Effect of platelet-derived growth factor on DNA synthesis and gene expression in bone marrow stromal cells derived from adult and old rats

The effects of platelet-derived growth factor (PDGF) on DNA synthesis and mRNA expression of osteoblast markers in marrow stromal cells derived from adult (6 months) and old (24 months) rats were examined. Treatment of stromal cells from adult rats with dexamethasone induced the appearance of osteoblast-like cells. PDGF partially also inhibited the differentiation of stromal cells induced by dexamethasone. In cultures of serum-starved stromal cells, PDGF stimulated [³H]-thymidine incorporation into DNA in a dose-dependent manner with a maximum stimulation of 15-fold at 500 ng/ml. By comparison, insulin-like growth factor (IGF-I) has a small effect on [³H] -thymidine incorporation. The effect of PDGF and IGF-I on DNA synthesis was additive. Treatment of the confluent stromal cells from adult rats with PDGF increased the mRNA level of osteopontin fourfold without any significant effect on alkaline phosphatase and type I collagen mRNAs. In contrast, dexamethasone stimulated the mRNA expression of alkaline phosphatase, type I collagen, and osteopontin 2.1-, 2.3-, and 14-fold, respectively. Addition of PDGF to dexamethasone-treated cells failed to induce any further increase in osteopontin expression whereas the expression of alkaline phosphatase and type I collagen was partially reduced. The expression of osteocalcin mRNA was negligible in stromal cells but stimulated several fold by dexamethasone and 1,25(OH)₂D₃. PDGF inhibited drastically the elevation of osteocalcin mRNA. In contrast, IGF-I stimulated type I collagen expression 100% without any appreciable effect on the expression of osteopontin and alkaline phosphatase. The stimulatory effect of PDGF on osteopontin expression was augmented by IGF-I. Furthermore, PDGF attenuated the stimulatory effect of IGF-I on type I collagen expression. The responses of cultured cells from old rats to growth factors were also examined. PDGF or PDGF plus IGF-I increased [³H]-thymidine incorporation in stromal cells from old rats but to a lesser extent. However, PDGF was equally effective in stimulating osteopontin expression in cells from both adult and old rats. We concluded that PDGF is a potent mitogen but that the response of stromal cells from old rats is impaired. In addition, PDGF stimulates osteopontin expression in stromal cells and this effect is not age dependent. © 1995 Wiley-Liss, Inc.     

Histamine production by alloantigen-activated mouse bone marrow cells

Histamine's contribution to the manifestations associated with graft-versus-host disease (GVHD) and/or hybrid resistance is unknown. Thus, we initiated studies to see whether or not mouse bone marrow cells could produce histamine upon alloantigen stimulation. Irradiated allogeneic spleen cells were shown to stimulate bone marrow cells to produce and secrete high levels of histamine. During 7 days of culture there was only a marginal increase in cell-associated histamine while the amount of histamine in the supernatant increased 10- to 20-fold. Optimal histamine production was dependent upon Lyt 1+2+ T cells resident in the bone marrow. Further, bone marrow cells from Nude mice failed to produce high levels of histamine following alloantigen stimulation. Soluble factors produced by alloantigen-stimulated bone marrow cells or by Con A-stimulated rat spleen cells induced high levels of histamine production in bone marrow cells in the absence of alloantigen. We suggest that histamine production by alloantigen-activated bone marrow cells may modulate immune functions following bone marrow transplantation.     

Regulation of osteoprotegerin secretion from primary cultures of human bone marrow stromal cells

Osteoprotegerin (OPG) is a soluble receptor for receptor activator of NF kappa B-ligand, a factor required for osteoclastogenesis. OPG secreted from bone marrow stromal cells is believed to inhibit osteoclast differentiation and several agents known to influence bone resorption have been demonstrated to regulate mRNA levels of OPG. In this report we have investigated the secretion of OPG protein from primary cultures of human bone marrow stromal cells. An ELISA was developed for measuring the concentration of OPG in culture medium. OPG secretion was decreased by 50% when the human bone marrow stromal cells were treated with 1 microM of prostaglandin E(2), possibly through activation of the protein kinase A-pathway since stimulation of protein kinase A by forskolin also inhibited OPG secretion. Treatment with phorbol 12,13 di butyrate, an activator of the protein kinase C-pathway, potently stimulated the secretion of OPG from human bone marrow stromal cells. The cells were also stimulated with inflammatory mediators and glucocorticoids. Treatment with interleukin-1 alpha (IL-1 alpha) and tumor necrosis factor-alpha (TNF-alpha) stimulated OPG secretion to 500% and 400% of control whereas dexamethasone decreased OPG production by 40%. In conclusion, an ELISA measuring OPG in cell culture media was developed. Using this ELISA, the amount of OPG secreted from human bone marrow stromal cells was clearly detectable, and the secretion of OPG-protein was potently regulated by prostaglandin E(2), forskolin, phorbol 12,13 di butyrate, IL-1 alpha, TNF-alpha, and dexamethasone.     

Influence of growth hormone on bone marrow adipogenesis in hypophysectomized rats

The hypophysectomized rat has been used as a model to study the effects of growth hormone deficiency on bone. Here, we have investigated the influence of growth hormone administration to hypophysectomized rats (HX) for 6 wk on accumulation of triglycerides in bone marrow and on the differentiation of primary marrow stromal cells into adipocytes under in vitro conditions. We found that hypophysectomy significantly increased triglyceride concentration in bone marrow, which was attenuated by growth hormone administration. Primary bone marrow stromal cells derived from HX rats also had more adipocytes at confluence compared with growth hormone-treated hypophysectomized (GH) rats. When stimulated with 3-isobutyl-1-methylxanthine plus dexamethasone (IBMX-Dex), preadipocyte colony counts increased more significantly in GH rats. Markers of adipocyte differentiation were higher in HX than in control or GH rats at confluence. However, after stimulation with IBMX-Dex, increased expression of markers was seen in GH compared with HX rats. In conclusion, growth hormone administration to hypophysectomized rats attenuated triglyceride accumulation in bone marrow and inhibited the differentiation of stromal cells into adipocytes in vitro.     

Effect of erythropoietin on the glucose transport of rat erythrocytes and bone marrow cells

The effect of Ep on radioactive glucose and methyl-alpha-D-glucoside transport by rat erythrocytes and bone marrow cells were studied. There is initial linearity followed by saturation kinetics of [14C]glucose transport by the erythrocytes of starved and starved plus Ep-treated rats at different concentrations of glucose. Starvation caused slight inhibition of glucose transport which increased markedly on Ep administration to starved rats. Normal animals failed to show any significant change in glucose transport after Ep treatment. Methyl-alpha-D-glucoside inhibited the Ep-stimulated glucose transport significantly. Ep also stimulated the transport of radioactive methyl-alpha-D-glucoside which was competitively inhibited in presence of D-glucose. Glucose transport in erythrocytes was found to be sensitive to metabolic inhibitors like azide and DNP. A sulfhydryl reagent and ouabain also inhibited the transport process. Ep stimulated glucose and methyl-alpha-D-glucoside transport in the bone marrow cells of starved rats. The sugar analog competitively inhibited the glucose transport in bone marrow cells and vice versa.     

Dexamethasone-induced inhibition of prostaglandin production dose not result from a direct action on phospholipase activities but is mediated through a steroid-inducible factor

Investigations were carried out to define the mechanisms of steroid-induced inhibition of prostaglandin secretion by rat renomedullary cells in tissue culture. Although it was strongly proposed that glucocorticoids may inhibit phospholipase A2 activity, we present several pieces of evidence against a direct action of dexamethasone on phospholipase activities. First, dexamethasone, which significantly decreases the release of labeled material from cells prelabeled with [3H]arachidonate, does not significantly alter the pattern of distribution of the radioactivity among the various classes of cell lipids. In addition, direct measurement of phospholipase A3 activity in dexamethasone-treated cells failed to show any significant decrease in the deacylation capacity. On the other hand, several indications suggest that dexamethasone may induce the secretion of a non-dialysable, transferable factor able to inhibit prostaglandin production, the mechanism of which remains to be investigated.     

Suppression of osteoprotegerin expression by prostaglandin E2 is crucially involved in lipopolysaccharide-induced osteoclast formation

LPS is a potent stimulator of bone resorption in inflammatory diseases. The mechanism by which LPS induces osteoclastogenesis was studied in cocultures of mouse osteoblasts and bone marrow cells. LPS stimulated osteoclast formation and PGE(2) production in cocultures of mouse osteoblasts and bone marrow cells, and the stimulation was completely inhibited by NS398, a cyclooxygenase-2 inhibitor. Osteoblasts, but not bone marrow cells, produced PGE(2) in response to LPS. LPS-induced osteoclast formation was also inhibited by osteoprotegerin (OPG), a decoy receptor of receptor activator of NF-kappaB ligand (RANKL), but not by anti-mouse TNFR1 Ab or IL-1 receptor antagonist. LPS induced both stimulation of RANKL mRNA expression and inhibition of OPG mRNA expression in osteoblasts. NS398 blocked LPS-induced down-regulation of OPG mRNA expression, but not LPS-induced up-regulation of RANKL mRNA expression, suggesting that down-regulation of OPG expression by PGE(2) is involved in LPS-induced osteoclast formation in the cocultures. NS398 failed to inhibit LPS-induced osteoclastogenesis in cocultures containing OPG knockout mouse-derived osteoblasts. IL-1 also stimulated PGE(2) production in osteoblasts and osteoclast formation in the cocultures, and the stimulation was inhibited by NS398. As seen with LPS, NS398 failed to inhibit IL-1-induced osteoclast formation in cocultures with OPG-deficient osteoblasts. These results suggest that IL-1 as well as LPS stimulates osteoclastogenesis through two parallel events: direct enhancement of RANKL expression and suppression of OPG expression, which is mediated by PGE(2) production.     

The induction of lymphokine synthesis and cell growth in IL 3-dependent cell lines using antigen-antibody complexes

Several IL 3-dependent murine bone marrow-derived cell lines can be stimulated to grow with antigen-antibody (Ag.Ab) complexes. The Ag.Ab complexes induced lymphokine gene expression and the synthesis of IL 2, GM-CSF, IL 3, and BSF-1 (IL 4). The lymphokines produced by these IL 3-dependent cells appeared to stimulate their own growth, as both IL 3 and BSF-1 (IL 4) stimulated the growth of IL 3-dependent cells. Ag.Ab complexes also stimulate the growth of primary cultures of bone marrow cells that have been previously activated with IL 3. Normal bone marrow, IL 2-, and GM-CSF-dependent bone marrow cell lines could bind Ag.Ab complexes, but binding did not result in the induction of lymphokine synthesis or cell growth. Hyperimmune serum from mice also stimulated lymphokine synthesis and cell growth in IL 3-dependent cells, and the stimulatory activity was removed by treatment with Staphylococcus aureus protein A, suggesting the presence of Ag.Ab complexes.     

Demonstration of altered signaling responses in bone marrow extracellular fluid during increased hematopoiesis in rats using a centrifugation method

The composition and characteristics of the bone marrow extracellular fluid supposedly modify the transport of cytokines, drugs, and other signaling molecules involved in the regulation of bone marrow function. Direct access to the bone marrow extracellular fluid surrounding hematopoietic cells is complicated by the virtually noncompliant surrounding bone tissue. We examined the applicability of a centrifugation method to obtain representative samples of bone marrow extracellular fluid from rats and humans. Perforated rat bones or human bone marrow biopsies were wrapped in nylon mesh baskets before being centrifuged at 180-239 g. In the rats, we found an only minor contribution of fluid from other sources than the bone marrow extracellular fluid as indicated by the average ratio of centrifugate-to-plasma activity of the extracellular tracer fluid 51Cr-labeled EDTA of 0.85. The colloid osmotic pressure in the centrifugate was consistently lower than that in the corresponding plasma in both species. In rats and humans, high-performance liquid chromatography showed a protein elution pattern from the bone marrow fluid similar to that of plasma, except for a peak eluting in the approximately 40-kDa molecular mass range. Western blotting of the cytokines erythropoietin and granulocyte colony-stimulating factor revealed generally higher amounts in the centrifugate than in the plasma. This difference was augmented during increased hematopoietic activity induced by inflammation or bleeding in rats. We conclude that the centrifugation method provides representative samples of bone marrow extracellular fluid and that extracellular signaling responses to altered hematopoiesis are more clearly reflected locally in the bone marrow interstitium than in plasma.     

Isozyme specificity of testosterone 7 alpha-hydroxylation in rat hepatic microsomes: is cytochrome P-450a the sole catalyst?

In the present study we show that monospecific antibody against cytochrome P-450a completely inhibits testosterone 7 alpha-hydroxylation in hepatic microsomes of untreated male or female rats or rats of either sex treated with dexamethasone. These data are in contrast with those of K. Nagata et al. (1987, J. Biol. Chem. 262, 2787-2793) who recently reported that an antibody prepared against cytochrome P-450a completely inhibited testosterone 7 alpha-hydroxylase activity in microsomes from untreated or 3-methylcholanthrene-treated rats but only inhibited 50% of the activity in microsomes from dexamethasone-treated rats. They proposed that dexamethasone treatment of rats induced another testosterone 7 alpha-hydroxylase in rat liver. The discrepancy in the two sets of data was due, at least in part, to the use of a chromatography system by Nagata et al. that is incapable of resolving a number of testosterone metabolites. Dexamethasone treatment of rats leads to a marked increase in the production of several testosterone metabolites, including 15 beta-hydroxytestosterone which is cochromatographic with 7 alpha-hydroxytestosterone in their chromatography system. Our results indicate that cytochrome P-450a accounts for all of the testosterone 7 alpha-hydroxylase activity in microsomes from dexamethasone-treated rats, and that testosterone 7 alpha-hydroxylation continues to be a useful marker for monitoring cytochrome P-450a in rat hepatic microsomes.     

New role for histamine in interleukin-3-induced proliferation of hematopoietic stem cells

This study investigated the effect of histamine generated by murine bone marrow cells in response to IL-3 on one particular biological activity of this growth factor, i.e., triggering of cells forming colonies in spleen (CFU-S) into S phase. Evidence is provided that i) IL-3-induced day-8 CFU-S cell cycling, evaluated by hydroxy-urea suicide, is completely abrogated when the binding of histamine to its H2 receptors is blocked by the specific antagonist oxmetidine, whereas cetirizine, a H1 receptor antagonist, is ineffective; and ii) the entry of day-8 CFU-S into S phase in response to IL-3 is likewise abolished when the histamine synthesis promoted by the growth factor is prevented by alpha-fluoromethylhistidine, a specific inhibitor of the histamine-forming enzyme, histidine decarboxylase. Similar results are obtained with both drugs, when a progenitor-enriched bone marrow population is used instead of total cells. Furthermore, i.v. injection of recombinant (r)IL-3 results within 2 hr in a substantial increase in bone marrow cell histamine synthesis together with triggering of day-8 CFU-S into cycle, the latter being completely abolished by a simultaneous injection of the H2 histamine receptor antagonist oxmetidine. Thus, our findings support the notion that both in vitro and in vivo the proliferation of early CFU-S in response to IL-3 is modulated by histamine via its H2 receptors. This conclusion is also consistent with the observation that dimaprit, a specific agonist of these receptors not only enhances the sensitivity of day-8 CFU-S to HU after a 2 hr incubation with bone marrow cells but also increases, to the same extent as IL-3, the number of colonies formed in irradiated spleens after a 5 hr pretreatment.     

On the mechanism of action of dexamethasone in a rat mast cell line (RBL-2H3 cells). Evidence for altered coupling of receptors and G-proteins

Prolonged exposure of rat basophilic leukemia (RBL-2H3) cells, a cultured analog of rat mast cells, to 0.1 microM dexamethasone resulted in global suppression of various stimulatory events in response to Ag and a global enhancement of the same stimulatory events to the adenosine analog, N-(ethylcarboxamide)adenosine (NECA). We had previously shown that Ag and NECA both activate phospholipase C but by different mechanisms; cells that had been treated with cholera or pertussis toxin, for example, responded to Ag but not to NECA with the release of inositol phosphates, increase in levels of cytosolic Ca2+, and secretion. Because the toxins still inhibited the responses to NECA in dexamethasone-treated cells, the effects of dexamethasone may have been exerted at the level of receptor/G-protein coupling rather than at the level of effector systems. Additional evidence for this was the following: 1) NECA-induced hydrolysis of the inositol phospholipids was still enhanced after permeabilizing (with streptolysin O or Staphylococcus alpha-toxin) and washing the cells; 2) the response to the G-protein stimulant, guanosine 5'-(3-O-thio)triphosphate was also enhanced in permeabilized, dexamethasone-treated cells and 3) binding and kinetic studies suggested that the enhanced responsiveness to NECA was attributable in part to an increase in receptor number. The suppressive action of dexamethasone on Ag-induced hydrolysis of inositol phospholipids, however, was readily lost by permeabilizing RBL-2H3 cells. The results indicate, therefore, that treatment with dexamethasone leads to changes in receptor-coupling mechanisms that are either resistant to (i.e., NECA-mediated responses) or reversed by (i.e., Ag-mediated responses) cell permeabilization.     

Regulating effect of bone marrow cells on human T-lymphocyte function

A study was made of the regulatory effect of human bone marrow cells in two experimental systems: lymphocyte proliferation in response to PHA, and spontaneous and PHA-induced production of macrophage migration inhibition factor (MIF) by peripheral blood lymphocytes. It was shown that bone marrow cells inhibit the proliferative activity of stimulated peripheral blood lymphocytes and induced MIF production. The effect of bone marrow cells on spontaneous MIF production was found to be inconclusive.     

Effect of 80 kDa protein of porcine follicular fluid on gonadotropin stimulated progesterone production in rat granulosa cells in vitro.

We reported the presence of a 80 kDa polypeptide in porcine follicular fluid that inhibited the binding of 125I-radiolabelled hFSH as well as hCG to the rat ovarian gonadotropin receptors. In the present study, the biological activity of the receptor binding inhibitor is determined using an in vitro bioassay procedure. Granulosa cells isolated from PMSG primed immature rat ovaries respond to exogenously added gonadotropins in terms of progesterone production. Addition of fractions containing the gonadotropin receptor binding inhibitory activity inhibited progesterone production stimulated by the gonadotropins in a dose-dependent fashion. The receptor binding inhibitory activity was also capable of inhibiting progesterone production stimulated by PMSG, which has both FSH- and LH-like activities in rats. In contrast, progesterone production stimulated by dbcAMP was not inhibited by the receptor binding inhibitor. This result indicates that the site of action of the inhibitor is proximal to the formation of the cAMP. The above observations point out to a possible role for this factor in modulating gonadotropin activity at the ovarian level.     

Inhibition of vesicular stomatitis virus replication in dexamethasone-treated L929 cells

We previously demonstrated that dexamethasone treatment of L929 cells inhibited plaque formation by vesicular stomatitis virus (VSV), encephalomyocarditis virus, or vaccinia virus. We now have characterized the antiviral effects of glucocorticoids in L929 cells. Dexamethasone did not directly inactivate VSV nor did steroid treatment of L929 cells affect virion adsorption or penetration. The VSV yield in L929 cells treated with dexamethasone for a period of only 4 or 8 hr was decreased by 50% when cells were infected the day following steroid treatment. Treating L929 cells with dexamethasone for a longer period resulted in greater inhibitions of virus synthesis. Interferon activity (less than 5 units/ml) was not detected in L929 cell culture fluids and cell sonicates from steroid-treated cells and the addition of antiserum to murine alpha/beta-interferon had no effect on the ability of dexamethasone to inhibit VSV replication. Dexamethasone treatment of L929 cells did not induce the production of double-stranded RNA-dependent protein kinase but did result in a slight elevation of 2-5A oligoadenylate synthetase activity, two enzymatic activities associated with the antiviral state induced by interferon. However, the elevated 2-5A synthetase activity was not associated with an inhibition of VSV RNA accumulation in dexamethasone-treated L929 cells. By contrast, the synthesis of all five VSV proteins was reduced by 50-75% in dexamethasone-treated L929 cells as early as 4 hr after infection. Thus, the dexamethasone-mediated inhibition of VSV replication in L929 cells is associated with decreased production of VSV structural proteins.     

Selective inhibition of arachidonic acid metabolite release from human lung tissue by antiinflammatory steroids

The effects of antiinflammatory steroids on arachidonic acid metabolite release from human lung fragments were analyzed. Incubation of lung fragments for 24 hr with 10(-6) M dexamethasone inhibited the net release of the prostacyclin metabolite 6-keto-PGF1 alpha, PGE2, and PGF2 alpha from lung fragments stimulated with anti-IgE but failed to inhibit the anti-IgE-induced release of PGD2, TXB2, and iLTC4. The IC50 of dexamethasone for inhibition of both spontaneous and anti-IgE-induced 6-keto-PGF1 alpha release was approximately 2 X 10(-8) M, and a 6-hr preincubation with the drug was required for 50% inhibition of prostaglandin release. Other agents were tested for activity in stimulating arachidonic acid metabolite release from human lung fragments. FMLP (fmet-leu-phe) stimulated the release of all metabolites tested (6-keto-PGF1 alpha, PGD2, PGE2, PGF2 alpha, TXB2, iLTC4); platelet-activating factor (PAF), but not lysoPAF, stimulated the release of PGD2, TXB2, and iLTC4. In contrast to the case with anti-IgE, where dexamethasone failed to inhibit net PGD2 and TXB2 release, the steroid inhibited the release of these metabolites stimulated by both FMLP and PAF. The steroid inhibited iLTC4 release induced by the highest concentration of PAF (10(-6)M) but did not inhibit iLTC4 release stimulated by either 10(-7) M PAF, FMLP, or anti-IgE. Because neither FMLP nor PAF caused the release of PGD2 or TXB2 from purified human lung mast cells, and because they also failed to induce histamine release from lung fragments, it is suggested that these stimuli produce PGD2 and TXB2 release in lung fragments through an action on a cell distinct from the mast cell. This suggestion is supported by the selective inhibition of the release of these arachidonic acid metabolites by dexamethasone. We suggest that the inhibitory action of steroids on arachidonic acid metabolite in human lung fragments contributes to their therapeutic efficacy in pulmonary diseases.     

Behavioral effects of L-tryptophan and p-chlorophenylalanine after adrenalectomy or dexamethasone administration in rats

Effects of acute administration of L-tryptophan (L-TRP. 250.0 mg/kg, i.p.) on active avoidance conditioning and "open-field" behavior were studied in male rats after adrenalectomy of dexamethasone administration. L-TRP inhibited the acquisition and reproduction of active avoidance reaction in adrenalectomized and dexamethasone-treated rats. Moreover, L-TRP decreased horizontal locomotor activity and grooming behavior in the "open field" on adrenalectomized rats. On the contrary, p-CPA restored the active avoidance conditioning in adrenalectomized rats and rats with excess of glucocorticoids. Also, p-CPA increased the total locomotor activity and grooming behavior in the "open field" in adrenalectomized rats, but decreased horizontal locomotor activity and enhanced emotional reaction in dexamethasone-treated rats in the "open field".     

Dexamethasone induction of osteoblast mRNAs in rat marrow stromal cell cultures

We have examined the ability of dexamethasone, retinoic acid, and vitamin D3 to induce osteogenic differentiation in rat marrow stromal cell cultures by measuring the expression of mRNAs associated with the differentiated osteoblast phenotype as well as analyzing collagen secretion and alkaline phosphatase activity. Marrow cells were cultured for 8 days in primary culture and 8 days in secondary culture, with and without 10 nM dexamethasone or 1 microM retinoic acid. Under all conditions, cultures produced high levels of osteonectin mRNA. Cells grown with dexamethasone in both primary and secondary culture contained elevated alkaline phosphatase mRNA and significant amounts of type I collagen and osteopontin mRNA. Addition of 1,25-dihydroxyvitamin D3 to these dexamethasone-treated cultures induced expression of osteocalcin mRNA and increased osteopontin mRNA. The levels of alkaline phosphatase, osteopontin, and osteocalcin mRNAs in Dex/Dex/VitD3 cultures were comparable to those of 1,25-dihydroxyvitamin D3-treated ROS 17/2.8 osteosarcoma cells. Omitting dexamethasone from either primary or secondary culture resulted in significantly less alkaline phosphatase mRNA, little osteopontin mRNA, and no osteocalcin mRNA. Retinoic acid increased alkaline phosphatase activity to a greater extent than did dexamethasone but did not have a parallel effect on the expression of alkaline phosphatase mRNA and induced neither osteopontin or osteocalcin mRNAs. In all conditions, marrow stromal cells synthesized and secreted a mixture of type I and III collagens. However, dexamethasone-treated cells also synthesized an additional collagen type, provisionally identified as type V. The synthesis and secretion of collagens type I and III was decreased by both dexamethasone and retinoic acid. Neither dexamethasone nor retinoic acid induced mRNAs associated with the chondrogenic phenotype. We conclude that dexamethasone, but not retinoic acid, promotes the expression of markers of the osteoblast phenotype in cultures of rat marrow stromal fibroblasts.