representing human MSCs from young, aging, and primary osteoporosis. (3) GSE70376 representing muscle stem cells (satellite cells) in homeostatic conditions and regenerative cardiotoxin injury-stimulated muscle stem cells from aged mice. (4) GSE47177 compared quiescent muscle stem cells (satellite cells) from uninjured mice, young and old. (5) GSE56560 representing passage 9 human MSCs from two donors at days 2 and 7 of culture. In all five datasets, we observed a similar decrease in mRNA expression of all eight Ndufs subunits during the aging process (Fig. S2). In particular, expression of Ndufs3, 4, and 6 was consistently and prominently decreased. We further analyzed the difference between Ndufs6 mRNA from young and old/ senescent cells and tissues using linear regression. A scatter plot for mRNA expression (relative to young) of all Ndufs subunits of mitochondria complex assembly I is shown in Fig. 1A. The mRNA expression of Ndufs6 is significant different between young and old cells or tissues (Fig. 1B). Linear regression of mRNA expression in the young is larger than that of the old samples (Fig. 1C). RT-PCR results showed that the mRNA level of Ndufs6 was much lower in aged-BM-MSCs derived from mice than young-BM-MSCs (Fig. S3). Further, the protein level of Ndufs6 was greatly reduced in aged-BM-MSCs derived from mice compared with young-BM-MSCs (Fig. 1D). This preliminary finding indicates that a dominant decline in expression of Ndufs6 may be a pivotal factor that provokes MSC ageing or a consequence in aged MSCs.

Reduced differentiation potential of BM-MSCs in the absence of Ndufs6
Given the important role of Ndufs6 in complex I activity and MSC ageing, we next characterized BM-MSCs from WT and Ndufs6 −/− mice 27 Fig. 1 mRNA expression change of Ndufs6 during aging in human or mouse MSCs. mRNA expression data were extracted from five microarray datasets representing young vs. old/senescent cells and tissues or early stages of MSC differentiation. A Scatter plot for mRNA expression (relative to young) for all subunits (Ndufs1-8) of mitochondria complex assembly I. B mRNA expression difference of Ndufs6 between young and old cells or tissues (n = 32). C Linear regression of mRNA expression of Ndufs6 between young and old cells or tissues (n = 30). D The protein level of Ndufs6 in young-BM-MSCs and aged-BM-MSCs isolated from mice was determined by western blotting. Significant difference between young and old cells or tissues, **p < 0.01 ***p < 0.001. (Fig. S4D-F). These results indicate that the differentiation functions of Ndufs6 −/− -BM-MSCs dramatically declined compared with WT BM-MSCs.

Ndufs6 low-BM-MSCs exhibits increased cellular senescence
Increasing evidence has shown that dysfunction of MSCs is caused by premature ageing of stem cells. We therefore compared the cellular senescence of BM-MSCs and Ndufs6 −/− -BM-MSCs. We first examined cellular proliferation by counting cell numbers of BM-MSCs and Ndufs6 −/− -BM-MSCs, and confirmed that the cellular proliferation of Ndufs6 −/− -BM-MSCs was significantly reduced (Fig. 2A). BM-MSCs exhibited a healthy spindle shape, whereas Ndufs6 −/− -BM-MSCs showed an enlarged and fried egg-like shape (Fig. 2B) with a notably increased cell area compared with BM-MSCs (Fig. 2C). Moreover, SA-β-gal staining showed that the number of SA-β-galpositive cells was greatly enhanced in Ndufs6 −/− -BM-MSCs compared with BM-MSCs (Fig. 2D, E). To further verify the senescence of Ndufs6 −/− -BM-MSCs, we performed Ki-67 staining to examine their proliferative capacity and showed that the of number Ki-67-positive cells was much lower in Ndufs6 −/− -BM-MSCs than BM-MSCs (Fig. 2F, G). RT-PCR results showed that the mRNA level of p53 and p21, senescence-associated markers, was significantly elevated in Ndufs6 −/− -BM-MSCs compared with BM-MSCs (Fig. 2H). Moreover, western blotting analysis also showed that the protein level of p53 and p21 was much higher than BM-MSCs (Fig. 2I). These results suggest that Ndufs6 may play a critical role in regulating cellular senescence.

Knockdown of Ndufs6 accelerates cellular senescence in BM-MSCs
To further explore the relationship between Ndufs6 and cellular senescence, we performed experiments with lossand gain-of-function of Ndufs6 in MSCs. First, we used Ndufs6-siRNA to treat BM-MSCs. RT-PCR results showed that the mRNA level of Ndufs6 was markedly reduced, whereas the mRNA level of p53 and p21 was significantly enhanced in Ndufs6-siRNA-treated BM-MSCs compared with BM-MSCs (Fig. 3A). Similarly, western blotting also showed that the protein level of Ndufs6 was markedly decreased in Ndufs6-siRNA-treated BM-MSCs compared with BM-MSCs (Fig. 3B). Further, the protein level of p53 and p21 was greatly upregulated compared with BM-MSCs (Fig. 3B). We then performed SA-β-gal staining to detect senescent MSCs among the two groups. Ndufs6-siRNA-treated BM-MSCs exhibited increased β-gal positivity compared with BM-MSCs (Fig. 3C, D). Moreover, Ki-67 staining demonstrated that the number of Ki-67-positive cells was significantly reduced in Ndufs6-siRNA-treated BM-MSCs (Fig. 3E, F).
Overall, these findings showed that knockdown of Ndufs6 enhanced cellular senescence of BM-MSCs.

Discussion
There are several major findings of this study. First, we have determined that a mitochondrial complex I accessory subunit, Ndufs6, is notably downregulated in both human and mice-aged MSCs. Second, we have verified that downregulation of Ndufs6 is not only a consequence  Over the past decades, the senescence of MSCs has attracted huge attention due to their critical role in regenerative medicine. Senescent MSCs exhibit decrease differentiation capacity, anti-inflammation capacity, and reduced immunomodulatory activity 30,[32][33][34] . Understanding the potential mechanisms that underlie MSC senescence is essential to the development of novel strategies to improve MSC quality, especially those from aged donors. It has been well documented that genetic alterations to MSCs contributes to cellular senescence. UBC knockdown induces senescence of MSCs and impairs their functions, whereas UBC overexpression ameliorates MSC senescence and improves their proliferative activity 35 . Conditional knockdown of Foxp1 in MSCs leads to premature senescence as evidenced by impaired bone mass and reduced MSC self-renewal capacity 36 . Knockout of WT p53-inducible phosphatase-1 in BM-MSCs results in premature characteristics of senescence, including a typical senescent morphology, increased β-gal activity, and decreased proliferative capacity 37 . Given the known evidence that a genetic defect is closely linked to cellular senescence, we sought to determine whether Ndufs6 mediates senescence in MSCs. We found that the differentiation capacity was greatly reduced in Ndufs6 −/− -BM-MSCs compared with WT MSCs. Moreover, Ndufs6 −/− -BM-MSCs exhibited enhanced β-gal activity and reduced proliferative ability, suggesting that Ndufs6 may be involved in cellular senescence. Nonetheless it has yet to be shown how Ndufs6 mediates MSC senescence.
Complex I is the major carrier of electron and subsequent superoxide production in mitochondrial membrane 38 . Complex I is the most vulnerable enzyme and the first to be damaged in many mitochondrial disorders. It has also been suggested that complex I malfunction is involved in the pathogenesis of diabetes and ageing. Nonetheless it is composed of at least 45 different subunits whose primary structures have only just been determined 39 . Understanding how the subunits of complex I contribute to stem cell ageing is just the beginning. Many subunits of complex I have been referred to as accessory (or supernumerary) subunits. Initially, these accessory subunits have been considered to be nonessential to the structure and function of mitochondrial complex I. Nonetheless recent research has questioned whether these "accessory subunits" are really accessory. Some of these subunits including Ndufs6 are absolutely essential for complex I function and loss or mutations of these subunits can lead to complex I defects.
Mutations of Ndufa9, a Q-module subunit for complex I, cause mitochondrial complex I assembly defect, leading to a severe and fatal neonatal phenotype 40 . Mutations in Ndufb10, an accessory subunit of complex I, results into complex I deficiency, thus impairing the oxidation 41 . Conditional ablation of Ndufa5 induces partial complex I deficiency, leading to lethargy and loss of motor skills in mice 42 . In the current study, following bioinformatics analysis that Ndufs6 downregulation is a hallmark of BM-MSCs ageing, we verified that Ndufs6 is also an inducer of MSC ageing, closely associated with downregulated Ndufs6-provoked complex I deficiency.
Accumulating evidence has demonstrated that complex I deficiency can cause increased ROS generation, leading to several devastating mitochondrial disorders. Leman et al. analyzed several fibroblast cell lines isolated from patients with inherited complex I deficiency and found that ROS was greatly enhanced 43 . Knockout of Ndufs4-induced complex I deficiency triggers ROS generation and lipid droplet accumulation in glia, and subsequent neurodegenerative disease in mice 44 . Accordingly, we also observed elevated ROS in Ndufs6 −/− -BM-MSCs compared with BM-MSCs. Further, Ndufs6-siRNA-treated BM-MSCs exhibited complex I deficiency and enhanced ROS, suggesting that complex I deficiency contributes to ROS generation. It is well established that ROS can trigger cellular senescence via regulation of multiple pathways. Inflammatory cytokine TNF-α can activate the ROS/NF-κB pathway to induce nucleus pulposus cell senescence 45 . Iron overload-induced ROS accumulation induces BM-MSC senescence via upregulation of the protein expression of p53, ERK, and p38 (ref. 46  p21 signaling pathway 47 . In this study, we used Mito-TEMPO to treat Ndufs6 −/− -BM-MSCs, and found that cellular senescence was greatly reduced along with a decreased level of ROS and protein level of p53/p21, indicating that the ROS/p53/p21 signaling pathway is involved in Ndufs6 knockout-induced cellular senescence of MSCs.
In summary, based on the above findings, we conclude that depressed Ndufs6 and MSC ageing interact in a manner of reciprocal causation. Ndufs6 defect impairs complex I function and activates the ROS/p53/ p21 signaling pathway, leading to cellular senescence of MSCs. Our study provides novel insight to precisely target the mitochondria subunit to prevent senescence of MSCs.