Dr Darwin J. Prockop, Center for Gene Therapy, MCP Hahnemann University, 245 North 15 Street, 10118 NCB, Mail Stop 421, Philadelphia, PA 19102-1192, U.S.A.
Marrow stromal cells (MSCs) were isolated from bone marrow obtained by aspirates of the iliac crest of normal volunteers. The cells were isolated by their adherence to plastic and then passed in culture. Some of the samples expanded through over 15 cell doublings from the time frozen stocks were prepared. Others ceased replicating after about four cell doublings. The replicative potential of the cells in culture was best predicted by a simple colony-forming assay in which samples from early passages were plated at low densities of about 10 cells per cm2. Samples with high colony-forming efficiency exhibited the greatest replicative potential. The colonies obtained by plating early passage cells at low density varied in size and morphology. The large colonies readily differentiated into osteoblasts and adipocytes when incubated in the appropriate medium. As samples were expanded in culture and approached senescence, they retained their ability to differentiate into osteoblasts. However, the cells failed to differentiate into adipocytes. The loss of multipotentiality following serial passage in culture may have important implications for the use of expanded MSCs for cell and gene therapy.
One of the most convenient sources of human MSCs is from aspirates taken from the iliac crest under local anaesthesia. Here we show that samples of MSCs obtained from bone marrow aspirates from normal volunteers varied widely in the expandability in culture, but the expandability could be predicted by a simple assay for CFUs. As reported previously by Bruder et al (1997), the cells retain the ability to differentiate into osteoblasts after extensive proliferation in culture. However, we show here that they lose some of their multipotentiality, since they can no longer be differentiated into adipocytes as they approach senescence in culture.
MATERIALS AND METHODS
Isolation and culture of hMSCs
hMSCs were obtained from 20 ml aspirates from the iliac crest of normal donors ranging in age from 19 to 49 years. Each 20 ml of aspirate was processed by first diluting the aspirated marrow with 20 ml Hanks' balanced salt (HBS; Gibco) and washing by gentle inversion several times. About 10 ml of ficoll (Ficoll-Paque; Pharmacia) was layered beneath 40 ml of sample and spun at 2500 g for 30 min at room temperature. The mononuclear cell layer was removed from the gradient interface and washed with Hanks' balanced salt. Cells were spun at 1500 g for 15 min and resuspended in complete medium (Minimum Essential Medium, alpha medium without deoxyribonucleotides or ribonucleotides, Gibco; 20% fetal calf serum lot-selected for rapid growth of hMSCs, FCS, Atlanta Biologicals; 100 units/ml penicillin, 100 μg/ml streptomycin, Gibco; and 2 mml-glutamine, Gibco). Cells were plated in a 25 cm2 tissue culture flask (Nunc) and incubated at 37°C with 5% humidified CO2. After 24 h, non-adherent cells were removed. Adherent cells were washed vigorously twice with PBS and shaking to remove adherent haemopoietic precursors, and fresh complete medium was added. The medium was replaced every 3 or 4 d and the cells grown to 70–90% confluency. They were harvested with 0.25% trypsin and 1 mm EDTA (Gibco) for 5 min at 37°C. The cells were re-plated in a 75 cm2 flask, again grown to confluency, and harvested. About 8 ml of complete medium was mixed with the trypsinized cells to inactivate the trypsin, and the cells were counted with an automated instrument (Coulter) or a haemacytometer. The cells were then slowly frozen in 5% DMSO and 30% FCS, and they were stored frozen in liquid nitrogen. To expand the cells through successive passages, they were plated at 5000 cells per cm2 (about a 1:3 dilution), grown to near confluency, and harvested with the same protocol. At the end of each passage, the cells were counted on a haemocytometer to calculate cell doublings.
hMSCs expanded in culture to 70–90% confluency were harvested with trypsin-EDTA and counted using a haemocytometer. A glass Pasteur pipette was flamed at its tip to reduce its diameter, and the cells were drawn through the narrowed pipette several times to ensure cell separation. Cells were diluted in complete medium, and they were plated at about 10 cells per cm2 in 100 mm tissue culture dishes (Falcon). After incubation for 10–14 d at 37°C in 5% humidified CO2, the cells were washed with PBS and stained with 0.5% Crystal Violet in methanol for 5–10 min at room temperature. Cells were washed with PBS twice and visible colonies were counted. To isolate colonies, unstained colonies were recovered using cloning cylinders and trypsin-EDTA.
Osteogenic differentiation of hMSCs
hMSCs were grown to 70–90% confluency in wells of 12-well tissue culture plates and then incubated in osteogenic medium (10−8m dexamethasone; 0.2 mm ascorbic acid, Sigma; and 10 mm beta-glycerolphosphate, Sigma). The medium was replaced every 3–4 d and deposition of mineral was observed after 2–3 weeks. To assess mineralization, cultures were washed with PBS and fixed in a solution of ice-cold 70% ethanol for 1 h. Cultures were rinsed with water and stained (Stanford et al, 1995) for 10 min with 1 ml of 40 mm Alizarin red (pH 4.1; Sigma) with rotation. Cultures were rinsed two to three times with PBS to reduce non-specific staining. Because the stained mineral was uniformly distributed and obscured the cells, the mineralization was assayed by examination of multiple fields for the area of mineralization as a percent of the total area of the confluent cultures.
Adipogenic differentiation of hMSCs
hMSCs in 85–95% confluent cultures (Kelly & Gimble, 1998) were incubated in complete medium supplemented with MHI (0.5 μm hydrocortisone, 0.5 mm isobutylmethylxanthine, and 60 μm indomethacin). The medium was replaced every 3–4 d. Cells containing lipid vacuoles were observed after 2–3 weeks. Cells were washed with PBS and fixed in 10% formalin for 10 min. Cells were then stained for 10–15 min with fresh Oil red-O solution (Houghton et al, 1998). The Oil red-O solution was prepared by vigorously mixing 3 parts stock solution (0.5% in isopropanol; Sigma) with 2 parts water for 5 min and filtering through a 0.4 μm filter. Plates were washed three times with water. The percent of adipocytes was assayed by counting 50–100 cells in multiple fields.
CFU-F assays predict life-span in culture
Nucleated cells were isolated from bone marrow aspirates of normal donors and hMSCs were isolated by their adherence to plastic culture dishes. After the cells were grown to confluency first in a 25 cm2 flask and then in a 75 cm2 flask, frozen stocks were prepared. The total number of subsequent population doublings obtained from samples from different donors was assayed by serial replating of the frozen stocks of cells. Recent reports indicated that cryopreservation does not significantly alter the properties of either canine MSCs (Hurwitz et al, 1997) or human MSCs (Bruder et al, 1997). Assays by FACS of confluent cultures obtained from the first plating of the frozen cells (second passage) indicated the cells were negative for CD45, a marker for haemopoietic cells (Clark & Keating, 1995). About 8% were strongly positive for Stro-1 (Simmons & Torok-Storb, 1991), and the positive cells were in the largest and most granular cell fraction. The remainder of the cells were weakly positive for Stro-1 (not shown). As indicated in Fig 1, there was a large variation in population doublings obtained with samples from different donors. Some expanded through over 15 cell doublings from the time frozen stocks were prepared. Others ceased replicating after about four cell doublings. The replicative potential of the cells was not reflected in their initial growth rate (compare donor 11 and donor 13 in Fig 1). Also, the variation was not related to the gender or age of the donors (not shown).
Assay of cells after the fourth passage indicated that there was also a large variation in the number of colonies obtained in a simple CFU assay in which the cells were plated at low density (10 cells per cm2). There was a positive correlation between number of mononucleated cells (MNCs) initially plated and the number of CFUs (r2 = 0.63, n = 8, P < 0.011), but there was no correlation when duplicate samples obtained at the same time from the same individual were compared. Specifically, there was no correlation between duplicate samples from donors 54, 56 and 57 (not shown). However, the number of population doublings obtained in culture was closely correlated (r2 = 0.9419, n = 5, P = 0.006) with values obtained in the CFU assay (Fig 2).
As expected, the number of CFUs declined as the cells were expanded in culture. The decline of CFUs with passage number was apparent from assays of a single sample (Fig 3, upper panel) and from assays of five samples from different donors compared at passage 4 and passage 12 (Fig 3, lower panel).
hMSC CFU-Fs vary in size, morphology and growth rate
As observed previously, colonies formed by sparsely plated hMSCs in CFU assays varied in size (see Mets & Verdonk, 1981; Bruder et al, 1997). As indicated in Figs 4 and 6 (upper and second panels), the largest colonies were consistently 2–3 times the size of the smallest. The larger colonies were composed of small spindle-shaped cells (Fig 6, second left panel). The smaller colonies were composed of broad flattened cells and tended to be less confluent than the larger colonies (Fig 6, second right panel).
Assays for differentiation into osteoblasts and adipocytes
To assess the multipotentiality of the hMSCs, cells from passage 2 were plated at low density and individual colonies were isolated with cloning rings. Duplicate aliquots of the cells from the colonies were then grown to near confluency and assayed for differentiation into either osteoblasts or adipocytes. All the clones isolated from early passage cultures differentiated into both phenotypes (Figs 5 and 6). However, there was considerable variation in the extent of differentiation. With subcultures of some colonies, the plates were almost completely covered with mineral after incubation with osteogenic medium. In duplicate samples, essentially all the cells contained Oil Red-O staining vacuoles after incubation in adipogenic medium. With subcultures of other colonies, the cells either preferentially differentiated into one phenotype, or poorly into either (Fig 5). The variation was seen even in comparisons of large colonies isolated from the same plate from the same donor.
Similar differentiation assays were carried out with late passage cells. Because the cells no longer generated colonies after sparse plating, the assays were carried out in near confluent cultures after plating at about a 1:3 dilution. The late passage cultures retained the ability to differentiate into osteoblasts (lower left panel in Fig 6, and Table I). In fact, a few cultures spontaneously began to deposit mineral (not shown). However, the cells failed to differentiate into adipocytes as they lost their ability to generate CFUs (lower right panel in Fig 6, and Table I).
Table 1. Table I. Differentiation of early and late passage hMSCs.* * With passage 2 cells, large colonies were isolated from a CFU assay with cloning cylinders, duplicate aliquots of the cells were grown to near confluency, and then the cultures were incubated in either adipogenic or osteogenic medium for 18 d. With passage 12 cells, the cells were grown to near confluency after plating at a 1:3 dilution and then duplicate cultures were incubated in either adipogenic or osteogenic medium for 18 d. As indicated in Fig 3, few if any CFUs were obtained with passage 12 cells from other donors.
MSCs isolated by their adherence to plastic culture surfaces have characteristic properties that have been well-defined by a number of investigators (see Friedenstein et al, 1976, 1987; Castro-Malaspina et al, 1980; Mets & Verdonk, 1981; Piersma et al, 1985; Owen & Friedenstein, 1988; Caplan, 1991; Prockop, 1997). However, there is considerable variability in cultures prepared from different sources and under different conditions. For example, initial cultures of MSCs from murine bone marrow are heavily contaminated by haemopoietic precursors (Clark & Keating, 1995), and the yields of MSCs are extremely low from some inbred strains (Phinney et al, 1999). In contrast, haemopoietic precursors appear to be shed rapidly from cultures of human MSCs. However, at least two morphologically distinct cells are present (see Mets & Verdonk, 1981; Satomura et al, 1998): type I cells that are spindle shaped and grow rapidly, and type II cells that are broad and grow slowly. The type II cells increase in number as the cells are passed and apparently arise from the type I cells. In addition, cells with intermediate morphologies are observed.
The results here demonstrated that samples of human MSCs obtained from iliac crest aspirates varied widely in their expandability in culture. The variation was not explained by gender or age of the donors. Also, it was not explained by the number of nucleated cells in the sample. Instead, it apparently reflected a sampling variation in aspirates of marrow from the iliac crest, since the variation was seen between two samples taken from the same volunteer at the same time. MSCs arise from the complex architectural structures of perivascular cells that incompletely separate the marrow itself from capillaries (see Prockop, 1997), and the yield of MSCs apparently varies with the presence of such architectural structures in the aspirate site.
The expandability of the MSC cultures was not predictable from the initial growth rates in the first or second passage. It was, however, predictable on the basis of a simple assay for CFUs. Although definitive experiments with genetic markers have not been reported, each colony in a CFU assay is generally assumed to arise from a single cell (see Castro-Malaspina et al, 1980; Kuznetsov et al, 1997; Satomura et al, 1998). Therefore the number of CFUs obtained from early passage cultures probably reflects the number of MSCs or precursors of MSCs in a sample of marrow. In samples with low numbers of colony-forming MSCs, the cells probably undergo many doublings before reaching confluency in the first plating and therefore have a limited potential for further expansion. The results here suggest that a simple CFU assay identifies samples that are rich in type I spindle-shaped cells, and therefore samples that have the greatest potential to expand in culture. Also, the samples rich in type I cells appear to have the greatest potential for differentiation. Therefore CFU assays in early passage hMSCs will probably be useful in identifying samples that are appropriate for extensive characterization or gene manipulation.
As recently demonstrated by Kuznetsov et al (1997) and Satomura et al (1998), the colonies obtained after sparse plating of early passage hMSCs were heterogenous in size, morphology, and potential to differentiate into osteoblasts in vivo. As shown here, clones from early passage cells were also heterogenous in their potential to differentiate into adipocytes in culture. Therefore the cells were not equally multipotential. As the cells were passed in culture, there was a further loss in multipotentiality. Bruder et al (1997) reported that samples of hMSCs could retain their potential to differentiate into osteoblasts even after 15 passages in culture. Those observations were confirmed here. In fact, some late passage hMSCs spontaneously differentiated into osteoblasts. The spontaneous differentiation into osteoblasts may reflect the presence of cytokines or other factors in the fetal calf serum in the medium that select for osteoblast precursors. However, as the cells approached senescence in culture, they lost their ability to differentiate into adipocytes. The results indicate, therefore, that some MSCs retain their multipotentiality through a number of passages (Fig 7), whereas others either become committed to a specific differentiation pathway or begin to senesce. The loss of multipotentiality as the cells are replicated in culture may have important implications for use of expanded cultures of MSCs for cell and gene therapy (Caplan, 1991; Goldberg & Caplan, 1994; Prockop, 1997; Azizi et al, 1998). Also, the results suggest that it is important to use more than one differentiation assay in evaluating the pluripotentiality of MSCs that are purified by immunoselection and other procedures (Gronthos et al, 1994; Gronthos & Simmons, 1995, 1996; Long et al, 1995; Waller et al, 1995; Rickard et al, 1996; Joyner et al, 1997).