Histochemistry and Cell Biology

, 132:547

Osteogenic properties of late adherent subpopulations of human bone marrow stromal cells


  • Elisa Leonardi
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
  • Gabriela Ciapetti
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
  • Serena Rubina Baglìo
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
  • Valentina Devescovi
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
  • Nicola Baldini
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
    • Laboratory for Orthopaedic Pathophysiology and Regenerative MedicineIstituto Ortopedico Rizzoli
Original Paper

DOI: 10.1007/s00418-009-0633-x

Cite this article as:
Leonardi, E., Ciapetti, G., Baglìo, S.R. et al. Histochem Cell Biol (2009) 132: 547. doi:10.1007/s00418-009-0633-x


The nonadherent (NA) population of bone-marrow-derived mononuclear cells (MNC) has been demonstrated to be a source of osteogenic precursors in addition to the plastic-adherent mesenchymal stromal cells (MSC). In the current study, two subpopulations of late adherent (LA) osteoprogenitors were obtained by subsequent replating of NA cells, and their phenotypic, functional, and molecular properties were compared with those of early adherent (EA) MSC. Approximately 35% of MNC were LA cells, and they acquired a homogeneous expression of MSC antigens later than EA cells. In EA-MSC, the alkaline phosphatase (ALP) activity increased significantly from time of seeding to the first confluence, whereas in LA cells it raised later, after the addition of mineralization medium. All subpopulations were able to produce type I collagen and to deposit extracellular matrix with organized collagen fibrils. The proportion of large colonies with more than 50% of ALP positive cells as well as the calcium content was higher in LA than in EA cells. Molecular analysis highlighted the upregulation of bone-related genes in LA-MSC, especially after the addition of mineralization medium. Our results confirm that bone marrow contains LA osteoprogenitors which exhibit a delay in the differentiation process, despite an osteogenic potential similar to or better than EA-MSC. LA cells represent a reservoir of osteoprogenitors to be recruited to gain an adequate bone tissue repair and regeneration when a depletion of the most differentiated component occurs. Bone tissue engineering and cell therapy strategies could take advantage of LA cells, since an adequate amount of osteogenic MSCs may be obtained while avoiding bone marrow manipulation and cell culture expansion.


Bone marrow stromal cellsNonadherent cellsOsteoprogenitorsBone repairGene expression


Adult bone marrow is the tissue most commonly used as a source of multipotent mesenchymal stromal cells (MSC), defined for their ability to differentiate in vitro into multiple mesenchymal lineages, such as osteoblasts, chondrocytes, and adipocytes (Bianco et al. 2001; Horwitz et al. 2005).

The osteogenic properties of MSCs make them attractive for bone repair and regeneration, both in cell therapy and tissue engineering applications (Ciapetti et al. 2006; Brooke et al. 2007; Dallari et al. 2007; Marcacci et al. 2007; Tseng et al. 2008). However, one of the main goals is still to optimize MSC recovery and ex-vivo expansion so as to achieve, in reasonable times, a vast number of functional cells (Neuhuber et al. 2008). MSCs with osteogenic potential are often obtained from autologous bone marrow or bone tissues, and in vitro cultured for a rather long time period (4–6 weeks) in order to obtain enough cells for clinical use. Among MSC isolation methods, the adherence to tissue culture plastic surface (TCPS) remains the oldest and most popular one, although not all authors agree on a common time to let the cells adhere, varying from a few hours to several days in similar protocols (Dominici et al. 2006). Moreover, culture expansion, which is labor intensive and therefore expensive, presents some risks, such as contamination with bacteria or viruses, and decrease of the proliferative and differentiative capacity of the osteoprogenitors prior to implantation (Muschler et al. 2004). Therefore, the large-scale clinical application of bone-marrow-derived cells for bone tissue repair has not been entirely realized (Brooke et al. 2007; Niemeyer et al. 2006; Valtieri and Sorrentino 2008; Granero-Molto et al. 2008). The concept that bone-marrow-derived osteogenic precursors reside exclusively in adherent cell population was overcome by the finding that also nonadherent cells are rich in osteoprogenitors, can differentiate into osteoblasts in vitro (Long et al. 1990), and are able to form bone when injected into the kidney parenchyma of mice (Wlodarski et al. 2004). As a consequence, some authors suggested an alternative approach in which bone marrow aspirated from the iliac crest was directly injected into the lesion site without ex-vivo culture expansion (Hernigou et al. 2006). To date no long-term data on the effectiveness of this surgical procedure are available, even though it has been suggested that nonadherent bone marrow cells have a proliferative and differentiative potential comparable to that of adherent bone marrow cells, and could improve the MSC bone regenerative properties (Wan et al. 2006). Knowledge about the biological features of nonadherent cell compartment opens exciting perspectives in basic research on bone tissue repair and regeneration. Indeed, it may or may not influence the choice of manipulating bone marrow aspirates, in order to obtain an adequate amount of osteogenic MSCs as an essential requirement for the success of bone tissue engineering or cell therapy. Relatively few papers have been published on this issue, and the results obtained by various authors suggest that in nonadherent bone-marrow compartment there is an undifferentiated quiescent subpopulation of mesenchymal progenitors that may become adherent in vitro, begin to proliferate, and differentiate into diverse tissue lineages besides bone, as well as rescue in vivo lethally irradiated mice (Falla et al. 1993; Eipers et al. 2000; Dominici et al. 2004; Mödder and Khosla 2008; Zhang et al. 2009). Nevertheless, the difference among the ‘early adherent’ (EA), ‘nonadherent’ (NA), and ‘late adherent’ (LA) bone marrow stromal cells has not been fully elucidated. In particular, there is currently little information available regarding the molecular mechanisms that govern the osteogenic potential and differentiation process of NA subpopulations. The aim of this study was to analyze the phenotypic, functional and molecular features of two subpopulations of late adherent osteoprogenitors derived from nonadherent bone marrow cells usually discarded during in vitro MSC expansion, and to compare their osteogenic properties with those of EA-MSC.

Materials and methods

Human bone marrow cell cultures and replating of nonadherent cells

Bone marrow samples were collected by reaming the metaphysis and proximal diaphysis of the femur of six donors (men, aged from 20 to 81 years), who had been scheduled for reconstructive joint surgery. The sample collection was approved by the Institutional Ethical Committee for human research. Mesenchymal stromal cells were obtained as previously reported (Ciapetti et al. 2006). Samples were filtered to remove bony fragments, diluted, and stratified on Ficoll-Hypaque gradient (Sigma–Aldrich, Milan, Italy) to collect mononuclear cells (MNC). MNC were seeded at a density of 250,000 cell/cm2, and maintained in μ-MEM supplemented with 10% FBS, 100 U/mL penicillin, 0.1 mg/mL streptomycin, and 100 μmol/L ascorbic acid-2 phosphate (Sigma) (basic medium) at 37°C in a 5% CO2-humidified atmosphere. After 4 days, NA cells (NA0) were removed, and replated at the same seeding density. The EA cells were re-fed with fresh medium along with the addition of 10−8 M dexamethasone (differentiation medium). Cells from the NA0 population which adhered to TCPS 4 more days later were defined as LA cells (LA0). Once again, the NA cells (NA1) were removed, and treated as NA0 to gain a new subset of LA cells, termed LA1 (Fig. 1). In order to obtain the formation of mineral nodules, 10 mM β-glycerophosphate (Sigma) was added to the culture medium (mineralization medium) after the second confluence. The protocol used for cell cultures and the time points selected to investigate the osteogenic properties of the different subpopulations are reported in Table 1.
Fig. 1

The picture shows the scheme for obtaining LA-MSC cultures, and the percentage of cells recovered at each step in comparison to the previous one. EA early adherent cells, NA non adherent cells, LA late adherent cells

Table 1

Protocol for cell cultures and time points selected to investigate the osteogenic properties of the different subpopulations





Culture medium

Basic (αMEM + FBS + ascorbic acid)

Differentiation (basic + dexamethasone)

Mineralization (differentiation + β-glycerophosphate)

Time pointa






Main events

MNC/NA isolation

Nonadherent cell removing, addition of differentiation medium

1st confluence, 1st passage

2nd confluence, addition of mineralization medium




4 days

Variable (14.2–15.3 days)

Variable (4.0–6.2 days)

14 days


Cell count, FC

Cell count, GEA

ALP activity, CFU assay, FC, GEA

ALP activity, calcium, FC

ALP activity, calcium, GEA, mineralization assay

ALP alkaline phosphatase, CFU colony forming unit, FC flow cytometry, GEA gene expression analysis

aTime points correspond to the main event occurring and characterization assays performed; more than one timepoint is needed to complete cell activity induced by specific medium composition (in brackets)

Characterization of cell cultures

At each time point, 0.04% trypan blue exclusion dye was used to analyze cell proliferation by cell counting and to discriminate the live from dead cells.

Immunophenotyping of MSC was performed at seeding, T1 and T2 by using monoclonal antibodies conjugated with fluorescein isothiocyanate (FITC), or R-Phycoerythrin (RD1), or RD1-Cyanin 5.1 (PC5) (Instrumentation Laboratory, Milan, Italy). They included CD45-FITC (leukocyte common antigen), CD166-RD1 (activated leukocyte cell adhesion molecule-ALCAM), CD90-PC5 (Thy 1), CD44-FITC (hyaluronic-acid receptor), CD105-RD1 (endoglin, TGF-β1 receptor), and CD117-PC5 (c-kit, stem cell factor receptor). Monoclonal antibodies and cells (105/test) were incubated for 20 min at 4°C, and the proportion of positive cells was evaluated on 10,000 events by using a flow cytometer EPICS XL-MCL (Beckman Coulter, Fullerton, CA, USA). Alkaline phosphatase (ALP) was measured using a biochemical method (Sigma, N7653) at T1, T2, and M1 in cell lysates obtained with 0.01% SDS and the ALP activity expressed as nanomoles of p-nitrophenol formed per minute normalized to the cell number.

The colony-forming unit (CFU) assay was performed at T1 as previously described (Ciapetti et al. 2006). After 14 days, duplicate wells were stained with cytochemical ALP (Sigma, kit no. 86-R). Aggregates with more than 25 cells were scored as small CFUs (25–50 cells) or large CFUs (>50 cells), and the percentages of CFUs with more than 50% of ALP positive cells were calculated (Veyrat-Masson et al. 2007).

The synthesis of type I collagen was assessed by measuring its metabolic product released in the culture supernatant. Levels of C-terminal propeptide of type I collagen (CICP) were quantified by enzyme immunoassay according to the manufacturer’s instructions (Quidel corporation, Heidelberg, Germany).

The deposition of extracellular matrix (ECM) was analyzed by Transmission Electronic Microscopy (TEM) on 1-week cultures in mineralizing medium on Permanox chamber slides (Nunc, Thermo Fisher Scientific); cell were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer pH 7.4 for 1 h, postfixed with 1% osmium tetroxide, dehydrated in a graded series of ethanol, and embedded in Epon. Samples were sectioned both longitudinally (‘on face’) and transversally. Thin sections were stained with uranyl acetate, tannic acid and lead citrate, and finally observed with a Zeiss EM 109 transmission electron microscope.

Mineralization was assessed after 14 days of culture in mineralization medium. The calcium content was quantified by a commercial colorimetric assay (Roche Diagnostics, Milan, Italy), and the results expressed as ng Ca2+ equivalents per cell. For the visualization of mineralized matrix areas, fixed cultures were stained with von Kossa/safranin O. Calcium deposits (Von Kossa staining) appear as dark amorphous aggregates, while safranin O stains proteoglycans (Karahuseyinoglu et al. 2007).

The gene expression was analyzed at T0, T1, and M1 by quantifying the transcripts of genes useful to monitor the MSC differentiation to osteoblasts (Table 2). RNA was extracted with a RNeasy mini kit (Qiagen, GmbH, Hilden, Germany) and the retrotranscription was performed with MuLV Reverse Transcriptase (Applied Biosystems, Foster City, CA, USA). Real-time PCR was performed by amplifying 1 μg of cDNA using the Light Cycler instrument and the Universal Probe Library system (Roche Applied Science, Monza, Italy) (Mouritzen et al. 2005). Probes and primers were selected using a web-based assay design software (ProbeFinder http://www.roche-applied-science.com). The results were expressed as a ratio between gene of interest and GAPDH reference gene.
Table 2

List of primers and probes selected to analyze the expression of genes related to the bone cell differentiation



NCBI reference number


Primer sequence (5′–3′)


Detection limit (μg)


Glyceraldehyde 3-phosphate dehydrogenase











Alkaline phosphatase






















Type 1 collagen, α1 chain











Type 12 collagen, α1 chain












Cartilage oligomeric matrix protein






















Frizzled 8











Bone sialoprotein






















Runt-related transcription factor 2

































Statistical analysis

Quantitative results were expressed as arithmetic mean plus or minus the standard error of the mean (SEM). The analysis of variance (Kruskal–Wallis test) was applied to detect the effects of the early/late adherence on the quantitative results. Mann–Whitney test was performed as a post hoc test of the multiple analyses, or as unpaired comparison for two independent variables. The p-values equal or below 0.05 were considered significant.


Phenotypic characterization of early and late adherent MSC subpopulations

After the first 4 days of culture, the percentages of plastic-adherent cells (EA) and NA0 cells suspended in the supernatant were highly variable, ranging from 45 to 97.5% and from 2.5 to 55.5% of the initial MNC population, respectively (Fig. 1). During the further 4 days of culture, 36.7–86.3% of the NA0 cells adhered to TCPS, becoming LA0 and corresponding to 11.2% ± 5 (range 0.9–34.8%) of the whole MNC population. The new NA cells (NA1) ranged from 13.7 to 63.3% of the total MNC, and a variable proportion (7.6–16.6%) became adherent during the subsequent 4 days. This last subpopulation (LA1), corresponded to 3.7% ± 3 of the whole MNC, ranging from 0.3 to 17.7%. No significant differences were found in the cell viability of all nonadherent subpopulations, being 92.8% ± 2 in MNC, 94% ± 2 in NA0 and 86.4% ± 4 in NA1. The proliferation rate of EA, LA0 and LA1 cells was similar, because they reached the first confluence on average in 2 weeks, i.e., 15.2 ± 2.9, 14.2 ± 2.5, and 15.3 ± 4.7 days, respectively. The adherent cells of all the subpopulations assumed a spindle-shaped morphology, and no differences in cell size or spatial distribution were visually observed.

The immunophenotype was characterized before the seeding, when EA, LA0 and LA1 coincided with MNC, NA0 and NA1, respectively (Table 3). In all the subpopulations, CD90 expression was lower than 2%, while CD44 was detectable in more than 25% of the cells. The percentage of CD45-positive cells, as well as of CD105 and CD166, was higher in NA0 and NA1 than in MNC, but in any case the differences were not statistically significant. At T1 and T2, confluent EA cells showed high levels of CD44, CD90, CD105, CD166 (>90%), while the percentage of CD45 and CD117 was very low (<1%). Confluent LA0 and LA1 highly expressed CD44, CD166, CD90, and CD105, but the complete disappearance of CD45 positivity was obtained only at T2 time point. The expression of CD90 and CD166 was significantly lower in first confluent LA0 than in EA (p = 0.02), while, at the second confluence, the expression of MSC markers was similar in all the subpopulations.
Table 3

Percentage (mean value ± SEM) of cells expressing markers typically positive (CD44, CD90, CD105, CD166) and negative (CD45, CD117) in MSC









26.3 ± 19

29.8 ± 21

27.1 ± 16


1.9 ± 1

0.4 ± 0.1

1.8 ± 0.9


5.3 ± 5

18.0 ± 11

11.3 ± 6


2.6 ± 2

14.8 ± 10

21.7 ± 15


39.5 ± 3

68.1 ± 19

64.2 ± 17


0.5 ± 0

0.5 ± 0.3

0.8 ± 0.8



99.0 ± 0.5

84.2 ± 10

86.8 ± 9


99.8 ± 0.1

88.8 ± 7 p = 0.02 versus EA

93.3 ± 3


90.3 ± 2

86.1 ± 7

86.8 ± 1


99.6 ± 0.1

74.4 ± 13 p = 0.02 versus EA

85.1 ± 7


0.4 ± 0.1

12.1 ± 6

10.4 ± 8


0.2 ± 0

1.6 ± 1

0.7 ± 0.3



94.6 ± 3

86.3 ± 6

96.1 ± 2


97.9 ± 0.8

91.8 ± 4

98.6 ± 1


94.9 ± 2

88.7 ± 5

94.3 ± 1


90.8 ± 5

94.2 ± 3

92.1 ± 5


1.6 ± 1

0.3 ± 0.1

1.0 ± 0.5


1.1 ± 0.7

1.2 ± 1

1.3 ± 0.7

aAt seeding EA, LA0 and LA1 corresponded to MNC, NA0 and NA1, respectively

The ALP content largely varied in all the cultures (Fig. 2a). The enzyme activity was always detectable in EA samples, while it was often under the detection limit in LA cells. The ALP reduction was statistically significant at T2 for all the LA cells (p = 0.03), and at M1 only for the LA0 subpopulation (p = 0.04). The temporal changes in ALP production were different in EA and LA cells: the enzyme activity of EA increased significantly from T1 to T2, and tended to decrease during the mineralization process. On the contrary, LA subpopulations showed a continuous, slight increase in ALP activity, which persisted also after the addition of the mineralization medium.
Fig. 2

(a) Biochemical activity of ALP. The enzyme activity was expressed as mean ± SEM of the ratios between ‘nanomoles of substrate converted in 1 min’ and ‘cell number’. EA-T2 versus LA0-T2 and LA1-T2: p = 0.03; EA-M1 versus LA0-M1: p = 0.03; (b) Collagen type I production quantified in the supernatant by ELISA. Graph visualized mean ± SEM of ‘μg of collagen’ normalized to ‘cell number’

The type I collagen production was higher in EA cells in comparison with LA cells, but, due to the variability in EA results, the difference was not significant; collagen deposition steadily increased during the differentiation phase (p < 0.05), while slightly decreased after the addition of β-glycerophosphate, in all populations (Fig. 2b).

Functional characterization of early and late adherent MSC subpopulations

All the subpopulations were able to form colonies, even though the total number of CFU slightly decreased in LA cells compared to EA (Fig. 3a). The proliferation rate of clonogenic LA0 cells seemed to be greater, as suggested by the higher number of large CFUs compared to EA and LA1 cells (Fig. 3b). The mean number of ALP positive colonies in EA and LA0 was similar, but it decreased in LA1 cells (Fig. 3c). We found that the proportion of colonies with more than 50% ALP-positive cells was higher in LA0 cells than in the other subpopulations (EA: 21.4 ± 11%, LA0: 34.3 ± 10%, and LA1 9.3 ± 7%), even if differences were not statistically significant.
Fig. 3

CFU assay. Bars represent the mean ± SEM of CFUs generated by EA, LA0, and LA1. Data on the total number (a), size (b), and ALP positivity (c) of CFUs were normalized to number of seeded cells. The representative pictures demonstrate that CFUs reside in all the subpopulations, and their ability to differentiate in bone-forming cells is suggested by the presence of ALP-positive colonies

Morphological analysis with TEM highlighted that in all subpopulations, under mineralizing conditions, elongated cells were able to organize in 5–8 layers, as observed on cross sections (Fig. 4a–c). Among cells, a consistent extracellular matrix was evident in all samples and EA cells showed the thickest layer of ECM. Collagen was easily recognized in all cell cultures and the typical bandage could be noted at high magnification. The presence of collagen fibrils in intercellular spaces and between cell layers was observed in both sections. On face images added details on fibrillar architecture, showing that in small gaps between cells the fibrils were organized as parallel bundles, whereas in wider spaces they were less regularly arranged (Fig. 4d–f).
Fig. 4

TEM analysis of extracellular matrix architecture. Multilayers of EA cells (a, d), LA0 (b, e) and LA1 (c, f) were sectioned both perpendicularly (ac) and on face (df); Collagen packed into fibrils is evident in all subpopulations

EA and LA cells showed similar mineralization ability (Fig. 5). Three-dimensional structures recognized as mineral nodules were evident with Von Kossa staining, as well as dense matrix deposition areas and ‘spiral’ arrangement of confluent multilayered cells were highlighted by safranin O counterstaining. To evaluate the biomineral formation, the deposition of calcium was determined by dissolving the mineral matrix in HCl. Mineral deposition tended to increase after the addition of mineralization medium, and after 2 weeks the difference in calcium content between T2 and M1 was significant in LA0 (p = 0.004) and LA1 (p = 0.003), but not in EA (p = 0.59). Indeed, at the M1 end point the highest calcium level was observed in LA1 (p = 0.03 vs. EA), followed by LA0 (p = 0.04 vs. EA).
Fig. 5

Mineralization assays. The calcium content was measured on cell lysates and expressed in the graph as mean ± SEM of the ratios between ‘ng Ca2+ equivalents’ and ‘cell number’. The representative pictures show in all the subpopulations the presence of mineral nodules, as dark deposits stained with von Kossa. The safranin O counterstaining highlights the osteoid areas

Molecular characterization of early and late adherent MSC subpopulations

The whole gene expression analysis is summarized in Fig. 6.
Fig. 6

Gene expression analysis. Graphs show the temporal changes in the transcription levels of a number of bone-related genes useful to monitor the differentiation of bone-forming cells. Data are expressed as mean ± SEM of the ratios between ‘gene of interest’ and ‘housekeeping gene (GAPDH)’

RUNX2 transcript levels did not show significant differences in EA and LA cells. The ALPL expression was lower in LA at the first confluence (EA vs. LA1 p = 0.02). During mineralization the ALPL transcription continued to rise in LA cells, thus approaching the production of EA cells at T1. The basal expression of COL1A1 was significantly higher in LA than in EA (T0: LA0 vs. EA p = 0.05; LA1 vs. EA p = 0.007), while no significant differences were observed at T1 and M1. At T0, COL12A1 expression was more evident in LA1 than in other subpopulations (p < 0.001 vs. EA; p = 0.03 vs. LA0), but also LA0 cells showed an increased expression compared to EA (p = 0.04). An inverse trend was observed at T1, as the COL12A1 expression in LA1 was significantly lower than in EA cells (p = 0.01). At the end of the mineralization process, the expression dramatically decreased in EA cells, while it was stable in LA subpopulations. With regard to the non-collagenous proteins, the expression of BGLAP, CLEC3B, and COMP was similar in EA and LA during the whole culture time. On the contrary, the expression of IBSP, POSTN, and SPARC was significantly different in LA cells. At T0, the gene expression was higher in LA than in EA cells (IBSP: EA vs. LA0 p = 0.02; POSTN: EA vs. LA0 p = 0.03, EA vs. LA1 p = 0.003; LA0 vs. LA1 p = 0.02; SPARC: EA vs. LA0 p = 0.002, EA vs. LA1 p = 0.06). In EA cells, transcripts of the above proteins significantly increased from T0 to T1 (about 2 log), and after the addition of mineralization medium they remained stable. In LA cells, the temporal change was less pronounced, despite being extended until M1, when the transcript levels were the same as those of EA cells at the first confluence. In EA cells the highest expression of TNFRSF11B was found at T1, while in LA the increase was observed at M1. Finally, the FZD8 expression tended to increase during the mineralization phase, but it was similar in EA and LA cells.


In this study, we found that the bone marrow in vitro contains variable proportions of NA cells which acquire the ability to adhere to TCPS later. The property of late adhesion is not irrelevant, because it was found in approximately 35% of the marrow-derived MNC. Similar results were obtained by other authors, who showed that the collection of NA subpopulations allowed the total number of bone marrow-derived MSC to increase up to 36–37% (Wan et al. 2006). Phenotypic and functional tests proved that the differentiation process of LA-MSC was similar to the one of EA-MSC, even if apparently delayed. A possible explanation for a delay in the differentiation process could be the high proportion of cells belonging to the hematopoietic lineage found in the NA subpopulations. In fact, over 60% of the replated cells were positive for the CD45 pan-leukocyte antigen, and this marker disappeared only after the second confluence, while in the EA-MSC it was lost earlier. In addition, biochemical and molecular data showed that the ALP production of LA cells was reduced in comparison with EA-MSC, but the enzyme activity in EA cells decreased after the first confluence, while in LA subpopulations it steadily rose until the mineralization endpoint. Furthermore, type I collagen production resulted higher in EA cells, as supported by a more abundant layer of extracellular matrix observed at TEM morphological analysis.

The immunophenotype analysis showed that the percentage of CD90 and CD166 positive cells was significantly lower in the first confluent LA0 subpopulation. Other authors found a decreased expression of CD90 in MSC with more marked osteogenic properties (Wiesmann et al. 2006), and it is reasonable to assume that LA0 cells have such capability. The clonogenic assay partially confirmed this hypothesis, because we found that the total number of CFU slightly diminished in LA cells, but LA0 exhibited a higher number of large CFUs, and a higher proportion of colonies with more than 50% ALP positive cells. TEM images suggested in all the subpopulations an active production of an extracellular matrix providing sites for deposition of calcium phosphate. Mineralization assay further confirmed the ability of cells to form mineral nodules, but the significantly higher calcium accumulation in LA versus EA cultures suggested a high osteogenic potential of LA cells. These results are in agreement with recent studies which showed that the osteogenic activity is tenfold higher in MSC derived from the NA cells than in EA-MSC (Dominici et al. 2004); as well, Wlodarski et al. (2004) observed that floating stromal cells depleted of their adherent component gave better results in terms of bone formation than freshly isolated bone marrow cells.

The gene expression analysis was very useful to understand whether the molecular mechanisms governing the osteogenic differentiation varied between EA and LA cells. Notably, the results of gene expression analysis were more reproducible than those observed with the other biochemical and functional tests. The expression of some genes, namely RUNX2, BGLAP, CLEC3B, COMP, and FZD8, was similar in EA and LA cells throughout the whole period of culture. Excluding the transcription factor RUNX2 which is involved in the transcriptional regulation of osteoblast commitment and differentiation (Huang et al. 2007) the other genes are upregulated during the late phase of the differentiation process (Boskey et al. 1998; Di Cesare et al. 2000; Iba et al. 2001; Katoh 2008). The increased expression of the above genes, also after the induction of mineralization, is consistent with the hypothesis that the differentiation process is ongoing in both EA and LA cells.

At T0, both the LA subpopulations showed higher levels of bone-related genes in comparison with EA, namely COL1A1, IBSP, SPARC, and POSTN. This finding suggests that the NA cells are enriched in osteoprogenitors, and agrees with the hypothesis of more marked osteogenic properties of LA cells, which is further supported by the upregulation of IBSP and POSTN at the end of the culture period. Both Type I collagen and the non-collagenous proteins coded by IBSP and SPARC are major structural proteins which regulate many biological functions of bone-cells, and are expressed in areas of active remodeling (Motamed 1999; Ogata 2008), while POSTN, even if expressed also in other tissues, plays a critical role in osteoblast differentiation (Litvin et al. 2004). In EA-MSC, the expression of ALPL, COL12A1, SPARC, and TNFRSF11B was significantly raised from T0 to the first confluence, whereas in LA cells the upregulation was detected later, during the mineralization process. COL12A1 promotes the interactions between Type I collagen fibrils and the surrounding matrix (Wälchli et al. 1994), and TNFRSF11B encodes osteoprotegerin, whose expression levels increase during the differentiation to osteoblasts (Gori et al. 2000). The behavior of these genes confirms once again that the osteogenic differentiation of LA is delayed.

All data so far discussed confirm the presence of quiescent osteoprogenitors in NA bone marrow stromal cells. In our hands, the adhesion to TCPS is a triggering event by which the NA cells activate a proper transcriptional program to reach the osteoblast phenotype, and exhibit marked osteogenic properties. The ability to adhere seems to be acquired under particular conditions, i.e., the depletion of osteogenic precursors that is obtained in vitro after the early adhesion of bone marrow stromal cells. The existence of a reservoir of osteoprogenitors recruitable when the most differentiated component is exhausted, also suggested by transplantation assays, may have a key role in the bone pathophysiology and tissue regeneration and repair (Baksh et al. 2004; Zhang et al. 2009). One accredited hypothesis is that the NA bone-marrow cells are able to enter the circulation and develop a fundamental role in cortical bone remodeling (Mödder and Khosla 2008). They increase in number during augmented bone formation, such as during pubertal growth, or following fractures, or in conditions of augmented bone remodeling, such as aging (Mödder and Khosla 2008; Kumagai et al. 2008; Eghbali-Fatourechi et al. 2007).

In conclusion, this study confirms that the nonadherent cell population of bone marrow cultures contains osteogenic precursors, and these cells differentiate to osteoblasts when they become adherent following a depletion of the early adherent MSCs. The late adherent MSCs exhibit osteogenic properties similar to or better than the EA-MSCs. Bone tissue engineering and cellular therapy strategies for injured or genetically defective bone could take advantage of LA cells, since an adequate amount of MSCs with enhanced osteogenic potential may be obtained, while avoiding bone marrow manipulation and cell culture expansion.

Further studies are necessary to better understand the clinical role of these quiescent osteoprogenitors which could play a crucial role in bone homeostasis as well as in bone repair and regeneration.


This study was funded by grants from the Italian Ministry of the Health. The Authors thank Nicoletta Zini (Laboratory for cell biology and electron microscopy, Rizzoli Orthopedic Institute, Bologna, Italy) for her help with TEM analysis.

Copyright information

© Springer-Verlag 2009