• Open Access

Age-related intrinsic changes in human bone-marrow-derived mesenchymal stem cells and their differentiation to osteoblasts


Julie Glowacki, Department of Orthopedic Surgery, Brigham & Women's Hospital, Harvard Medical School, 75 Francis St., Boston, MA 02115, USA. Tel.: (617) 732-5397; fax: (617) 732 6937; e-mail: jglowacki@rics.bwh.harvard.edu


In vivo and in vitro studies indicate that a subpopulation of human marrow-derived stromal cells (MSCs, also known as mesenchymal stem cells) has potential to differentiate into multiple cell types, including osteoblasts. In this study, we tested the hypothesis that there are intrinsic effects of age in human MSCs (17–90 years). We tested the effect of age on senescence-associated β-galactosidase, proliferation, apoptosis, p53 pathway genes, and osteoblast differentiation in confluent monolayers by alkaline phosphatase activity and osteoblast gene expression analysis. There were fourfold more human bone MSCs (hMSCs) positive for senescence-associated β-galactosidase in samples from older than younger subjects (P < 0.001; n = 17). Doubling time of hMSCs was 1.7-fold longer in cells from the older than the younger subjects, and was positively correlated with age (P = 0.002; n = 19). Novel age-related changes were identified. With age, more cells were apoptotic (P = 0.016; n = 10). Further, there were age-related increases in expression of p53 and its pathway genes, p21 and BAX. Consistent with other experiments, there was a significant age-related decrease in generation of osteoblasts both in the STRO-1+ cells (P = 0.047; n = 8) and in adherent MSCs (P < 0.001; n = 10). In sum, there is an age-dependent decrease in proliferation and osteoblast differentiation, and an increase in senescence-associated β-galactosidase-positive cells and apoptosis in hMSCs. Up-regulation of the p53 pathway with age may have a critical role in mediating the reduction in both proliferation and osteoblastogenesis of hMSCs. These findings support the view that there are intrinsic alterations in human MSCs with aging that may contribute to the process of skeletal aging in humans.


In humans, peak bone mass is attained during the third decade of life. Subsequently, bone mass declines slowly with advancing age (Mautalen & Oliveri, 1999). Although there is currently intense activity to define the process of acute bone loss associated with sex steroid deficiency and development of osteoporosis, there is little information about the mechanism(s) by which the aging process influences bone loss. A better understanding of age-related changes in cells and tissues may offer new approaches to mitigate or avoid loss of bone with aging.

Recent findings suggest that a decline in the numbers or potential of stem cell populations in adult organs may contribute to human aging and age-related disease (Rao & Mattson, 2001; Van Zant & Liang, 2003; Carrington, 2005). Human adult bone-marrow-derived mesenchymal stem cells, or marrow stromal cells (MSCs), have been shown to be precursors of several different cellular lineages, including osteoblast, chondrocyte, myoblast, adipocyte, and fibroblast (Prockop, 1997; Krebsbach et al., 1999; Pittenger et al., 1999; Caterson et al., 2002). It has been shown with mouse marrow cells that there are age-associated differences in osteoblast differentiation and signaling (Kahn et al., 1995; Moerman et al., 2004). Less is known about the effects of age on the capacity of human marrow-derived cells for self-renewal and osteoblast differentiation. Some data from colony assays with marrow aspirates conflict about the effects of age on osteoblast potential of adult human marrow (Majors et al., 1997; Oreffo et al., 1998a,b; D’Ippolito et al., 1999; Nishida et al., 1999; Muschler et al., 2001; Stenderup et al., 2001; Justesen et al., 2002). Using molecular measures, we found an age-related decline in osteogenic potential of human bone MSCs (hMSCs) from men (Mueller & Glowacki, 2001). In this study, we tested the hypothesis that there are intrinsic effects of aging on the biology of hMSCs. We tested the effect of age on proliferation, cell cycling, apoptosis, senescence-associated β-galactosidase (SA-β-gal), and osteoblast differentiation.


Yield of STRO-1+ cells from human marrow mononuclear cells (MMCs)

Sixteen samples (42–75 years of age) of low-density MMCs were sorted for STRO-1+ cells by flow cytometry. The mean percent yield of STRO-1+ cells was 0.062 ± 0.038%. There was no significant difference in percentage yield of STRO-1+ cells with age of the subjects (Spearman correlation, r = –0.297, P = 0.242) (Fig. 1).

Figure 1.

Yield of STRO-1+ cells in human marrow. Low-density mononuclear cells were sorted for STRO-1 by FACS. With the number available, there was no apparent relationship between percentage yield of STRO-1+ human marrow mononuclear cells and age (n = 16).

Effects of age on SA-β-gal in hMSCs

Seventeen samples of hMSCs (passage 2, 17–90 years old) were assayed for percent of cells positive for SA-β-gal (Fig. 2). These data suggest that there may not be a linear increase in positive cells with age, but rather a threshold age of its appearance. Frequency of hMSCs positive for SA-β-gal was fourfold greater in the group = 55 years (9.1 ± 3.0, n = 12) than in the group younger than 50 years (2.3 ± 1.8, n = 5, P < 0.001, nonparametric Mann–Whitney test) (Fig. 2 inset).

Figure 2.

Effect of age on senescence-associated β-galactosidase (SA-β-gal)-positive human bone marrow stromal cells (hMSCs). hMSCs isolated from 17 subjects were stained for SA-β-gal. Value of percent positive cells is expressed for each subject as mean ± standard deviation with four replicates. Inset: Frequency of hMSCs positive for SA-β-gal was fourfold higher in the group of subjects older than 55 years (n = 12) than in the group younger than 50 years (n = 5) (P < 0.001, Mann–Whitney test).

Effects of age on proliferation of hMSCs

Population kinetics were evaluated with short-term cultures of freshly isolated hMSCs obtained from young (< 50 years old, n = 3) and older (≥ 55 years old, n = 16) subjects. Cell expansion was slower in cells from the older subjects (Fig. 3A). Cell population doubling time (CPDT) was 1.7-fold longer in cells from the older (76.1 ± 23.4 h) than the younger (44.0 ± 7.7 h) subjects (P = 0.0021, Mann–Whitney test) (Fig. 3B inset), and was positively correlated with age (Spearman r = 0.62, P = 0.005, n = 19) (Fig. 3B).

Figure 3.

Effect of age on proliferation of human bone marrow stromal cells (hMSCs). (A) Proliferation rate was determined by counting cells with a hemacytometer. The hMSCs isolated from younger subjects (< 50 years) expanded more rapidly than hMSCs obtained from older (≥ 55 years) subjects (only four shown here for legibility). (B) Cell population doubling time (hours) of hMSCs obtained from all 19 subjects showed a significant correlation with age (r = 0.62, P = 0.005, Spearman rank order correlation). Inset: hMSCs isolated from 16 older subjects had a significantly longer cell doubling time than cells obtained from three younger subjects (P = 0.0021, Mann–Whitney test).

Cell cycle and effect of age on apoptosis of hMSCs

To determine whether the age-related increase in doubling time results from disproportionately prolonged dwelling in a specific phase(s) of the cell cycle, we subjected cells from ten subjects age 17–90 years to flow cytometric analysis of distribution in G0/G1, S, and G2/M phase. On average, 91.3 ± 0.5% of the cells were in G0/G1 with 6.2 ± 0.5% in S phase and 1.8 ± 0.2% of the cells in G2/M phase, with no effect of age apparent with the numbers available. These preliminary data suggest that with age, there is a uniform prolongation of the duration of all phases of the cycle. There was a range of 0.2–4.2% apoptotic cells, with a significant increase with age (r = 0.719, P = 0.016, Spearman rank order correlation test) (Fig. 4).

Figure 4.

Effects of age on apoptosis of human bone marrow stromal cells. Cells from ten subjects were analyzed with flow cytometric analysis for apoptotic cells. There was a significant increase of apoptotic cells with age (r = 0.719, P = 0.0157, Spearman rank order correlation test).

Effects of age on p53 pathway in hMSCs

Human MSCs (passage 2) were cultured in MEM-α with 10% fetal bovine serum-heat inactivated (FBS-HI) and antibiotics. Upon reaching 50% confluence, total RNA was isolated with TRIzol reagents (Invitrogen, Carlsbad, CA, USA). To test the effects of age on p53 pathway gene expression, we examined the expression of p53, p21, and BAX by semiquantitative reverse transcriptase–polymerase chain reaction (RT–PCR) for hMSCs from 12 subjects (17–90 years old) (Fig. 5A). There was significantly greater expression (Fig. 5B) of p53 (2.3-fold, P < 0.01, Mann–Whitney test), p21 (2.1-fold, P < 0.01), and BAX (2.2-fold, P < 0.05) in hMSCs obtained from older subjects (n = 9, ≥ 55 years) compared with younger subjects (n = 3, < 50 years). Analyses of correlations between p53 expression and each of those two target genes showed significance with p21 (Spearman r = 0.846, P = 0.0009) and BAX (Spearman r = 0.587, P = 0.049).

Figure 5.

Effect of age on p53 pathway genes in human bone marrow stromal cells (hMSCs). (A) An age-related increase in p53, p21, and BAX gene expression was detected by reverse transcriptase–polymerase chain reaction (RT–PCR). (B) Semiquantitative RT–PCR analysis indicates significantly more expression of p53, p21, and BAX genes (normalized to internal control GAPDH) in hMSCs obtained from older subjects (n = 9,  ≥ 55 years), compared with hMSCs obtained from younger subjects (n = 3, < 50 years) (*P < 0.01 and #P < 0.05, Mann–Whitney test). Each lane indicates age and gender of the subject.

Effects of age on osteoblastogenesis in hMSCs and STRO-1+ MMCs

After hMSCs (n = 10, 17–90 years) were cultured for 2 weeks in osteoblastogenic medium, alkaline phosphatase (AlkP) enzymatic activity was measured as an index of osteoblast differentiation. There was a significant decrease of AlkP activity with age (Spearman r = –0.82, P < 0.001) (Fig. 6A). In addition, semiquantitative RT–PCR analysis (Fig. 6B) showed significantly greater expression of osteoblast marker genes Cbfa1/Runx2, Osterix, AlkP, bone sialoprotein (BSP), and osteocalcin (OC) in hMSCs obtained from younger subjects (n = 3, < 50 years), compared with hMSCs from older subjects (n = 7, ≥ 55 years) (P < 0.05, Mann–Whitney test). There was no significant difference in COL I gene expression, a marker for fibroblast and osteoblast phenotypes.

Figure 6.

Effects of age on osteoblastogenesis. (A) Osteoblast differentiation of human bone marrow stromal cells (hMSCs) was assessed with alkaline phosphatase (AlkP) enzyme activity assays, after 2 weeks’ culture in osteogenic medium. There was a significant decrease of AlkP activity with age (Spearman correlation, r = –0.82, P < 0.001) in hMSCs obtained from ten women. (B) Semiquantitative reverse transcriptase–polymerase chain reaction showed that there is significantly less expression of Runx2, Osterix, AlkP, BSP, and OC genes (normalized to internal control GAPDH) in hMSCs obtained from old subjects (n = 7,  ≥ 55 years old) compared with hMSCs obtained from younger subjects (n = 3, < 50 years old) (P < 0.05, Mann–Whitney test). (C) Osteoblast differentiation of STRO-1+ cells from eight men was assessed by fluorescent immunoreactivity for AlkP, 4 weeks after seeding at 1–50 cells per well in osteogenic medium. Data are presented as the percentage of wells at each seeding density with clear fluorescein isothiocyanate signal that was above background and control levels. There was an age-dependent decrease in cells that differentiated into AlkP-positive osteoblasts (Spearman correlation, r = –0.714, P = 0.0466, n = 8 for cumulative data).

In addition, STRO-1+ human MMCs (n = 8, 40–83 years old) were assessed for differentiation to osteoblasts 4 weeks after seeding at very low densities, from 1 to 50 cells per well. There was a striking age-dependent decrease in the percent of the sorted cells that differentiated into AlkP-positive osteoblastic cells (r = –0.714, P = 0.047 for cumulative data) (Fig. 6C).


Several intrinsic properties of human MSCs show effects related to the age of the person from whom the cells were obtained. Stem and progenitor cells are located throughout the adult body and are believed to be involved in continuous maintenance and repair of tissues (Fehrer & Lepperdinger, 2005). Questions arise as to what extent adult mesenchymal stem/stromal cells are either subject to, or causes of aging, and whether age-related changes in those cells are caused by intrinsic factors or are induced by the extrinsic somatic environment, e.g. by declines in circulating hormones. Mesenchymal stem cells or MSCs derived from various sources such as bone marrow or fat can differentiate in vitro into osteoblasts, chondroblasts, adipocytes, myoblasts, and fibroblasts (Prockop, 1997; Krebsbach et al., 1999; Pittenger et al., 1999; Caterson et al., 2002). The relationship between age-related skeletal diseases, such as osteoporosis and osteoarthritis, and aging of MSCs is not yet well documented.

To determine whether there are intrinsic, age-related changes of hMSCs in vivo, we evaluated either freshly isolated marrow cells or stromal cells at very early passage. This approach was used to avoid changes in cell behaviors that are associated with prolonged culture, such as in vitro senescence or culture stress. Like other normal mammalian cells, such as human osteoblasts (Kassem et al., 1997), when cultured for many passages, MSCs display what is termed ‘in vitro senescence’, i.e. decreased proliferation, replicative quiescence, enlargement, increase in SA-β-gal activity, and erosion of telomeres (reviewed in Fehrer & Lepperdinger, 2005; Sethe et al., 2006). There is controversy to what degree such in vitro changes recapitulate organismal aging (Bird et al., 2003; Herbig et al., 2006). The results obtained herein, however, should reflect the effects of in vivo aging because cells from young and old individuals were treated the same way and evaluated upon isolation or at early passage.

Age and proliferation

There were striking age-dependent decreases in proliferation and osteoblast differentiation, and increases in SA-β-gal activity, apoptosis, and p53 pathway genes in hMSCs obtained from individuals aged 17–90 years. Understanding the normal processes of cell replacement may be crucial to understanding the imbalance between bone destruction and bone formation in the aging individual. Stem and progenitor cell dynamics may be the determinant of whether tissues show effects of aging that take place at the cellular level (Carrington, 2005). In this study of short-term culture of freshly isolated cells, the doubling time of hMSCs was longer for cells from the older than the younger subjects, and was correlated with age. Theoretically, the prolonged doubling time could have been due, for example, to a prolonged G1 or G2 arrest with fewer numbers of cells entering S or M, respectively. Because there was no detectable difference in the distribution of cells in each phase of the cell cycle, however, those pilot data suggest that aging may uniformly increase the duration of all phases of cell cycle. The significant increase in apoptotic cells in MSCs obtained from older subjects suggests that loss of cells may contribute to the differences in population kinetics. Alternately, the cells from elders may have been less stimulated by, or more sensitive to harmful elements in fetal calf serum or to other stressful aspects of cell isolation or culture. Another study with marrow aspirates showed that MSCs from young subjects had more population doublings in long-term culture than did the MSCs from older subjects (Stenderup et al., 2003; Kassem, 2006).

Consistent with age-related decreased proliferation and increased apoptosis of hMSCs, there was an age-related increase in p53 and its targets p21 and BAX, which are known to mediate those cellular functions, respectively. p53 Is known to play critical roles in senescence and apoptotic responses to dysfunctional telomeres (Artandi & Attardi, 2005) and genotoxic stress (Helmbold et al., 2006). Once activated, p53 induces p21 and BAX, which mediate different aspects of p53 action on senescence and apoptosis, respectively (Vousden & Lu, 2002; Artandi & Attardi, 2005). Our data for age-related increases in p53, p21, and BAX gene expression in hMSCs suggest that up-regulation of the p53/p21 pathway may regulate age-related decreases in proliferation, and the p53/BAX pathway may mediate age-related increases in apoptosis. Further, finding correlations between p53 and each of those target genes suggests constitutive relationships in this pathway even after isolation of the cells.

In this study, there was a higher frequency of SA-β-gal-positive cells in human MSCs isolated from older subjects, compared with younger ones. This marker is best known for skin fibroblasts; SA-β-gal positive fibroblasts accumulate with age in skin samples and have been shown to produce degradative enzymes and inflammatory cytokines, which can disrupt tissue organization and function (Campisi, 1998, 2005). A previous small study with marrow aspirates showed no effect of age on SA-β-gal, although all specimens showed the expected increase with time in culture (Stenderup et al., 2003). In that study, SA-β-gal was measured after considerable expansion of the stromal cells, that is, ‘to less than 50% of in vitro lifespan’. Those cultures from young and older donors showed 17% and 16% SA-β-gal-positive cells, values much greater than those reported here. It is possible that those findings reflect in vitro abolishment of detectable differences. Although SA-β-gal is widely used in studies of in vitro replicative senescence and of pathological tissues, finding an age-associated increase in this marker in early passage human MSCs extends its utility as a correlate for in vivo age. There is a potential theoretical limitation of using surgical discarded tissue for studying normal aging because of a possible systemic impact of the diagnosis of osteoarthritis, but this has been minimized by studying marrow from young and old subjects with similar stage of advanced joint disease and by excluding comorbidities and medications that may confound interpretation. We consider finding an age-related increase in SA-β-gal-positive cells in hMSCs as evidence in support of using orthopedic surgical discarded tissue for studies of aging. Scarcity of positive cells in samples from younger subjects (< 50 years of age) indicates that this marker is unrelated to the diagnosis of osteoarthritis and is related to the age of the subject.

Age and osteoblast differentiation

There are apparently conflicting data about the effect of age on osteoblast potential of human marrow-derived cells (reviewed in Mueller & Glowacki, 2001). Finding either no effect or an age-related decline in osteoblast potential may in part be attributable to use of marrow aspirates, frozen cells, growth-factor-supplemented media, and use of different assays for the number of colonies that stain for AlkP. Colony assays have inherent inaccuracies regarding the threshold volume of each counted colony, intensity of stain, and problems with necrosis in cultures of longer duration. In addition, studies that use colony size or number have been criticized for inappropriate use of parametric statistical methods (Dobson et al., 1999). We previously used a three-dimensional (3-D) culture system and quantitative RT–PCR assays to measure generation of AlkP-positive osteoblasts from human marrow from men, and found a striking age-dependent decline (Mueller & Glowacki, 2001).

Human MSCs isolated from marrow aspirates by different sampling methods for small amounts of marrow may result in heterogeneity of the cell population, which affects in vitro cell behavior (Siddappa et al., 2007). A more recent review concluded that differences in isolation methods (whole marrow, low-density fraction, adherent stroma), sources (biopsies and necropsies), anatomical sites, colony assay protocols, and ambiguous terminology confound integration of the literature and resolution of apparently conflicting conclusions (Sethe et al., 2006).

The studies reported herein provide assessment of osteoblast potential without the confounders present with colony assays or because of expansion needed with small samples of cells. The use of all the bones discarded from each subject during orthopedic surgery ensures a large enough population of marrow cells to minimize sampling heterogeneity. Further, early evidence of osteoblastogenesis was measured. In addition, to avoid confounding effects of proliferation and differential responses to serum growth factors, osteoblast differentiation assays were always done with confluent, contact-inhibited cultures in osteogenic medium with low serum supplementation. Using the same conditions for culture, treatment, and analysis of hMSCs from groups of younger and older subjects, we report a significant age-related decrease of AlkP activity and immunoreactivity as well as bone marker genes, such as Cbfa1/Runx2, Osterix, AlkP, BSP, and OC, in hMSCs after culture in osteogenic medium. Although there is heterogeneity in hMSC cultures (Phinney et al., 1999), our samples of bone marrow obtained from each subject's whole femoral head provide large numbers of bone MMCs for testing reproducibility of the findings (as many as 800 million low-density cells).

Our experiments with microwell cultures of STRO-1+ marrow cells also showed a striking age-dependent decrease in differentiation into AlkP-positive osteoblastic cells. The observation that the number of STRO-1+ cells in the low-density mononuclear fraction of fresh marrow did not decline with age is supported by other studies of osteogenic stem cells (Stenderup et al., 2001) and muscle precursor cells (Conboy et al., 2003). Although the number of STRO-1+ cells may not decrease with age, their potential for osteoblastic differentiation does. Thus, our cumulative experiments with three different assays for osteoblastic differentiation, including a 3-D culture system (Mueller & Glowacki, 2001), all indicate that there is an age-related decline of osteoblastogenesis in human marrow cells. MSCs are characterized by the expression of numerous surface antigens, but none of them appears to be exclusively expressed on MSCs (Bobis et al., 2006). Antibodies against STRO-1 bind to stromal cell precursors and also to nucleated erythroid precursors (Simmons & Torok-Storb, 1991). An exclusive cell surface marker for MSCs will be critical for studies of age-related changes of MSCs in vivo.

Thus, some data reported here were obtained from two-dimensional (2-D) culture with assays of AlkP enzyme activity and of osteoblast gene expression; other data were obtained from microwell cultures of STRO-1+ cells and fluorescence detection of antibodies against AlkP. Use of those different methods strengthens the conclusions from the literature that support the view that human marrow-derived cells show an age-related decrease in osteoblast potential.

Very recent information from mouse studies indicates that p53 is a negative regulator of osteoblast differentiation. Unlike information with established cancer cell lines that show p53 suppression of proliferation and stimulation of differentiation, p53 is a negative regulator of osteoblast differentiation in vitro and in vivo (Lengner et al., 2006; Wang et al., 2006). Those observations with developing mice may apply more generally as our data with human MSCs show an age-related increase in p53 and decrease in osteoblast differentiation.


The age-related changes in human MSCs described here for proliferation; SA-β-gal; expression of p53, p21, and BAX; and osteoblast potential add to other age-related properties we previously reported for similar marrow samples. Marrow-derived hMSCs also show an age-related increase in constitutive in vitro secretion of the osteolytic cytokines interleukin-6 and -11 (Cheleuitte et al., 1998); increase in magnitude of interleukin-1β-stimulated secretion of those cytokines (Cheleuitte et al., 1998); increase in constitutive secretion of IGF-binding protein-3 (Rosen et al., 1997); and decrease in expression of the osteoclastogenesis-inhibitory factor, osteoprotegerin (Makhluf et al., 2000). Taken together with the finding of an age-related increase in osteoclastogenesis in human marrow cultures (Glowacki, 1995), those findings and the results reported herein support the hypothesis that human marrow cells and their products can contribute to skeletal aging by decreasing the renewal of bone-forming osteoblasts and increasing the generation of bone-resorbing osteoclasts.

In conclusion, there were age-related decreases in proliferation and osteoblast differentiation in human MSCs and increases in apoptosis and in percent of SA-β-gal-positive cells. Up-regulation of the p53 pathway with age may have a critical role in mediating the reduction in both proliferation and osteoblastogenesis of hMSCs. To our knowledge, this is the first report of age-related increases in p53, its targets p21 and BAX, and apoptosis in human MSCs. These findings open new avenues for research on the underlying mechanisms of human skeletal aging. Use of several different measures with monolayer cultures in these studies adds new evidence to the controversy about the effects of age on osteoblast differentiation; these studies reinforce the view that there is an age-related decline in osteoblast differentiation in human MSCs. These results support the conclusion that there are intrinsic alterations in human MSCs with aging that may explain skeletal, cellular, and tissue aging in humans.

Experimental procedures


Bone marrow samples were obtained with IRB approval as femoral tissue discarded during primary hip arthroplasty for osteoarthritis, that is, erosion of the articular cartilage that results in pain and reduced joint mobility. Although the prevalence of osteoarthritis increases with age, there are younger people with this diagnosis who require the same surgery, in which femoral bone and marrow are removed in order for the surgeon to implant the femoral component of the joint prosthesis. Thus, we identified from the surgical schedule a wide range of ages of subjects with similar stage of advanced hip osteoarthritis in order to minimize any possible impact of the joint disease on effects of age on marrow. Criteria for exclusion are rheumatoid arthritis, cancer, and other comorbid conditions that may influence skeletal metabolism, that is, renal insufficiency, alcoholism, active liver disease, malabsorption, hyperthyroidism, ankylosing spondylitis, aseptic necrosis, hyperparathyroidism, morbid obesity, and diabetes. Also excluded were patients who were taking medications that may influence skeletal metabolism (e.g. thyroid hormone, glucocorticoids, NSAIDs, and bisphosphonates). A total of 57 subjects, 33 women and 24 men, age from 17 years old to 90 years old, were included in this study. Of them, 17 subjects were classified for convenience as younger (≤ 50 years, mean 41 ± 10 years) and 40 were classified as older (≥ 55 years, mean 67 ± 8 years). Not all specimens could be included in every experiment because of the surgical schedule and numbers of cells needed for each assay. In each experiment, standardized conditions were used for all samples, for example, early cell passage, identical medium, serum, and reagents. For cell cycle, RNA, and AlkP assays, cells obtained from different subjects were cultured, treated, and stored for analysis at the same time to avoid technical differences between assays.

Preparation of hMSCs

Low-density MMCs were isolated by density centrifugation on Ficoll/Histopaque 1077 (Sigma, St Louis, MO, USA) (Cheleuitte et al., 1998; Zhou et al., 2005b). This procedure enriches for undifferentiated cells. Low-density MMC preparations include a population of non-adherent hematopoietic cells and a fraction capable of adherence and differentiation into musculoskeletal cells. Adherent human MSCs were expanded in 2-D monolayer culture with phenol red-free α-MEM medium, 10% FBS-HI, 100 U mL−1 penicillin, and 100 µg mL−1 streptomycin (Invitrogen).

STRO-1+ human bone MMCs

For some experiments, STRO-1+ cells were isolated from the low-density mononuclear cell suspensions with monoclonal anti-STRO-1 primary antibody (Mouse IgM anti-human marrow stromal progenitor cells obtained from Dr Paul J. Simmons, Adelaide, Australia) (Simmons & Torok-Storb, 1991) and flow cytometric sorting. Samples from 12 men and 4 women between 40 years and 83 years of age, containing more than approximately 10 million cells were divided into three groups (unstained control or for incubation with either STRO-1 antibody or isotype-matched control antibody, Mouse IgM, BD Pharmingen, San Diego, CA, USA) for 15 min at 4 °C. Following two washes with phosphate-buffered saline (PBS) with 2% FBS, the samples were incubated with fluorescein isothiocyanate (FITC)-labeled goat anti-mouse IgM antibody (Jackson Immuno Research Laboratories, Inc, Westgrove, PA, USA) for an additional 15 min at 4 °C. The STRO-1+ cells were isolated with a Mo-Flo high-speed sorter (DAKO, Fort Collins, CO, USA). Positive populations were enumerated by morphology (forward scatter for size, and side scatter for granularity) and by level of fluorescence greater than autofluorescence of unstained cells and than nonspecific fluorescence of the isotype-matched control as described (Simmons & Torok-Storb, 1991). Percent STRO-1 yield was calculated for each sample. For assays of osteoblast potential, STRO-1+ human bone MMCs from eight men were plated in 1% gelatin-coated Terasaki microwell plates (72-well plates) at densities ranging from 1 to 50 cells per well. Cells were cultured for 4 weeks in medium supplemented with 10 nm dexamethasone, 5 mmβ-glycerophosphate, and 50 µg mL−1 ascorbate-2-phosphate. At day 28, viable cells in all wells were counted and stained for AlkP as follows. After wells were rinsed, 7 µL of monoclonal antibone AlkP (B4-78 hybridoma, DSHB, University of Iowa) or NS-1 normal serum (negative control) was added and incubated for 20 min. Cells were rinsed thrice with PBS and 2% bovine serum albumin (BSA), and were incubated with goat anti-mouse FITC-conjugated secondary antibody (Cappel, Solon, OH, USA) for 20 min. Following three rinses with PBS and 2% BSA, the cells were scored using fluorescence microscopy. Wells were scored as AlkP positive if they contained cells with clear FITC signal that was above background and control levels. Data are normalized as percent of number of wells with viable cells.

Staining of cells for SA-β-gal activity

Human MSCs from 17 subjects (17–90 years old; 14 women, 3 men) at passage 2 were seeded at 4 × 104 cells in each of four 35 mm dishes (5000 cells cm−2). After 4 days in α-MEM with 10% FBS-HI, cells were stained for SA-β-gal (Dimri et al., 1995). In brief, the cells were rinsed twice with PBS solution, fixed for 5 min in 3% formaldehyde, and rinsed with PBS. Two milliliters of staining solution was added to each 35 mm dish. After overnight incubation at 37 °C, the staining solution was removed, the cells were rinsed with PBS, and a 25 mm coverslip was placed in each dish with glycerol vinyl alcohol mounting solution (Zymed, South San Francisco, CA, USA). The number of stained cells and the total number of cells were enumerated with bright light and phase microscopy, respectively. Values are expressed as percentages of cells that were stained blue.


Human MSCs at passage 2 were seeded at 1 × 104 cells in each of three 35 mm dishes for each time point (0 days, 2 days, 4 days, 6 days, 8 days, and 10 days) in α-MEM with 10% FBS-HI. Cells were suspended with 0.5 mL of 0.05% trypsin–ethylenediaminetetraacetic acid (Invitrogen), and cell number was determined by hemacytometer. Cell population doubling time was calculated with the formula: CPDT = (t – t0) × log2(log[N/N0])−1, with t as time and N as cell number.

Cell cycle and apoptosis analysis

Cell cycle distribution was measured by flow analysis for DNA content (Epperly et al., 1999). Cells (hMSCs, passage 2) at approximately 80% confluence were trypsinized, combined with the floating cells, centrifuged for 10 min at 1500 r.p.m., washed three times in PBS, and resuspended in 1 mL of ice-cold 70% ethanol added dropwise with mixing. The cells were stored for at least 24 h at –20 °C. The cells were centrifuged for 5 min at 3000 r.p.m., and the supernatant was partially decanted, with approximately 200 µL of 70% ethanol retained with the pellet. The cells were vortexed, and 1 mL of staining solution (50 µg mL−1 propidium iodide, 0.1 mg mL−1 RNase A in a solution of 1 g glucose in 1000 mL PBS) was added. The cells were incubated at room temperature for 30 min and stored at 4 °C until analysis with a Mo-Flo high-speed sorter (DAKO). Modfit software (Verity Software House, Topsham, ME, USA) was used to express the data as the percent of cells in phases of the cell cycle, G0/G1, S, and G2/M, and cells that were apoptotic.

Conditions for osteoblast differentiation

For each sample to be assayed for AlkP activity, 2 × 104 cells/well were seeded in triplicate in 12-well-plates in α-MEM with 10% FBS-HI until confluence; this required different times depending upon rates of proliferation. Therafter, cultures were changed to osteogenic medium (α-MEM with 1% FBS-HI, 100 U mL−1 penicillin, 100 µg mL−1 streptomycin plus 10 nm dexamethasone, 5 mmβ-glycerophosphate, 50 µg mL−1 ascorbate-2-phosphate) for 14 days. Reduction of serum to 1% for differentiation was designed to minimize possible differences in proliferation that could confound interpretation of effects of age on osteoblastogenesis. For each sample to be assayed for osteogenic gene expression by RT–PCR, cells were cultured in 100 mm dishes in α-MEM with 10% FBS-HI. Upon confluence, the medium was changed into osteogenic medium for 14 days.

RNA isolation and RT–PCR

Total RNA was isolated from human MSCs with TRIzol reagent (Invitrogen). For RT–PCR, 2 µg of total RNA was reverse transcribed into cDNA with SuperScript II (Invitrogen), following the manufacturer's instructions. One-tenth or one-twentieth of the cDNA was used in each 50 µL PCR reaction (30–40 cycles of 94 °C for 1 min, 55–60 °C for 1 min, and 72 °C for 2 min) as described (Zhou et al., 2005b). The gene-specific primers for human p53 (Vakifahmetoglu et al., 2006); p21 (Lohr et al., 2003); BAX (Tirado et al., 2005); Cbfa1/RUNX2, Osterix, and bone sialoprotein (D’Ippolito et al., 2006); AlkP (Winn et al., 1999), COL I, and Osteocalcin (Lomri et al., 1999) were used for amplification. Gene expression levels were measured by semiquantitative PCR. Polymerase chain reaction products were quantitated by densitometry of captured gel images with KODAK Gel Logic 200 Imaging System and measured by KODAK Molecular Imaging Software, following the manufacturer's instructions (KODAK, Molecular Imaging Systems, New Haven, CT, USA). Quantitative data were expressed by normalizing the densitometric units to GAPDH (internal control).

AlkP enzyme assay

Alkaline phosphatase enzyme activity was measured in ten samples (nine women, one man; 17–90 years old). Dishes were rinsed with PBS, 200 µL of lysis buffer was added, and the plates were stored at –80 °C until assays could be done together (Zhou et al., 2005a).

Statistical analyses

All experiments were performed at least in triplicate. Group data are presented as mean values ± standard deviation. Quantitative data were analyzed with nonparametric tools, either the Mann–Whitney test for group comparisons or Spearman correlation test. A value of P < 0.05 was considered significant.


This study was presented in part at the 27th ASBMR annual meeting, 2005, in Nashville, TN, USA, and 2006 AIMM/ASBMR meeting at Snowmass, CO, USA. The authors greatly appreciate help from I. Amato, K. D. Johnson, N. A. Glass, A. Tilt, and Dr S. Mizuno for aspects of these experiments. This study was supported by grants from the National Institutes of Health R01 AG 025015 and R01 AG 028114. The discarded marrow was obtained and studied with approval and annual review from the Partners Human Research Committee.