Human bone marrow collection was approved by the Seoul St. Mary’s Hospital Institutional Review Board (IRB), and written informed consent was obtained from a healthy donor (IRB number #KC15CSSE0336). Isolated MSCs were characterized for several cell surface antigens by flow cytometry (FACS Calibur; BD Biosciences). The MSCs were positive for CD29, CD44, CD73, CD90, and CD105, but negative for CD14, CD31, CD104, and HLA- DR.

In this study, MSCs cultured at three different cell densities were used for all experimental analyses: low-density cultured MSCs [LD-MSCs (LD), 50 cells/cm²], middle- density cultured MSCs [MD-MSCs (MD), 1,000 cells/cm²], and high-density cultured MSCs [HD-MSCs (HD), 4,000 cells/cm²]. LD and MD were passaged at intervals of 7∼10 days, whereas HD were passaged every 3∼5 days and cultured for long term until late passage. To observe the effects of ROS scavengers on MSC proliferation, cells were treated with ascorbic acid (AA; 25 μg/ml). Cell prolife-ration was determined by cell counting. Proliferation fold increase (FI), population doubling time (PDT), and population doubling level (PDL) were calculated by counting initial seeding numbers, final cell numbers obtained from each passage, and culture period per passage.

Adipogenic and osteogenic differentiations were performed using the protocol described previously (18). For quantitative analysis, Oil red O-stained cells were completely dried and de-stained using 100% isopropanol. The absorbance of the de-stained solution was measured at 500 nm. Similarly, Alizarin Red S-stained cells were completely dried, de-stained with 10% acetic acid, and the absorbance was measured at 405 nm.

Cellular senescence was determined using a β-galacto-sidase staining kit (Cell Biolabs, San Diego, CA, USA), following the manufacturer's instructions. Briefly, cells were washed with phosphate-buffered saline, fixed with 4% formaldehyde, and incubated overnight at 37℃ with β-galactosidase staining solution. Positive (blue) cells were observed under a light microscope. For quantitative evaluation of senescence, 15,000 cells were assayed using the Quantitative Cellular Senescence Assay Kit (Cell Biolabs, San Diego, CA, USA). Results were expressed as the relative fluorescence unit (RFU).

Intracellular ROS were measured using DCFDA Cellular ROS Detection Assay Kit (ab113852, Abcam, Cambridge, UK), following the manufacturer’s instructions. Cells (2×10⁴) were seeded in each well of a 96-well plate and fluorescence was measured using a fluorescence plate reader (Biotek, Synergy H1; excitation=485 nm and emission=535 nm) to quantitate ROS production.

To determine oxidative DNA damage, 8-hydroxy-2’-deo-xyguanosine (8-OHdG) was measured using OxiSelect DNA Oxidative Damage ELISA Kit (STA-320, Cell Biolabs), following the manufacturer’s instructions. The absorbance was measured with a microplate reader at 450 nm.

Telomerase activity was determined using Telomerase PCR ELISA Kit (Roche Diagnostics Scandinavia AB, Stockholm, Sweden), following the manufacturer’s instruc-tions.

Total RNA was isolated using TRIzol (Invitrogen, Paisley, Scotland) according to the manufacturer’s instructions. The PCR reactions were performed on an Applied Biosystems StepOnePlus^TM PCR machine using 5 μl SYBR^Ⓡ Green PCR Master Mix (Life Technologies, Grand Island, NY, USA). Real-time reverse transcriptase polymerase chain reaction (RT-PCR) was used to quantify by specific marker genes. The primers used in this study are as follows. P15, p16, PCNA (proliferating cell nuclear antigen), PPARγ (peroxisome proliferator-activated receptor gamma), FABP4 (fatty acid-binding protein 4), ALPP (alkaline phosphatase), SPP1 (secreted phosphoprotein 1). All primers were obtained from Qiagen (Hilden, Germany). The delta-delta-Ct method was employed to determine the relative gene expression normalized to the endogenous controls glyceraldehyde-3-phosphate (GAPDH).

All results are presented as mean±standard deviation (SD). The experimental data were analyzed by one-way analysis of variance (ANOVA). A p-value of <0.05 was considered significant. Statistical tests were performed using GraphPad Prism 6.0 software (Graphpad; California, US).

To determine whether culturing conditions affect aging-associated changes, we observed morphological changes in MSCs by culturing them at different cell densities for long term. Three different cell densities were selected: 50 (LD), 1,000 (MD), and 4,000 (HD) cells/cm². Morphological changes in MSCs were observed from passage 5 to passage 15. At passage 15, HD showed flattened and enlarged shapes, which is the normally observed morphology of senescent cells. On the contrary, such changes were observed neither in LD nor in MD (Fig. 1a). To observe changes in cell size and granularity, we performed flow cytometry. At passage 5, no density-dependent difference in cell size was observed. However, at passages 10 and 15, cell size was significantly increased in HD compared to that in LD and MD. There was no difference in cell sizes between MD and LD (Fig. 1b). On the contrary, we found that granularity increased significantly with increasing cell density at all passages (Fig. 1c). These observations suggest that morphological changes in MSCs can be modulated by lowering cell density in long-term cultures.

To confirm whether the morphological changes shown in Fig. 1 were aging-dependent, cellular senescence was analyzed using senescence-associated β-galactosidase (SA-β-gal) staining. Cell density-dependent morphological changes in MSCs were not observed at passages 5 and 10. However, at passage 15, SA-β-gal positive cells were increased in HD compared to those in LD and MD (Fig. 2a). Quantitative analysis of fluorescence intensity in SA-β-gal positive cells showed the most prominent induction in HD at passage 15 (Fig. 2b).

Next, we performed experiments to observe differences in the expression of senescence-related genes, such as p15, p16, and proliferating cell nuclear antigen (PCNA), in MSCs cultured at three different cell densities. At passage 15, the expression of p15 and p16 was compared among LD, MD, and HD. We observed that the expression levels of both genes were decreased by lowering the cell densities. However, PCNA, a marker for proliferation, was decreased with increasing cell density (Fig. 2c). These results imply that senescence of MSCs can be delayed under low-density conditions.

MSC senescence is reportedly related to a decrease in telomerase activity (19-21). Thus, we analyzed telomerase activities in MSCs cultured at the three cell densities. Here, we found no significant difference in telomerase activity among different conditions of passage and cell density (Fig. 2d).

As the proliferation ability of MSCs gradually decreases upon senescence, measuring proliferation ability is one of the tools for estimating senescence (22, 23). We analyzed proliferation potentials of cells at the abovementioned conditions of passage and cell density. FI, PDT, and PDL were calculated from initial cell number, final cell number, and culture period.

Throughout the passage stages, PDT tended to increase while both FI and PDL tended to decrease with increasing passage numbers at all densities. We noted that the decrease in proliferation ability during long-term culture was slower at low cell densities (Table 1). These results indicate that the lifespan of MSCs can be extended in vitro in long-term cultures with low density.

We analyzed differentiation potentials to check whether they were affected by varying cell densities. Both adipogenic and osteogenic differentiation potentials were compared among MSCs cultured at the abovementioned conditions. Although adipogenic potential, determined by Oil red O staining at passage 15 was decreased when compared to that at early passages, the difference was not significant across different cell densities (Fig. 3a and 3b). Adiopogenic differentiation potential of MSCs was also analyzed by specific marker expressions, such as PPARγ and FABP4. No significant difference across different cell densities was observed (Fig. 3c and 3d).

However, osteogenic differentiation potential, determined by Alizarin Red-S staining with mineralization quantitation was significantly decreased in a passage-dependent manner (Fig. 3e and 3f). There was no difference in differentiation potential regarding varying cell densities at passages 5 and 10. At passage 15, however, osteogenic differentiation potential was almost completely lost in HD but was maintained in LD and MD. It is well correlated with the reduction of osteogenic differentiation markers expressions, ALPP and SPP1, at passage 15, compared to those at passages 5 and 10. (Fig. 3g and 3h). These results indicate that osteogenic potential of MSCs can be maintained under low cell densities even after prolonged passage.

As cells become senescent, they show a direct correlation between decrease in stem cell function and accumulation of DNA damage (24-27). In particular, ROS induce oxidative DNA damage and decrease stem cell function, leading to MSC aging (25, 27). Accordingly, total cellular ROS were determined to identify whether ROS production was influenced by cell density. At passage 5, there were no significant differences in ROS production in LD, MD, and HD. On the contrary, at passages 10 and 15, ROS production was significantly decreased when cell density was lowered. Specifically, ROS production was significantly increased in HD compared to that in LD or MD (Fig. 4a). To confirm the above results, 8-OHdG, a DNA damage marker (28), was measured to assess ROS-dependent DNA damage. In our experiments, 8-OHdG production did not show any difference among the three cell densities at passages 5 and 10; however, at passages 15, it increased significantly with increase in cell density (Fig. 4b). These results indicate that DNA damage was induced by ROS at high cell density, resulting in increase in cell senescence.

To determine whether MSC proliferation was influenced by the ROS generated in high cell density cultures, HD were treated with AA, a well-known ROS scavenger. At passages 11∼15, AA was added to MSCs and their proliferative abilities were compared with those of untreated MSCs. At passage 11 of HD, FI values were 2.6±0.32 and 3.8±0.51 for AA-treated and untreated MSCs, respectively. At passage 15, the corresponding values were 1.6±0.14 and 2.5±0.27 (Fig. 4c). In addition, a decrease in ROS production due to AA treatment was confirmed at passage 15 (Fig. 4d). Overall, MSC proliferation ability was increased by ROS inhibition and it was maintained even after late passage. This suggests that the lifespan of MSCs can be improved via scavenging ROS. To confirm the above results, we examined the expressions of senescence-related genes, such as p15, p16, and PCNA in MSCs cultured at passage 15 of HD in the presence or absence of AA. According to scavenging ROS by treatment of AA, the expression levels of both p15 and p16 were decreased, but the expression of a proliferation marker, PCNA, was increased. (Fig. 4e). Moreover, the production of 8-OHdG, DNA damage marker, was significantly reduced by AA treatment, scavenging ROS (Fig. 4f). These results imply that senescence of MSCs can be delayed via scavenging ROS under high-density conditions.