Role of p16INK4a and BMI-1 in oxidative stress-induced premature senescence in human dental pulp stem cells.

Human dental pulp stem cells (hDPSCs) are a source for cell therapy. Before implantation, an in vitro expansion step is necessary, with the inconvenience that hDPSCs undergo senescence following a certain number of passages, loosing their stemness properties. Long-term in vitro culture of hDPSCs at 21% (ambient oxygen tension) compared with 3-6% oxygen tension (physiological oxygen tension) caused an oxidative stress-related premature senescence, as evidenced by increased β-galactosidase activity and increased lysil oxidase expression, which is mediated by p16INK4a pathway. Furthermore, hDPSCs cultured at 21% oxygen tension underwent a downregulation of OCT4, SOX2, KLF4 and c-MYC factors, which was recued by BMI-1 silencing. Thus, p16INK4a and BMI-1 might play a role in the oxidative stress-associated premature senescence. We show that it is important for clinical applications to culture cells at physiological pO2 to retain their stemness characteristics and to delay senescence.

Introduction

Organismal aging is associated with a loss of the homeostasis. One of the elements that contribute to this deterioration is increased cell senescence [1]. Hayflick originally described senescence as a permanent cell cycle arrest due to the limited replicative potential of cultured human fibroblasts [2]. Senescence after a number of cell doublings during in vitro culture is inevitable under current culture conditions, resulting in cellular phenotypic changes and growth arrest [3], [4], [5]. This observation of cellular senescence has been extrapolated to somatic stem cells in vivo and might reflect the aging process of the whole organism [4].

In vitro cellular senescence refers to both replicative and premature senescence [6]. Premature or accelerated senescence can be induced by stress signals, such as activation of oncogenes, strong mitogenic signals, and/or reactive oxygen species (ROS) levels. As we previously reported, oxidative stress is responsible for the low proliferation rate under ambient oxygen tension (21% pO2) through p38, p21, and NRF-2 activation [7]. Cell culture-inherent oxidative stress can cause critical telomere attrition, accumulation of DNA damage and de-repression of the INK4/ARF locus, leading to stress-induced premature senescence (SIPS) [8]. Lysyl oxidase enzymes (LOXL1 and LOXL2) have been also shown to be oxidative stress-sensitive. Among other roles, such as cell motility and cell adhesion, they have been related to cell growth control and cellular senescence [9].

To maintain their replicative and self-renewing potential stem cells have in place mechanisms to repress activation of cell death pathways. The Polycomb-group transcriptional repressor BMI-1 has emerged as a key regulator in several cellular processes including stem cell self-renewal and cancer cell proliferation. BMI-1 was first identified in 1991 as a frequent target of Moloney virus insertion in virally accelerated B-lymphoid tumours of E mu-myc transgenic mice [10]. Through repression of target gene expression in a lineage and context- dependent manner, BMI-1 regulates a myriad of cellular processes critical for cell growth, cell fate decision, development, senescence, aging, DNA damage repair, apoptosis, and self-renewal of stem cells [11]. The most studied and validated BMI-1 target is the INK/ARF locus, which encodes two structurally distinct proteins, p16INK4a and p14ARF, both of which restrict cellular proliferation in response to aberrant mitogenic signalling. Thus, collectively BMI-1 regulates p53/pRb axis through repression of the INK/ARF locus, which has been described as the principle barrier to the initiation and maintenance of neoplastic transformation [12]. BMI-1 is known to repress the INK/ARF locus expression, which encodes two structurally distinct proteins, p16INK4a and p14ARF, both of which restrict cellular proliferation in response to aberrant mitogenic signalling [12]. BMI-1 has been implicated in the modulation of self-renewal in several types of stem cells, including hematopoietic [13], neural [14], and mammary [15].

Self-renewal of stem cells is critical for their persistence through life, however the capacity to maintain this characteristic declines with age [16], [17]. Pluripotency genes, OCT4, SOX2, KLF4 and C-MYC (OSKM) [18], are expressed in both pluripotent and adult stem cells, such as mesenchymal stem cells (MSCs) and are down-regulated upon long-term in vitro expansion and differentiation [19].

Our main purpose was to analyse the role of p16INK4a and BMI-1 in oxidative stress-induced senescence in long term human dental pulp stem cells (hDPSCs) cultures. In this study we demonstrate that non-physiological in vitro cell culture conditions at 21% pO2 induces premature senescence of hDPSCs, which is mediated by downregulation of BMI-1 leading to an activation of p16INK4a pathway. By restoring BMI-1 levels, we were able to rescue SOX2 and OCT4 expression under oxidative stress conditions, reflecting that BMI-1 is not only involved in stem cell self-renewal, but also in stemness maintenance. In summary, we show that oxygen tension is critical when culturing hDPSCs. Ambient oxygen tension (21% pO2) induces premature hDPSCs senescence compared with physiological oxygen tension (3% pO2) due to activation of p16INK4a pathway. Moreover, this is accompanied by a BMI-1-dependent stemness potential loss. This is of importance in regenerative medicine and also in stem cell banking for clinical use.

Material and methods

Dental pulp stem cells isolation and culture

Intact third molars were collected from men and women (aged from 15 to 35 years old). All patients were informed and agreed freely to participate and signed the informed consent by contributing the extracted tooth, which was always extracted for reasons independent of this study. The study was approved by the institutional review board of the University of Valencia. Cells cultured from dental pulps did not exhibit any clinical and/or radiological sign or symptom of inflammation and/or infection. To isolate the cells, the pulps were firstly fragmented by trituration, then chemically digested with 2 mg/mL EDTA in Krebs-Henseleit buffer, and finally digested with a combination of type I collagenase and type II dispase at a final concentration of 4 mg/mL during 30 min in a humid incubator at 37 °C, 5% CO2, and 3% pO2. Digested pulp fragments were centrifuged at 1000g for 2 min, and the precipitate was resuspended and seeded in culture flasks with complete DMEM (Dulbecco's Eagle Modified Medium with low glucose supplement 1g/L, 10% heat-inactivated FBS and 1% antibiotic) under the same conditions of temperature and oxygen pressure.

After the first passage, hDPSCs were divided in two groups: one group was moved to a humid incubator with an oxygen pressure of 21%, while the other group was kept in the same incubator used for the isolation at 3% oxygen tension. Cells were then cultured for 7 months.

All reagents were purchased from Gibco, Invitrogen.

siRNA transfection for BMI-1 knockdown

Young hDPSCs (5 passages) were seeded in a six well culture plate, at 2×105 cells per well in 2 mL antibiotic-free normal growth medium supplemented with 10% FBS and incubated until the cells were 60–80% confluent. For each transfection, 0.8 mL of siRNA Transfection Medium was added to each tube containing the siRNA Transfection Reagent mixture (Solution A+Solution B) following manufacturer's instructions (Santa Cruz Biotechnologies). The mixture was overlayed onto the washed cells prior to a 6 h incubation at 37 °C in a CO2 incubator. 1 mL of normal growth medium containing 2 times the normal serum and antibiotics concentration was added post-incubation without removing the transfection mixture. 24 h later, the medium was replaced with fresh 1x normal growth medium and cells were assayed using the appropriate protocol 24 h after the addition of fresh medium in the step above.

Reactive oxygen species and mitochondrial membrane potential determination by flow cytometry

Cells were washed twice with warm PBS and treated with trypsin (Gibco, Invitrogen) and then resuspended in serum-free DMEM containing 1 g/L glucose. To detect intracellular peroxide levels, cells were stained with DHR123 (dihydrorhodamine-123, Thermo Fisher Scientific) at a final concentration of 1 μg/mL. Cells were then incubated for 30 min at 37 °C in the dark. Mitochondrial membrane potential was measured after cell staining with 1 μg/mL TMRM (tetramethylrhodamine methyl ester), Thermo Fisher Scientific for 30 min at 37 °C in the dark. After incubation, values were read by FACS-Verse flow cytometry until 20,000 events were recorded.

Lipid peroxidation measured using high-performance liquid chromatography

hDPSCs lipid peroxidation was determined as malondialdehyde (MDA) levels, which were detected using high-performance liquid chromatography (HPLC) as an MDA-thiobarbituric acid (TBA) adduct following a method described previously [20]. This method is based on the hydrolysis of lipoperoxides and subsequent formation of an adduct between TBA and MDA (TBA-MDA2). This adduct was detected using HPLC in reverse phase and quantified at 532 nm. The chromatographic technique was performed under isocratic conditions, the mobile phase being a mixture of monopotassium phosphate 50 mM (pH 6.8) and acetonitrile (70:30).

Protein carbonylation measured using high-performance liquid chromatography

The carbonyl groups in the protein side chains were derivatized to 2,4-dinitrophenylhydrazone by reacting with 2,4-dinitrophenylhydrazine so that they could be detected using Western blotting using specific antibodies. Oxidative modification of total proteins was assessed by immunoblot detection of protein carbonyl groups using the OxyBlot Protein Oxidation Detection kit in accordance with the manufacturer's instructions (Merk Millipore). The procedure to quantify total protein carbonyls using the OxyBlot kit was densitometry of the Oxyblot and Ponceau staining followed by finding the ratio between the total density in the Oxyblot and the Ponceau.

RNA extraction and RT-qPCR analysis

Total RNA was isolated from hDPSCs by using TRIzol reagent (Invitrogen), according to the manufacturer's instruction. RNA was quantified by measuring the absorbance at 260 nm. The purity of the RNA preparations was assessed by the 260/280 ratio. cDNA was synthesized from 0.5 μg total RNA using a MultiScribe reverse transcriptase (RT) system kit of Applied Biosystems (High-Capacity cDNA Reverse Transcription Kits). The reaction was incubated as recommended by the manufacturer, for 10 min at 25 °C, followed by 120 min at 37 °C, and then for 5 min at 85 °C, and finally cooled to 4 °C to collect the cDNA and then stored at −20 °C prior to the real-time PCR assay.

The quantitative PCR was performed using the detection system 7900HT Fast Real-Time PCR System (Applied Biosystems). Target and control were run in separate wells following procedure: 10 min at 95 °C and then 35 cycles of denaturation at 95 °C for 15 s and annealing and extension at 62 °C for 1 min per cycle. All experiments were repeated at least three times for each sample.

Gene-specific primer pairs and probes for BMI-1 (3’-CCAGGGCTTTTCAAAAATGA-5’ and 5’-GCATCACAGTCATTGCTGCT-3’), OCT4 (3’-GATCCTCGGACCTGGCTAAG-5’ and 5’-GACTCCTGCTTCACCCTCAG-3’), SOX2 (3’-AAAACAGCCCGGACCGCGTC-5’ and 5’-CTCGTCGATGAACGGCCGCT-3’), KLF4 (3’-CCCACATGAAGCGACTTCCC−5’ and 5’-CAGGTCCAGGAGATCGTTGAA−3’), C-MYC (3’-CGCCCTCCTACGTTGCGGTC-5’ and 5’-CGTCGTCCGGGTCGCAGATG-3’), p16INK4a (3’-GGGGGCACCAGAGGCAGT-5’ and 5’-GGTTGTGGCGGGGGCAGTT-3’) and p14ARF (3’-CCCTCGTGCTGATGCTACTG-5’ and 5’-CATCATGACCTGGTCTTCTAGGAA-3’) were assayed together with Maxima SYBR Green/ROX qPCR Master Mix (2X) (Fermentas) and normalized against GAPDH (3’-TGAACGGGAAGCTCACTGG-5’ and 5’-TCCACCACCCTGTTGCTGTA-3’) housekeeping gene. Relative expression was analysed using the standard curve method.

Gene-specific primer pairs and probes for LOXL1 (Hs00935937_m1), LOXL2 (Hs00158757_m1), and TET1 (Hs04189344_g1), were used together with 1x TaqMan® Universal PCR Master Mix (Applied Biosystems) and normalized against GAPDH (Hs00375015_m1). In this case, the expression was calculated according to the 2−ΔΔCt method.

Senescence-associated β-galactosidase staining by flow cytometry

SA-β-Gal staining was performed using FluoReporter® LacZ Kit (Molecular Probes) following manufacturer's instructions. 100 uL of resuspended cells (107 cells/mL) in staining medium were placed into an appropriate flow cytometer tube and treated with 100 uL of prewarmed fluorescein di-β-D-galactopyranoside (FDG) 2 mM working solution for exactly one minute at 37 °C. FDG loading was stopped by adding 1.8 mL ice-cold staining medium containing 1.5 μM propidium iodide. FDG values were read by flow cytometry until 20,000 events were recorded.

Protein analysis using western blotting

Total protein was harvested by lysing the cells in a lysis buffer containing a protease inhibitor cocktail (Roche Products). Protein content was determined by a modified Lowry method [21]. 30 μg of protein from each sample was separated on SDS-12.5% polyacrylamide gels and transferred onto a PVDF membrane (BioRad). Membranes were blocked with 0.01 g/mL BSA in TBS-0.05% Tween 20 (TBS-T) and incubated with the following antibodies: anti-BMI-1 (1:200), anti-Tubulin (1:1000) and anti-Mouse (1:10,000). The protein bands were detected by chemiluminiscence.

Statistical analysis

Quantitative variables are expressed as means and SD. Qualitative data are expressed as total number and percentage. Statistical analysis consisted of Student's t-test for 2 means and ANOVA to compare 2 means with one variation factor. If the n is not the same in all the groups, the comparison of Scheffé was used. All values are means±SD of measurements in at least three different cultures (three replicates each). Significance was defined as *p<0.05, **p<0.01, and ***p<0.001.

Results

Ambient oxygen tension induces oxidative stress in hDPSCs long term culture

hDPSCs cultured under ambient oxygen tension showed significantly higher levels of ROS, lipid oxidation and protein carbonylation than those cultured under physiological oxygen tension, as well as a loss of their mitochondrial membrane potential. These differences were found from early passages and were maintained during long term culture, showing that hDPSCs at passage 15 at 3% pO2 were less damaged than their counterparts at 21% pO2 (Fig. 1A–D). According to this, we found that hDPSCs cultured at 21% pO2 showed significantly increased expression of the antioxidant enzymes manganese superoxide dismutase (MnSOD), catalase and glutathione peroxidase (GPx) (Fig. 2). This increase of antioxidant shield might be due to an attempt to hormetically adapt the metabolism to the increased oxidative stress at 21% pO2.

Fig. 1

Oxidative stress related parameters in hDPSCs cultured at 21% or 3% oxygen tension along passages. (A) ROS levels measured by dihydrorhodamine-123 (DHR123), (B) Mitochondrial membrane potential levels measured by tetramethylrhodamine methyl ester (TMRM), (C) Lipid oxidation levels measured by malondialdehyde (MDA) and (D) protein oxidation levels. The data are shown as means±SD (n=5). The statistical significance is expressed as *p<0.05; ***p<0.001 versus 3% pO2.

Fig. 2

Antioxidant genes expression in hDPSCs cultured at 21% or 3% oxygen tension. Manganese superoxide dismutase (MnSOD) levels, glutathione peroxidase (GPx) levels and catalase levels. The data are shown as means±SD (n=5). The statistical significance is expressed as ***p<0.001 versus 3% pO2.

Oxidative stress induces premature senescence (SIPS) under ambient oxygen tension during long term culture of hDPSCs

hDPSCs were cultured under ambient (21% pO2) or physiological oxygen tension (3% pO2) serially until the cells were exhausted. hDPSCs cultured under ambient oxygen tension only reached 15 passages, while those cultured under physiological oxygen tension, were able to achieve 25 passages (Fig. 3A). Moreover, we could observe that hDPSCs cultured at 21% pO2 began to show enlarged and flattened shapes around passage 15, while hDPSCs cultured at 3% pO2 at passage 25 were still morphologically thinner (Fig. S1).

Fig. 3

Oxidative stress induces premature senescence in hDPSCs during long term culture at 21% oxygen tension. (A) Number of passages reached (upper pannel) and survival curve (lower pannel), (B) β-galactosidase activity measured by fluorescein di-β-D-galactopyranoside (FDG) load, (C) LOXL1 and (D) LOXL2 relative mRNA expression levels. The data are shown as means±SD (n=5). The statistical significance is expressed as ***p<0.001 versus 3% pO2.

β-Galactosidase staining is one of the most commonly used markers of senescence. hDPSCs cultured at 21% pO2 had significantly higher levels of β-galactosidase activity at any passage analysed (5, 10, 15) when compared with 3% pO2. This difference increased along passages. Moreover, hDPSCs cultured at 3% pO2 were even less senescent at passage 25 than hDPSCs cultured at 21% pO2 at passage 15 (Fig. 3B).

LOXL1 and LOXL2 are lysyl oxidase enzymes involved in cell cycle regulation. We show in Fig. 3C and D, that hDPSCs cultured at 3% pO2 had lower mRNA levels of both enzymes in comparison to those cultured at 21% pO2. Furthermore, as passaging number increased, so did LOXL1 and LOXL2 mRNA levels when cultured at 21% pO2 but not at 3% pO2.

Therefore, hDPSCs cultured at 21% pO2 undergo premature senescence compared to those cells cultured at 3% pO2.

Oxidative stress-induced premature senescence (SIPS) is mediated by p16INK4a pathway

We analysed the mRNA expression pattern of both p14 ARF and p16INK4a in hDPSCs long term culture at 3% and 21% pO2. p14ARF mRNA expression levels were not affected by long-term culture or by oxygen concentration. However, p16INK4a mRNA levels revealed an expression pattern very similar to β-galactosidase activity levels, i.e., there was an increase of its mRNA expression along the passages and it was always higher at 21% pO2 (Fig. 4A and B).

Fig. 4

Oxidative stress-induced premature senescence. Correlation with p16and p14. (A) p14ARF mRNA levels, (B) p16INK4a mRNA levels and (C) p16INK4a mRNA levels when treatment with 50 μM Trolox. The data are shown as means ±SD (n=5). The statistical significance is expressed as ***p<0.001 versus 3% pO2 and ###p<0.001 versus 21% pO2+Trolox.

In order to demonstrate that oxidative stress was mediating the p16INK4a induced premature senescence at 21% pO2, we cultured hDPSCs at this oxygen tension with 50 μM Trolox, a hydrosoluble antioxidant analogue of Vitamin E. We found that Trolox reversed the effect of ambient oxygen tension on p16INK4a mRNA expression (Fig. 4C). Therefore, oxidative stress increases p16INK4a expression, which in turn accelerates senescence in hDPSCs cultured under ambient oxygen tension.

Loss of stemness under ambient oxygen tension during long term culture of hDPSCs

Oxidative stress can affect stemness, therefore we measured the mRNA expression levels of OSKM transcription factors. SOX2 and OCT4 are implicated in pluripotency induction, while KLF4 and C-MYC are involved in pluripotency maintenance. Our results show that, comparing hDPSCs cultured at 3% pO2 vs 21% pO2, SOX2 and OCT4 mRNA expression was significantly higher at passage 5, and KLF4 and C-MYC mRNA expression was significantly higher at passage 15 (Fig. 5A). TET1 is one member of a family of enzymes that alter the methylation status of DNA. They are involved in stem cell self-renewal, proliferation and differentiation. We observed that TET1 mRNA levels were downregulated in hDPSCs cultured at 21% pO2, in comparison to 3% pO2 (Fig. 5B).

Fig. 5

Pluripotency markers in hDPSCs cultured at 21% vs 3% oxygen tension along passages. (A) OCT4, SOX2, KLF4 and C-MYC mRNA levels and (B) TET1 mRNA levels relative expressions. The data are shown as means±SD (n=5). The statistical significance is expressed as *p<0.05; **p<0.01; ***p<0.001 versus 3% pO2.

Therefore, we show that culturing hDPSCs at 3% pO2 increases OSKM transcription factors compared to 21% pO2.

BMI-1 can rescue SOX2 and OCT4 expression under ambient oxygen tension

BMI-1 protein levels were significantly higher in hDPSCs cultured at 21% pO2 at passage 5 when compared to 3% pO2. However, they decreased very rapidly with passages and were significantly lower at passage 15, in comparison to those cultured at 3% pO2 (Fig. 6A and B).

Fig. 6

expression level in hDPSCs cultured at 21% vs 3% oxygen tension along passages. (A) BMI-1 protein levels, and (B) representative western-blot images of BMI-1 protein levels. The data are shown as means ±SD (n=5). The statistical significance is expressed as *p<0.05; ***p<0.001 versus 3% pO2.

We used siRNA transfection in order to obtain a mild BMI-1 knockdown so that hDPSCs at passage 5 cultured at 3% or 21% pO2 had the same BMI-1 protein expression level (Fig. S2). BMI-1 knockdown did not have any effect on p16INK4a expression (data not shown), however it restored SOX2 and OCT4 mRNA levels in hDPSCs cultured at 21% pO2 (Fig. 7). This reflects a relationship between BMI-1 and pluripotency transcription factors.

Fig. 7

knockdown effect onandexpression in hDPSCs cultured at 21% vs 3% oxygen tension. (A) SOX2 mRNA expression levels and (B) OCT4 mRNA expression levels. The data are shown as means±SD (n=3). *p<0.05; **p<0,01; ***p<0.001.

Discussion

hDPSCs normally reside in low oxygen concentrations. In mammals including humans, hDPSCs are located in perivascular niches close to the vascular structure in almost all tissues [22], [23], [24]. By the time oxygen reaches the organs and tissues, oxygen concentration drops to 2–9%, with a mean of 3% in the dental pulp tissue [25], [26]. Despite this fact, it is still common to culture cells at high non- physiological 21% pO2. Here we demonstrate that long term culture of hDPSCs in a physiological oxygen tension (3% pO2) has beneficial effects on both cellular senescence and stemness potential maintenance in comparison to culturing cells under ambient oxygen tension.

In the present study, we show that reduction of the pO2 level led to decreased intracellular oxidative stress and cellular components damage during hDPSCs long term culture. This was accompanied by reduced ROS levels, less protein and lipid damage, as well as a better conservation of the mitochondrial membrane potential. In accordance to this, it has been demonstrated that high concentrations of oxygen can cause oxidative stress via production of reactive oxygen species (ROS) and free radicals that damage lipids, proteins and DNA [27]. In this work, we also linked oxygen tension to altered mRNA expression of MnSOD, CAT and GPx. The increment of antioxidant enzyme activities indicated that cellular anti-oxidative system was triggered to resist oxidative damage. Fan and colleagues already demonstrated that higher oxygen concentrations resulted in more H2O2 generation in human cells, which implied that the increase of H2O2 levels could enhance transcription of MnSOD, CAT and GPx [28].

As passage number increased, cells cultured under ambient oxygen tension began to show flattened or lengthened shapes, and debris in the culture medium increased. These morphological changes, have already been described as a characteristic of senescence [29], [30]. In addition to this, hDPSCs cultured at 21% oxygen tension showed higher levels of senescence related β-galactosidase activity as well as p16INK4a expression. This behaviour in long term in vitro culture leads to senescence and is collectively referred to as oxidative stress induced premature senescence (SIPS) [31]. Lysyl oxidase activity has been shown to increase under oxidative stress conditions [32]. hDPSCs cultured at 21% pO2 have higher mRNA expression levels of both LOXL1 and LOXL2. Moreover, it has been described that an increased level of these enzymes may contribute to escape from cellular senescence [9]. This could be a defensive mechanism of hDPSCs cultured at 21% pO2 to escape from oxidative stress-induced senescence. In addition, lysyl oxidases participate in the carbonylation of several proteins, and further increase the H2O2 levels as subproduct of this reaction [33]. These results reinforce the observations of Fan and colleagues [28] and also may contribute to the induction of the antioxidant shield in hDPSCs incubated at 21% pO2, as we described in this investigation.

The p16INK4a/pRb and p14 ARF/p21/p53 cell cycle inhibitory pathways represent two important pathways controlling proliferation, and their inactivation can extend the limited division number of mitotic cells in culture [34]. Given the role of p16INK4a in cell cycle regulation and the recent implication of oxidative stress in stem cell senescence, we observed a potential link between ROS and p16INK4a regulation. Our results show that hDPSCs cultured at 21% pO2 have higher levels of ROS, as well as increasing p16INK4a expression, suggesting that they are approaching senescence. In fact, stress signals such as ROS stimulate the activation of p16INK4a transcription and play important roles in initiation, as well as maintenance, of cellular senescence [35], [36], [37]. Ito and colleagues described, both in vitro and in vivo, that activation of p16INK4a/pRb gene product pathway in response to elevated ROS led to the failure of hematopoietic stem cells (HSCs) function, and that treatment with antioxidant agents restored the constitutive capacity of HSCs, resulting in the prevention of bone marrow failure [38], [39]. These results support our data that treatment with 50 μM Trolox can rescue p16INK4a levels in hDPSCs long term culture at 21% pO2.

Interestingly, p16INK4a and p14ARF are both encoded by a single locus; however, ROS specifically affects p16INK4a but not p14ARF. It may be that the p16INK4a pathway is of particular importance in the senescence of stem cells [40], [41] as it is considered to be a robust biomarker for cellular senescence, and at the forefront of cell cycle inhibition as it binds specifically to the CDKs, displacing cyclin-D and thereby arresting cells in G1 phase [42].

Taken together, the fact that hDPSCs cultured at 21% pO2 show increased p16INK4a expression, higher β-galactosidase activity, overexpressed LOXL1 and LOXL2, and senescent like morphology, earlier than those cultured at 3% pO2, means that 21% oxygen induced- oxidative stress causes premature senescence.

It is well known that one of the BMI-1 downstream targets are p16INK4a and p14ARF [43], [44]. p16INK4a contributes to the regulation of cell cycle progression by inhibiting the S phase [45]. BMI-1 is a repressor, so, high levels of BMI-1 should then be followed by a decrease in p16INK4a expression, and subsequently by a hyper proliferation rate like in cancer cells [44]. However, we found that BMI-1 overexpression in hDPSC cultured at 21% pO2 at early passages was not followed by a p16INK4a downregulation. ROS and BMI-1 play an opposite role on p16INK4a regulation. In fact, oxidative stress induced by 21% oxygen tension would be strong enough to counteract BMI-1 downstream effects on the INK/ARF locus, as it has been shown that it can up-regulate p16INK4a expression [46]. As passages succeed, BMI-1 levels plummet because hDPSCs cultured under ambient oxygen tension suffer an accelerated ageing accompanied by a loss of their capacity to face oxidative stress effects.

Although p16INK4a and p14ARF have been shown to be BMI-1 targets in the context of stem cell self-renewal, they do not account for all BMI-1 actions, and other downstream effectors are being sought [47], [48]. Recently, a relationship between BMI-1 and SOX2 has been described [49]. SOX2, as well as OCT4, are transcription factors that are key players in the induction of pluripotency and stemness [18]. In the present study, we confirmed a loss in all OSKM transcription factors expression, as well as a deregulation in TET1 expression, during hDPSCs in vitro long term culture at 21% pO2. TET proteins are dioxygenases that regulate 5 hydroxyl-methylcytosine (5-hmC) levels in genes implicated in self-renewal, proliferation and differentiation [50], [51]. Our results show that TET1 expression levels are influenced by culture oxygen pressure and passaging number. In fact, when hDPSCs undergo long passages, TET1 levels are downregulated, so decreasing the maintenance of stemness in hDPSCs [51], [52]. Koh and colleagues also proposed that OCT4 and SOX2 regulate the expression of TET1 [51]. Interestingly, our results show that the effect of cell passage is greater than oxygen pressure effect, suggesting that OCT4, SOX2 and NANOG are more relevant in stem cell maintenance than TET1. In accordance with this, it has been shown that cells maintained at 21% pO2 expressed significantly less OCT4, SOX2 and NANOG than those cultured at 5% pO2 [53]. Furthermore, it has also been described that physiological oxygen tension inhibits senescence and maintains stem cell properties [54], [55].

Our results show that BMI-1 downregulation can rescue SOX2 and OCT4 levels without affecting p16INK4a expression levels in hDPSCs cultured at 21% pO2. Again, the opposite effect of BMI-1 and oxidative stress plays a role in maintaining p16INK4a expression levels.

As it has also recently been described by Izpisúa and colleagues [56], the in vitro induction of OSKM can ameliorate some cellular markers of ageing, such as cellular senescence. hDPSCs cultured at 3% pO2 are able to maintain the expression of this factors during a larger number of passages, so the preservation of the OSKM factors could play an important role in the delayed onset of cellular phenotypes associated with ageing observed in this cells.

In conclusion, the present study suggests that p16INK4a and BMI-1 are involved in the cellular premature senescence of hDPSCs triggered by oxidative stress. It is important considering this fact when culturing primary culture cells to improve the extrinsic culture environment, in order to retain their stemness properties and to delay the process of senescence prior to clinical application.

Author contributions

CMB, JVA, MI, JSR and JSIC performed experimental work; JG and JLGG, JV and CB directed experimental work, CMB, JV and CB wrote the paper and CB designed research and directed the project.

Funded sources

This work was supported by the Spanish Ministry of Education and Science (MEC) (SAF2010-19498; SAF2013-44663-R; SAF2016-75508-R); “Red Tematica de investigacion cooperativa en envejecimiento y fragilidad” (RETICEF, ISCIII2006-RED13-027 and ISCIII2012-RED-43-029); CIBERFES (ISCIII2016-CIBER); Valencia International Campus of Excellence (VLC/CAMPUS of the University of Valencia) "Conselleria de Educación, Cultura y Deporte" (PROMETEO2010/074 and ACIF2014/165); and European Union (CM1001 and FRAILOMIC-HEALTH.2012.2.1.1-2). ADVANTADGE project/EU). This study has been co-financed by FEDER funds from the European Union.

The authors state no conflicts of interests.