Portions of this work were presented at the 1995 meeting of the American Society for Bone and Mineral Research and the 1995 meeting of the International Society for Experimental Hematology.
Bone formation and hematopoiesis are anatomically juxtaposed and share common regulatory mechanisms. However, little is known about the interrelationship between these two processes. We have previously shown that the senescence accelerated mouse-P6 (SAMP6) exhibits decreased osteoblastogenesis in the bone marrow that is temporally linked with a low rate of bone formation and decreased bone mineral density. Here we report that in contrast to decreased osteoblastogenesis, ex vivo bone marrow cultures from SAMP6 mice exhibited an increase in the number of colony-forming unit adipocytes, as well as an increase in the number of fully differentiated marrow adipocytes, compared with SAMR1 (nonosteopenic) controls. Further, long-term bone marrow cultures from SAMP6 produced an adherent stromal layer more rapidly, generated significantly more myeloid progenitors and produced more IL-6 and colony-stimulating activity. Consistent with this, the number of myeloid cells in freshly isolated marrow from SAMP6 mice was increased, as was the number of granulocytes in peripheral blood. The evidence that SAMP6 mice exhibit decreased osteoblastogenesis, and increased adipogenesis and myelopoiesis, strongly suggests that a switch in the differentiation program of multipotential mesenchymal progenitors may underlie the abnormal phenotype manifested in the skeleton and other tissues of these animals. Moreover, these observations support the contention for the existence of a reciprocal relationship between osteoblastogenesis and adipogenesis that may explain the association of decreased bone formation and the resulting osteopenia with the increased adiposity of the marrow seen with advancing age in animals and humans.
OSTEOBLASTS ORIGINATE from multipotent mesenchymal stem cells which also give rise to adipocytes and fibroblastic stromal cells of the bone marrow as well as chondrocytes and muscle cells.1 In ex vivo cultures, these progenitors proliferate and differentiate to form colonies of fibroblastoid cells, designated colony forming unit-fibroblasts (CFU-Fs). Under specific culture conditions and in the presence of certain factors, a particular differentiation pathway can be stimulated. However, the interrelationships of the differentiation of the various lineages and the precise mechanisms for determining the commitment to each lineage are poorly understood.
It is now well appreciated that aberrant osteoclast and osteoblast development in the bone marrow, relative to the demand for these cells, is a fundamental problem in osteoporosis. Thus, an oversupply of osteoclasts, relative to the need for remodeling, and an undersupply of osteoblasts, relative to the need for cavity repair, are critical pathogenetic changes in postmenopausal and age-related osteopenia, respectively.2 Using SAMP6 mice, a model of early senescence and osteopenia, we have previously demonstrated an association between decreased osteoblastogenesis in the bone marrow, with a decreased rate of bone formation and low bone mineral density.3
In contrast to the negative effect of aging on osteoblast production, it is well documented that the proportion of fatty marrow increases with age.4–6 Thus, in neonatal mammals, adipocytes are all but absent in the bone marrow, which at this stage is primarily hematopoietic; but with advancing age, the number of adipocytes increases, resulting in the appearance of fatty marrow. By the third decade of life in humans, most of the femoral cavity is occupied by fat.
Changes in mesenchymal cell differentiation in the bone marrow, besides determining the rate of production of the progeny of mesenchymal progenitors, should also affect hematopoiesis because this process depends heavily on support provided by certain fibroblastic/stromal and adipocytic cells as well as osteoblastic cells.4,7,8 This dependency is most likely due to the ability of these cells to produce growth factors and cytokines that act in a paracrine fashion to influence the differentiation of hematopoietic progenitors. It has been recently appreciated that the same sets of cytokines and growth factors can also act in an autocrine fashion to modulate the differentiation of mesenchymal progenitors as well.9–11
Prompted by our earlier findings that osteoblastogenesis is defective in SAMP6 mice, we employed this model to examine whether the development of other progeny of mesenchymal precursors is also affected in these animals. Specifically, we sought to determine whether adipogenesis was altered. In addition, and because of the evidence for the dependency of myelopoiesis on mesenchymal cell differentiation,12 we also examined myeloid cell differentiation.
MATERIALS AND METHODS
Alpha modified essential medium (α-MEM) was obtained from ICN Biochemicals (Aurora, OH, U.S.A.); fetal bovine serum (FBS) was from Hyclone (Logan, UT, U.S.A.); RPMI 1640, horse serum (HS), methylcellulose, β-mercaptoethanol, isobutylmethylxanthine (IBMX), hydrocortisone, indomethacin, Oil Red O, and Thyazolyl blue (MTT) were from Sigma (St. Louis, MO, U.S.A.). Fluoresceinisothiocyanate (FITC)-labeled F4/80 monoclonal antibody was purchased from CALTAG Laboratories (San Francisco, CA, U.S.A.). Recombinant murine IL-6 was from Upstate Biotechnology Inc. (Lake Placid, NY, U.S.A.); and the factor-dependent mouse-mouse 7TD1 hybridoma cell line was from American Type Culture Collection (Rockville, MD, U.S.A.). The Fast Track mRNA Isolation Kit was obtained from Invitrogen Corp. (San Diego, CA, U.S.A.); and nylon membranes were purchased from Bio-Rad Laboratories (Richmond, CA, U.S.A.). The Decaprime Labeling Kit was from Ambion Corp. (Austin, TX, U.S.A.). Autoradiography was done on X-ray film (XAR-5) purchased from Kodak (Rochester, NY, U.S.A.).
Senescence accelerated mice, SAMP6 and the control SAMR1 strain, were obtained from a colony established from breeders, kindly provided by Dr. Takeda at Kyoto University, Kyoto, Japan. Animals were maintained in accordance with National Institute of Health (NIH) guidelines on the care and use of laboratory animals. Three to 5-month-old female animals were used in the studies. For analysis of peripheral blood, samples were collected by cardiac puncture under CO2 anesthesia. Packed red cell volume was measured using a microhematocrit. White blood cell count was performed in a hemocytometer, and blood smears were used for differential counts of peripheral white blood cells. Marrow cells were removed from femora and tibiae by flushing with α-MEM. Unless indicated otherwise, combined femoral and tibial cells from each animal were analyzed separately. Cell numbers were determined using a model ZF Coulter counter (Coulter Electronics, Hialeah, FL, U.S.A.).
Analysis of bone marrow cells
Differential bone marrow cell counts were performed on cytospin preparations according to established criteria.13 Flow cytometric analysis was performed with a Coulter Flow Cytometer Profile (Coulter Corporation), using FITC-labeled F4/80 monoclonal antibody. For colony forming units-granulocytes and macrophages (CFU-GM) analysis, bone marrow cells (25 × 103) were cultured on 35-mm dishes (Corning, Corning, NY, U.S.A.) in the presence of 10% murine lung conditioned medium, 0.8% methylcellulose, and 30% FBS in a humidified 5% CO2 atmosphere at 37°C. Colonies of ≥50 cells were counted on day 7 after the initiation of culture with an inverted microscope. Adherent CFU-Fs were assayed by culturing marrow cells at 1 × 106/10 cm2 in α-MEM containing 15% FBS.14 One-half of the medium was replaced with fresh medium on days 5 and 11. Cells were fixed with methanol and stained with Giemsa on day 14. Colonies of more than 50 cells were counted as CFU-Fs. Adipogenesis was stimulated in CFU-F cultures by adding 0.5 mM IBMX, 60 μM indomethacin, and 0.5 μM hydrocortisone on day 11.15 Cells were fixed on day 14, and adipocytes were visualized by staining with Oil Red O and counterstaining with Harris-hematoxylin. CFU-F colonies with more than 10% of cells staining positive for Oil Red O were considered as colony forming units-adipocytes (CFU-ADs).
Dexter type long-term bone marrow cultures (LTBMCs) were established as we have previously described to analyze hematopoietic activity of the bone marrow in vitro.14 Briefly, bone marrow cells were cultured in 12.5 cm2 (Falcon, Meylan, France) in α-MEM containing 25% HS, 10−6 M hydrocortisone in a 5% CO2 atmosphere at 33°C. At weekly intervals, the supernatant was removed, the cells were recovered by centrifugation, and the supernatant cells were counted. Cultures were refed with 60% fresh medium and 40% of cell-free supernatant.
Colony stimulating activity (CSA) in LTBMC was measured by assessing colony number when 5 × 104 BMCs were cultured for CFU-GM in the presence of 10% LTBMC supernatant.16 Supernatants were collected at 0 and 24 h after medium change, when the peak of CSA production has been previously shown to occur.16 Interleukin-6 (IL-6) in LTBMC supernatants was measured 12 h after medium change using the factor-dependent mouse–mouse 7TD1 hybridoma cell line and the MTT colorimetric assay.17 IL-6 was quantified using a standard curve set up with known amounts of recombinant murine IL-6.
Lipoprotein lipase assay
Three weeks after establishing LTBMC from SAMP6 and SAMR1 mice, the cell monolayers were rinsed with PBS and incubated for 1 h at 37°C in α-MEM containing 10 U/ml of heparin. Lipoprotein lipase (LPL) activity was determined using [3H]triolein as described previously.18 Briefly, after incubation for 1 h at 37°C, the reaction was stopped by the addition of a mixture of chloform-methanol-heptane, and liberated [3H]free fatty acids were separated and quantitated by liquid scintillation.
mRNA isolation and Northern blot analysis
Poly(A+) RNA was prepared from LTBMC stroma, established 3 weeks previously, employing a Fast Track mRNA isolation kit (Invitrogen). mRNA, 1.5 μg, was electrophoresed on 1.2% agarose gels containing 1.9% formaldehyde and blotted onto a nylon membrane.19 Specific mRNAs were detected on Northern blots using cDNA inserts representing a 1430 bp fragment of murine LPL and a 700 bp insert for glyceraldehyde-3-phosphate dehydrogenase (GAPD). Human cDNA for GAPD was used because the homology of these transcripts between the two species is greater than 80%. Probes were labeled to high specific activity using the Decaprime Labeling Kit (Ambion Inc., Austin, TX, U.S.A.), and hybridization was performed under high stringency conditions (hybridization: 16 h at 65°C, 0.57 M Na+; two washes, 20 minutes each, were at 65°C, 0.26 M Na+, followed by the third 20 minute wash at 65°C 0.1 M Na+). Filters were exposed to X-ray film, and autoradiographic bans were quantitated within the near-linear range with a model 300 Å laser densitometer using Image Quant software (Molecular Dynamics, Sunnyvale, CA, U.S.A.) and corrected for variations in loading, using GAPD.
Results are expressed as mean ± SEM. Statistically significant changes were detected using Student's t-test.
In experiments where bone marrow cells were cultured in 15% FBS, the number of adherent colonies containing adipocytes (i.e., CFU-ADs) was 14.7 times higher in SAMP6 than in SAMR1 (Table 1), whereas the number of CFU-F colonies was indistinguishable between the two strains. In parallel cultures, bone marrow cells were maintained in the presence of IBMX, indomethacin, and hydrocortisone in order to stimulate adipogenesis. Under these conditions, the number of CFU-ADs was 5-fold higher in cultures derived from SAMP6 compared with SAMR1 (Table 1). The difference in CFU-AD numbers was more pronounced when the results were expressed per animal (i.e., femora and tibiae), as compared with when they were corrected for the number of cells seeded at the initiation of each culture. This was due to differences in the cellularity of bone marrow aspirates from SAMP6 as compared with SAMR1. Increased cellularity of the bone marrow aspirates from SAMP6 mice was primarily due to an increase in myelopoietic elements including recognizable mitotic and postmitotic myeloid pools such as promyelocytes, myelocytes, metamyelocytes, and polymorphonucleocytes (Table 2). However, the number of monocyte/macrophages, determined by flow-cytometric analysis of marrow using an antibody against F4/80, was similar in the two strains. Similarly, the number of erythroid cells were indistinguishable in the two strains; but, when erythroid cells were expressed as a percentage of the total marrow cell population, they were significantly lower in SAMP6 because of the increase in myeloid cells. In separate experiments, we found that the number of CFU-GM per femur + tibia was similar in SAMP6 and SAMR1 mice (25 ± 4 × 103 vs. 18 ± 3 × 103, respectively, n = 4 per group, p = 0.17), as we had previously published.3
Table Table 1.. COLONY FORMING UNITS-ADIPOCYTES (CFU-AD) IN BONE MARROW FROM SAMP6 AND SAMR1
Table Table 2.. ANALYSIS OF BONE MARROW CELLULARITY
In LTBMCs maintained in the presence of 25% horse serum and 1 μM hydrocortisone, cells from SAMP6 formed an adherent monolayer capable of supporting hematopoiesis more rapidly than cells from SAMR1 as demonstrated by a more abundant adherent layer in SAMP6 (Figs. 1A and 1B). At 3 weeks after initiation of the cultures, the adherent cell number was significantly higher in SAMP6 as compared with SAMR1 (2.65 ± 0.06 × 106 vs. 1.70 ± 0.15 × 106, respectively [p < 0.05]). As shown in Figs. 1C and 1D, the LTBMC stroma from SAMP6 also contained more adipocytes than SAMR1. Consistent with this observation, mRNA of lipoprotein lipase, a marker of adipogenesis, was increased in SAMP6 (Fig. 1E); and LPL activity was higher in SAMP6 than in SAMR1 (38.3 ± 4.0 nmol/ng of protein vs. 14.6 ± 1.3 nmol/ng of protein, respectively [p = 0.001]).
Faster stroma formation in SAMP6 was accompanied by increased production of the nonadherent myeloid cells, which are found in the supernatant of these cultures. As shown in Fig. 2A, significantly more cells in the culture supernatant were produced by cultures from SAMP6 mice compared with SAMR1 after 4 weeks. Cumulative production after 10 weeks averaged 25.27 ± 1.75 × 106 cells in SAMP6 compared with 10.65 ± 0.78 × 106 cells in SAMR1 (p < 0.001). To establish that hemotopoietic progenitors were present in the supernatant cells, we measured CFU-GMs at weekly intervals (Fig. 2B). At each time point, the number of CFU-GMs per flask was higher in SAMP6 and paralleled that of total supernatant cell number. After 7 weeks, CFU-GMs in SAMP6 were 1943 ± 176 versus 259 ± 29 in SAMR1 (p < 0.0001). In addition, both CSA and IL-6 activity in supernatants of LTBMCs from SAMP6 mice at 4 weeks were significantly increased (Fig. 3). However, in a separate experiment, when corrected for the number of adherent cells, we found that IL-6 levels in LTBMC culture supernatants were similar, indicating that increased IL-6 as well as CSA in SAMP6 cultures are related to increased cell proliferation.
Finally, comparison of peripheral blood counts in SAMP6 and SAMR1 mice revealed a slight increase in packed cell volume in the former (43.7 ± 1.1% vs. 40.5 ± 0.8%, respectively; p < 0.05). As shown in Table 3, the total white cell count was not different, but peripheral blood neutrophils were elevated whether expressed as neutrophil count or as a percentage of total white blood cells. Lymphocyte, basophil, and eosinophil counts were similar in the two strains.
Table Table 3.. PERIPHERAL BLOOD COUNT
Evidence obtained previously has strongly suggested that the bone defect exhibited by SAMP6 mice, a model of low turnover osteopenia, is due to a premature decrease in the ability of mesenchymal progenitors of the bone marrow to differentiate toward the osteoblast lineage.3 The results presented in this paper demonstrate that besides decreased osteoblastogenesis, other aspects of mesenchymal stem cell differentiation, as well as hematopoiesis, are abberant in these animals. Specifically, our findings indicate that the bone marrow of SAMP6 mice, at 3–5 months of age, exhibits increased adipogenesis as well as increased myelopoiesis that is associated with an increase in granulocyte counts in peripheral blood. Increased adipogenesis in ex vivo cultures of marrow cells from these mice has also been observed in studies by Suda and colleagues.20
Because of the descriptive nature of the experimental results shown here, we cannot exclude the possibility that the coexistence of decreased osteoblastogenesis on the one hand and increased bone marrow adipogenesis and myelopoiesis on the other hand are unrelated. Nonetheless, several lines of evidence support the contention that the decreased osteoblastogenesis and increased adipogenesis, seen in the bone marrow of SAMP6 mice, are interrelated phenomena that may well reflect a reciprocal relationship between the two processes. Indeed, the existence of a common progenitor for osteoblasts and adipocytes has been demonstrated by several groups using primary cultures of bone marrow, as well as clonal cell lines, capable of differentiating either to osteoblasts or adipocytes, when maintained under appropriate culture conditions.15,21–23 Furthermore, Beresford et al. have shown that in bone marrow cultures from adult rats, addition or removal of steroids at different times of the culture led to reciprocal changes in adipogenesis and osteoblastogenesis.24 In other words, stimulation of the former process led to a decrease in the latter, and vice versa. In line with these findings, we have recently established clonal cell lines from SAMR1, the nonosteopenic control strain for SAMP6, and found that they are indeed capable of either forming fat-laden adipocytes or mineral depositing osteoblasts, depending on the culture conditions.25 Consistent with the in vitro evidence that common progenitors can differentiate either toward the osteoblastic or adipocyte lineage, an inverse relationship between bone marrow fat and bone mass has been observed in elderly osteoporotic humans, as well as in animal models of bone loss induced by immobilization, weightlessness, ovariectomy, or glucocorticosteroid excess.26–29
The stromal cells of the bone marrow exhibit an extensive phenotypic overlap with osteoblastic cells, and we and others have referred to them as cells of the stromal/osteoblastic lineage.2,8 Heretofore, it has remained unclear whether the bone marrow stromal cells that provide support for myelopoiesis are different from the cells that differentiate into osteoblasts. The observation that there is increased myelopoiesis in the marrow of SAMP6 mice, whereas osteoblastogenesis is decreased, suggests that the support of myelopoiesis is provided by cells that are distinct from those destined to become fully differentiated osteoblasts.
In contrast to the inverse relationship between osteoblastogenesis and myelopoiesis, we found that in the SAMP6 mice, the latter process is positively correlated with adipogenesis. To the extent that the LTBMCs reflect the in vivo situation, our data suggest that increased stromal cell proliferation and the consequent increase in IL-6 and CSA are responsible in part for the increased myelopoiesis observed in SAMP6 mice. This finding is consistent with evidence that marrow adipocytic cell lines support granulopoiesis12 and that agents that stimulate adipogenesis also stimulate myelopoiesis,30,31 whereas agents that suppress adipocyte differentiation inhibit myelopoiesis.10
We had previously demonstrated that the defect in osteoblast development in the bone marrow of SAMP6 mice is responsible for a secondary defect in osteoclast development and that loss of sex steroids fails to up-regulate osteoclastogenesis in these mice.3 In addition, we have determined more recently that in this model of defective osteoblastogenesis, loss of sex steroids does not up-regulate bone remodeling and does not cause loss of bone.32 The number of CFU-GMs, the common progenitor of osteoclasts, macrophages, and neutrophils, were not different in SAMP6 as compared with SAMR1 controls. These pieces of evidence taken together with the current observation that whereas osteoclastogenesis is decreased, myelopoiesis is increased in SAMP6, suggests that the adipogenic cells that promote the later stages of myeloid cell differentiation to macrophages and neutrophils are different from the cells that support commitment to the osteoclast lineage. In addition, our previous finding that osteoclastogenesis is suppressed in SAMP6 mice,3 even though results of the present and other studies33 indicate that IL-6 is elevated in these mice, suggests that normal osteoblast differentiation is required for IL-6 to exert its osteoclastogenic effects. This notion is supported by evidence that stimulation of osteoclast formation by IL-6 requires the expression of the IL-6 receptor on stromal/osteoblastic cells.34
A number of factors produced in the bone marrow microenvironment are capable of influencing not only hematopoiesis but also osteoblastogenesis and adipogenesis.9–11 In the present study, we found that CSA and IL-6 are increased; and in preliminary reports, Kodama and coworkers obtained evidence that the production and/or the action of IL-11 are deficient in the bone marrow of SAMP6.35 At this stage, it is not known whether the cytokine changes seen by us and Kodama et al. are the cause of the cellular changes or epiphenomena reflecting the presence of different mesenchymal cell constituents in the marrow of SAMP6 versus SAMR1. In any event, it is interesting to note that IL-6–type cytokines as well as members of TGF-β superfamily inhibit adipogenesis but stimulate osteoblastogenesis.9–11,36,37 Hence, a causative role of these agents in the cellular changes seen in the bone marrow of SAMP6 mice is at least theoretically possible.
In conclusion, the results of the present studies suggest that changes in the differentiation of mesenchymal progenitors of the bone marrow may not only provide a potential mechanism for the development of senile osteoporosis but may also explain the cellular basis for the association between osteoporosis and increased bone marrow fat. Moreover, our findings add to the validity of the SAMP6 mouse as a model of age-related osteopenia.
The authors thank Virginia Fitzhugh and Elena Moerman for excellent technical assistance, Mary Jo Moody for flow cytometry analysis, and Dr. Philip Kern for LPL cDNA probe. This work was supported by the National Institutes of Health (PO1 AG/AR 13918, AG09458; AR43003), the Department of Veterans Affairs, and the Arkansas Experimental Program to Stimulate Competitive Research (funded by the National Science Foundation, the Arkansas Science and Technology Authority and the University of Arkansas for Medical Sciences).