Cell Culture
D1 cells were maintained in Dulbecco modified Eagle medium (GibcoBRL, Gaithersburg, Maryland) containing 10 per cent fetal bovine serum (Hyclone Laboratories, Logan, Utah), fifty milligrams of sodium ascorbate per milliliter, and antibiotics (100 units of penicillin G per milliliter and 100 micrograms of streptomycin per milliliter) in a humidified atmosphere of 5 per cent carbon dioxide at 37 degrees Celsius. For all experiments, the D1 cells were plated at 1 x 104 cells per square centimeter in tissue-culture dishes. The medium was changed after twenty-four hours and every forty-eight hours thereafter. Under these conditions, the cells reached confluence by three to five days, and the experiment was started at this time-point, which was designated day zero. To study the effect of the concentration of carbon dioxide and fetal bovine serum on the differentiation of adipocytes, parallel cultures of D1 cells were maintained in medium containing 20 per cent fetal bovine serum in an atmosphere of 10 per cent carbon dioxide.
Differentiation of D1 Cells into Adipocytes
Medium containing 10-9, 10-8, or 10-7-molar dexamethasone (Sigma Chemical, St. Louis, Missouri) was added to the cells. The cells were treated with dexamethasone for either forty-eight hours, after which they were maintained in culture without the steroid, or the entire study period. Some cells were not treated with dexamethasone and served as controls. Cell morphology was monitored on the basis of the appearance of cytoplasmic lipid droplets as seen with a phase-contrast microscope and was photographed with Kodak Technical Pan Film (Eastman Kodak, Rochester, New York).
Cells in culture were fixed in 10 per cent buffered formalin phosphate (Hydrol Chemical, Yeadon, Pennsylvania) for ten minutes and were stained with Sudan IV for two minutes with use of hematoxylin as a counterstain. In each dish, an area of twenty-two square millimeters was examined under a Leitz microscope (Allendale, New Jersey) equipped with a color video camera and frame grabber. Nine areas in each dish were sampled, with the first area located at the approximate center of the dish and the remaining eight located in the intersection of an imaginary grid with eight-millimeter vertical and horizontal spacings. Each area, which consisted of a square of 200 by 200 square micrometers at a resolution of 106 pixels per 100 millimeters, was sampled as a digital image at each intersection point. Adipocytes were counted within the sampled images with use of an unbiased counting frame calibrated with pixel dimensions of 100 by 100 square millimeters. The average number of adipocytes in each well was calculated and was used as the basis for statistical analysis.
Analysis of the RNA
Cells in culture were homogenized in Trizol reagent (GibcoBRL). Total RNA was separated with chloroform and was precipitated with isopropyl alcohol. After the pellet had been washed with 75 per cent ethanol, the RNA was redissolved in RNAse-free water and the concentration was determined by measuring absorbance at 260 nanometers. Twenty micrograms of total RNA, denatured in 2.2-molar formaldehyde, 50 per cent formamide in 1x 3-(N -morpholino)propanesulfonic acid buffer at 65 degrees Celsius for ten minutes, was separated by electrophoresis on a 1.2 per cent agarose gel containing 2.2-molar formaldehyde in 1x 3-( N-morpholino)propanesulfonic acid running buffer (125 volts for two hours), was transferred overnight by capillary blotting to a Zeta Probe membrane (Bio-Rad, Hercules, California) in 20x saline, sodium phosphate, and EDTA buffer (3.6-molar NaCl, 0.2-molar Na2HPO4 7H2O, and twenty-millimolar EDTA), and was cross-linked to the membrane by irradiation with ultraviolet light for one minute.
Dr. Daniel Lane (Johns Hopkins University, Baltimore, Maryland) provided 422(aP2) cDNA (700 base pairs), Dr. David Rowe (University of Connecticut, Farmington, Connecticut) provided rat cDNA pa1R1 (1600 base pairs), and Dr. John Wozney (Genetics Institute, Cambridge, Massachusetts) provided osteocalcin cDNA (500 base pairs). These were labeled with [32P]deoxycytidine triphosphate with use of a random-primed DNA labeling kit (Boehringer, Mannheim, Germany) to at least 1 x 108 counts per minute and were used for hybridization at 65 degrees Celsius overnight in a hybridization buffer containing 1 per cent bovine serum albumin, 0.25-molar sodium phosphate, 7 per cent sodium dodecyl sulfate, one-millimolar EDTA, and 100 micrograms of denatured salmon sperm DNA per milliliter. The membrane was washed twice for ten minutes each in 2X 3.6-molar saline, 0.2-molar sodium phosphate, and twenty-millimolar EDTA-1 per cent sodium dodecyl sulfate at room temperature, and then it was washed three times in 0.1X 3.6-molar saline, 0.2-molar sodium phosphate, and twenty-millimolar EDTA-0.1 per cent sodium dodecyl sulfate, with the first two of these three washes at room temperature for ten minutes each and the final wash for one hour at 65 degrees Celsius. Membranes were exposed to intensifying screens, which were scanned on a PhosphorImager, and the radioactivity was quantitated with use of ImageQuant Software (both from Molecular Dynamics, Sunnyvale, California). Probes were removed by washing the membrane twice for one hour each at 70 degrees Celsius with a solution containing 0.25-molar Tris, 0.2-molar EDTA, and 20 per cent sodium dodecyl sulfate. The RNA on the membranes was then hybridized with other cDNA probes.
Differentiation of D1 Cells into Adipocytes
D1 cells are a relatively homogeneous population of pluripotential mesenchymal cells. At confluence, the cells become polygonal and form a tightly packed monolayer. When confluent monolayers were treated with increasing (10-9, 10-8, and 10-7-molar) concentrations of dexamethasone for forty-eight hours, the cells accumulated triglyceride vesicles, which were small initially and increased in size with time. In contrast, cells that were not treated with dexamethasone exhibited osteogenic properties. On the sixth and eighth days, the cells containing triglyceride vesicles were clearly distinguishable from the surrounding cells both on phase-contrast microscopy and on staining with Sudan IV, a stain for neutral lipids in fat cells (Figs. 1-A, 1-B, and 1-C). The number of adipocytic cells in culture after treatment with the steroids was dependent on the concentration of dexamethasone and on the duration of treatment; it was highest in the dishes that were treated with 10-7-molar dexamethasone continuously and was lowest in the dishes that were treated with 10-9-molar dexamethasone for the initial forty-eight hours only (Table I and Figs. 2-A, 2-B, 2-C, and 2-D). Adipogenic changes were not found in the D1 cells that had not been treated with dexamethasone. Furthermore, changing the conditions of the cell culture from 10 to 20 per cent fetal bovine serum and from 5 to 10 per cent carbon dioxide did not produce the adipocytes observed in culture with dexamethasone; this suggests that changes in the concentration of fetal bovine serum and carbon dioxide did not have a substantial effect on adipogenesis.
Expression of Adipose-Specific 422(aP2) mRNA
To investigate the effect of dexamethasone on gene expression, D1 cells were treated continuously for six days, starting at day zero, with 10-9, 10-8, or 10-7-molar dexamethasone. Total RNA analyzed by Northern blot hybridization with 422(aP2) cDNA showed that the expression of 422(aP2) mRNA with 10-7-molar dexamethasone was two and five times greater than that with 10-8 and 10-9-molar dexamethasone, respectively, whereas no 422(aP2) mRNA was detectable in the D1 cells that had not been treated with the steroid (Fig. 3).
To evaluate the effect of the duration of treatment with dexamethasone on the expression of 422(aP2) mRNA, D1 cells were treated with 10-7-molar dexamethasone either continuously from day zero or for an initial forty-eight hours, after which they were maintained in culture without dexamethasone, and the total RNA was analyzed. The expression of 422(aP2) mRNA in the cells treated with 10-7-molar dexamethasone for forty-eight hours reached a maximum level by two days and declined by four days, whereas the expression in the cells treated continuously continued to increase with time, reaching a maximum level between four and six days (Fig. 4). The results indicate that continuous treatment with dexamethasone produces an increasing level of 422(aP2) mRNA expression, whereas the cells from which dexamethasone has been removed after an initial treatment period show diminished expression.
Osteoblastic Gene Expression during Steroid-Induced Adipogenic Differentiation
Beginning at day zero, D1 cells in culture were treated either with or without 10-7-molar dexamethasone for six days. Total RNA hybridized with 422(aP2) cDNA showed expression of 422(aP2) mRNA in D1 cells treated with the steroid, whereas 422(aP2) mRNA was not detectable in the cultures of control cells that had not been treated with the steroid. After the 422(aP2) cDNA had been removed from the membrane, it was reprobed with a1 type-I collagen cDNA, which showed decreased expression of a1 type-I collagen mRNA in D1 cells that had been treated with dexamethasone. Northern blot hybridization with osteocalcin cDNA also showed that treatment with dexamethasone decreased expression of osteocalcin mRNA in D1 cells (Fig. 5).
In the absence of dexamethasone, the pluripotential mesenchymal cell D1 is osteogenic and differentiates mainly into osteoblasts. In the presence of dexamethasone, differentiation into osteoblasts is decreased and differentiation into adipocytes is greatly increased. The rapid appearance of 422(aP2) mRNA after treatment with the steroid suggests that the expression of this gene, which encodes a putative lipid-binding protein6, is one of the events that is followed by the accumulation of fat-containing vesicles with increasing duration of treatment. Only occasional adipocytes were observed in cultures that had not been treated with steroids. This is in contrast to all previously described preadipocyte cell lines, such as 3T3-L1 and 3T3-F422A, which differentiate under normal culture conditions into lipid-containing cells if the cultures are maintained after the cells reach confluence24-26,57. Thus, dexamethasone appears to be required for the expression of the fat-cell phenotype in the D1-cell line that was isolated from bone marrow.
A group of cDNAs corresponding to messages whose levels are greatly increased during adipocyte differentiation has been isolated7,40,51. One of these cDNAs, pAL422, which encodes an adipocyte homologue of myelin P2, has sequence homology to a class of fatty acid-binding proteins found in lipogenic tissues7,12. Expression of the 422 gene is transcriptionally activated during differentiation of 3T3 preadipocytes, which leads to a fifty to 100-fold accumulation of its 0.7 kilobase mRNA, which is expressed during differentiation of adipocytes6,12,13. It has been demonstrated that the 5'-flanking region of the 422(aP2) gene contains elements that mediate activation by glucocorticoid and cyclic adenosine monophosphate13, agents that promote differentiation of 3T3-L1 preadipocytes.
The D1 clone isolated from mouse bone marrow is derived from a cell that is pluripotential and differentiates into progeny with both osteogenic and adipogenic properties in vitro and in vivo17,19. Under conditions that promote differentiation into chondrocytes, D1 cells also have the ability to express the cartilage phenotype2. These observations support the hypothesis that there is a common precursor for several cell lines, including osteogenic, adipogenic, and chondrogenic lineages5,19. However, expression of the fat-cell phenotype in vitro occurred only after the addition of dexamethasone and was dose-dependent, indicating that dexamethasone is at least one of the factors required for these bone-marrow cell cultures to change from a primarily osteogenic nature to an adipogenic nature. The results indicate that glucocorticoids may control the processes of pluripotential cells differentiating into adipocytes. They may also stimulate the transdifferentiation of osteoblast-like cells into adipocytes5. However, the regulation of adipogenesis may involve complex mechanisms and interactions between multiple regulatory elements20,61,65.
The increased use of steroids for immunosuppression after organ transplantation, as treatment for rheumatoid diseases, and for chemotherapy has resulted in an increased risk of osteonecrosis1,8-10,22,30,33,43,52,60. Glucocorticoids have been the focus of studies on the pathogenesis of osteonecrosis. Although statistical data show that steroids may be implicated in one-third of all cases of osteonecrosis3,34,50, the precise mechanism of action of the steroids has not been determined3,15,38,47,59, to our knowledge.
More than ten hypotheses have been proposed3,36,47,59. Fat embolism35, microfracture16, intraosseous hypertension31, vasculitis48, and intravascular coagulation37 are some well accepted theories. Fat-cell hypertrophy and abnormal metabolism of fat have been demonstrated both in patients who have osteonecrosis and in animals that have been treated with steroids41,62. The results of the present study indicate that fat-cell hypertrophy and hyperplasia in bone marrow may be a direct result of treatment with steroids.
Hormones are clearly involved in the control of a variety of processes during development11. Dexamethasone has been shown to accelerate the appearance of adipocytes in some of the preadipocyte cell lines11,27,44,55,56 and to stimulate the expression of several mRNA species that are specific to adipose tissue11,54. Dexamethasone has been used to induce adipocyte differentiation in some of the preadipocyte cell lines7,13,46,56 and in the bone-derived28 cell line RCG3.1. In the present study, we demonstrated that the multipotential bone-marrow cell line D1 responds to treatment with dexamethasone by differentiating into fat cells and by expressing an adipose-specific gene, 422(aP2), which was enhanced by treatment with increasing concentrations of the steroid and extended time-periods. The results indicate that glucocorticoids not only stimulate but also regulate the process of adipogenesis in bone-marrow stroma. Decreased expression of a1 type-I collagen mRNA and osteocalcin mRNA indicates a decreased potential for matrix production by the D1 cells and a consequent diminution of osteogenic properties resulting from treatment with dexamethasone.
During the differentiation of bone cells from progenitor cells, dexamethasone appears to maintain the adipocyte phenotype28. Our data are consistent with reports that glucocorticoids affect osteoblast gene expression by downregulating type-I collagen and osteocalcin18,21. However, dexamethasone has also been reported to promote the expression of markers of the osteoblast phenotype in cultures of rat marrow stromal fibroblasts45. Glucocorticoids can have various effects on bone cells, depending on the state of cellular differentiation, the species of animal, and the dose of and duration of treatment with steroids29,53. The effect on human marrow cells is to inhibit matrix synthesis4.
Our findings are relevant with regard to a number of pathological conditions. An increase in the fat volume in bone marrow and a decrease in hematopoietic cells and bone tissue have been reported in patients who have osteoporosis49 and in models of steroid-induced osteonecrosis in animals39,42,63. The relationships between adipogenesis and osteonecrosis, osteoporosis, and other pathological changes need to be investigated. Adipocytes form an integral part of the stromal system of bone and of marrow and participate in the establishment and maintenance of the intramedullary hematopoietic microenvironment. The source of adipocytes could be preadipocytes that are committed to adipocyte differentiation or stem cells that are inducible by exogenous stimulation13,21,27. Dexamethasone may enhance the differentiation of preadipocytes as well as pluripotential bone-marrow cells into adipocytes. Since adipocytes and osteoblasts share a common pool of stem cells, when exogenous stimulators shift the differentiation of marrow stem cells into the adipocyte lineage the stem-cell pool may not be sufficient to provide enough osteoblasts to meet the need for bone-remodeling, fracture-healing, or repair of necrotic bone. Furthermore, the increase in volume of fatty marrow with a concomitant increase in intraosseous pressure62,64 could decrease the blood flow in a semi-intact osseous compartment, eventually leading to ischemia and osteonecrosis31. Thus, steroid-induced adipogenesis in bone-marrow stroma and systemic changes in lipid metabolism may be major contributing factors in steroid-related osteonecrosis and osteoporosis.
NOTE: The authors are grateful to Dr. Daniel Lane and to members of his research group at Johns Hopkins University for valuable discussions, and to Dr. George Alheid for assistance with the image analysis.