Cyclic AMP signaling in bone marrow stromal cells has reciprocal effects on the ability of mesenchymal stem cells to differentiate into mature osteoblasts versus mature adipocytes
- First Online:
- Cite this article as:
- Kao, R., Lu, W., Louie, A. et al. Endocrine (2012) 42: 622. doi:10.1007/s12020-012-9717-9
Stimulatory G protein-mediated cAMP signaling is intimately involved in skeletal homeostasis. However, limited information is available on the role of the cAMP signaling in regulating the differentiation of mesenchymal stem cells into mature osteoblasts and adipocytes. To investigate this, we treated primary mouse bone marrow stromal cells (BMSCs) with forskolin to stimulate cAMP signaling and determined the effect on osteoblast and adipocyte differentiation. Exposure of differentiating osteoblasts to forskolin markedly inhibited progression to the late stages of osteoblast differentiation, and this effect was replicated by continuous exposure to PTH. Strikingly, forskolin activation of cAMP signaling in BMSCs conditioned mesenchymal stem cells (MSCs) to undergo increased osteogenic differentiation and decreased adipogenic differentiation. PTH treatment of BMSCs also enhanced subsequent osteogenesis, but promoted an increased adipogenesis as well. Thus, activation of cAMP signaling alters the lineage commitment of MSCs, favoring osteogenesis at the expense of adipogenesis.
KeywordsCyclic AMP (cAMP) Osteogenesis Adipogenesis Parathyroid hormone (PTH) Mesenchymal stem cells Alkaline phosphatase
Fibrous dysplasia (FD, OMIM 174800) of bone is one of the manifestations of the McCune–Albright syndrome. Patients with McCune–Albright syndrome have activating missense mutations in GNAS, the gene encoding the α subunit of Gs [1, 2, 3]. These mutations have been demonstrated in various tissues, including bones [3, 4, 5] and samples from the monostotic form of fibrous dysplasia [6, 7]. In fibrous dysplasia, the expression of Gs protein and its transcript is upregulated during the maturation of precursor osteogenic cells to mature osteoblasts . Examination of the fibrotic area revealed an excess of cells with features of pre-osteoblasts, and the abnormal cellular features of fibrous dysplasia could be reproduced in vitro by the addition of excess exogenous cAMP to human osteogenic cells or by stable expression of constitutively active form of Gsα in human BMSCs [8, 9].
The role of the Gs/cAMP pathway in bone formation is further highlighted by several findings. Conditional deletion of the α subunit of Gs in osteoblast lineage cells resulted in reduced trabecular bone formation . Targeted expression in osteoblasts of constitutively active Gs-coupled G protein-coupled receptors (GPCRs) resulted in markedly increased trabecular bone mass [11, 12], whereas expression of a constitutively-active Gi-coupled receptor produced trabecular osteopenia . Furthermore, expression of an engineered constitutively active Gs-coupled receptor under the control of the osteoblast-specific Col1α1 2.3-kb promoter fragment in mice produced cellular features similar to that of the polyostotic fibrous dysplasia of McCune–Albright’s syndrome . Alteration of Gs and Gi signallings in osteoblast lineage cells was able to generate striking, but opposite effects on skeletal tissues.
Several extracellular regulators of skeletal function are known to exert their actions through GPCRs, which are located in the osteoblast membrane. The best-studied GPCR that regulates skeletal function is parathyroid hormone 1 receptor or PTH1R . This receptor is coupled to multiple G proteins, including Gs, Gq, and possibly Gi, resulting in the activation of diverse downstream effectors [15, 16, 17, 18]. Although PTH activates multiple signaling pathways, several lines of evidence indicate that PTH peptide fragments that specifically stimulate Gs signaling coupled to PTH1R could increase bone formation [19, 20, 21, 22]. PTH (1–34) was also able to regulate osteogenic development via cAMP signaling in a BMP-dependent mesenchymal differentiation system . PTH-induced signaling via Gs/cAMP pathway plays a particularly important role in skeletal function.
Mesenchymal stem cells are a pluripotent cell type that can differentiate into a variety of cell types, including adipocytes and osteoblasts. Lineage specification is achieved by the expression of transcription factors Cbfa1/Runx2 and Osterix in the case of osteogenesis, and PPARγ in the case of adipogenesis. The number of osteoblasts is partially determined by the lineage specification process. Mice haploinsufficient for PPARγ or Alox15, the lipoxygenase that generates oxidized lipid ligands of PPARγ, have high bone mass phenotype [24, 25]. The control of MSC differentiation into adipocytes and osteoblasts is important in maintaining a normal physiologic bone mass, but the signaling pathways and molecular mechanisms that regulate this process are not well understood.
Cyclic AMP/protein kinase A (PKA) signaling plays a prominent, but ambiguous, role in mesenchymal cell fate decision. Treating human MSCs with PTH inhibited adipocyte development induced by a specific cocktail of adipogenic inducers through the cAMP pathway . Activation of cAMP pathway in calcifying vascular cells derived from primary aortic medial cell cultures resulted in mineralization, enhanced alkaline phosphatase activity, and osteoblast-like differentiation . Pretreatment of human MSCs with a cAMP analog or forskolin, followed by transplantation, enhanced in vivo bone formation [28, 29]. Specific type 4 phosphodiesterase inhibitors could stimulate fibroblast-colony formation, alkaline phosphatase activity, and calcium deposition by rat bone marrow cells . In contrast, a high-throughput siRNA library screen identified components of cAMP pathway as osteogenic suppressors in human MSCs . Activation of PKA by IBMX and forskolin in transformed human MSCs enhanced adipogenesis, while downregulated gene expression of osteogenic gene markers by suppressing leptin . In another one of their reports, de Boer and coworkers  observed that cAMP signaling in MSCs was inhibitory on osteogenesis in rodent models, arguing that the difference seen in cAMP-stimulated osteogenic responses could be species-specific.
The role of cAMP pathway in mesenchymal cell fate decision and the subsequent cellular differentiation may depend on the developmental stage in which the cAMP signal is presented. It is important to consider the precise developmental stage of the cells in which the cAMP signal is activated. Activating cAMP signal in differentiating osteoblasts may promote a different osteogenic response from activating cAMP signal in MSCs. Likewise activating cAMP signal in differentiating adipocytes may also promote a different adipogenic response from activating cAMP signal in MSCs. Therefore, we sought to investigate the role of the cAMP signaling in regulating the differentiation of MSCs into mature osteoblasts and mature adipocytes.
Materials and methods
Bone marrow stromal cell (BMSCs) cultures
Primary bone marrow stromal cells were isolated from the femurs and tibiae of 8- to 10-week-old FVB/N male mice (The Jackson Laboratory). The cells were collected in primary culture medium (PCM) consisting of α-modification of Eagle’s medium, 10 % fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.25 μg/ml Fungizone (Gibco Cell Culture), and plated onto 10-cm cell culture dishes at a density of 7 × 106 cells/dish. Cells were incubated at 37 °C with 5 % CO2 and maintained undisturbed for five days to allow for cell attachment.
PTH, forskolin, and Wnt3a administrations
Human PTH [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34] (Bachem product # H-4835), forskolin (Sigma product # F6886), and recombinant mouse Wnt3a (R&D Systems product # 1324-WN) were dissolved in acetic acid (4.07 μM final concentration), ethanol (0.82 % final concentration), and PBS containing 0.1 % bovine serum albumin, respectively.
Day 5 to day 23 PTH treatment
After incubating BMSCs in PCM for five days, PCM was removed along with all non-adherent cells. The remaining adherent cells containing MSCs were exposed to 10−7 M hPTH in osteoblast differentiation medium (see “Initiation of osteoblast differentiation” section) from day 5 to day 23.
Day 5 to day 10 forskolin treatments
After removal of non-adherent cells, adherent MSCs were exposed to forskolin at final concentrations of 0.001, 0.01, 0.05, and 0.1 mM from day 5 to day 10 in osteoblast differentiation medium.
Day 0 to day 5 PTH, forskolin, and Wnt3a treatments
Day 0 to day 10 forskolin treatments
Freshly isolated BMSCs were exposed to forskolin at final concentrations of 0.001, 0.01, 0.05, and 0.1 mM from day 0 to day 5 in PCM. PCM containing forskolin was removed at day 5 along with non-adherent cells. Adherent cells were exposed to forskolin at these concentrations from day 5 to day 10 in osteoblast differentiation medium.
Initiation of osteoblast differentiation
After primary BMSCs had been maintained in primary medium for five days, the medium was aspirated and replaced with secondary osteogenic differentiation medium (primary medium containing 50 μg/ml ascorbic acid and 3 mM β-glycerolphosphate) to initiate osteoblast differentiation. Thereafter, secondary differentiation medium was replaced every two or three days (Fig. 1a).
Initiation of adipogenesis
BMSCs were plated in 6-well plates at 1 × 107 cells/well and remained in PCM for 10 days. At day 10, the medium was replaced with secondary adipogenic differentiation medium (addition of 1 μM rosiglitazone, 1 μM dexamethasone, 5 μg/ml insulin, and 500 μM IBMX), and cells were maintained in it for 2 days. After 2 days, this medium was replaced with PCM containing 5 μg/ml insulin, and cells were incubated in it for another 2 days. Cells were then maintained in PCM until day 19 (Fig. 1b).
Von Kossa and alkaline phosphatase (ALP) staining of osteoblasts and mineralized nodules
The presence of alkaline phosphatase activity was identified using Leukocyte Alkaline Phosphatase kit (Sigma product # 85L2-1KT). Two percent silver nitrate (Sigma product # S8157) solution was added to cell culture dishes for Von Kossa staining and UV-crosslinked for 10 min. Staining solution was then aspirated, and stained dishes were rinsed twice with distilled water. Stained cultures were scanned and quantified by means of Improvision Openlab software version 5.0.2.
Oil Red O staining for lipid accumulation in differentiated cells
Cells in 6-well plates were fixed in PBS containing 10 % formaldehyde. Following fixation, cells were stained in Oil Red O for 15 min. Cells were rinsed several times with distilled water to remove excess stain. Stained cultures were scanned and quantified by means of Improvision Openlab software version 5.0.2.
RNA extraction and quantitative real-time PCR (qRT-PCR)
Sequence of primers used in qRT-PCR
Data are presented as mean ± standard error of at least three independent experiments. Statistical differences between the means were examined by Student’s t test by means of GraphPad Prism 5 software, and significance was set at a P value of <0.05.
Effects of PTH and forskolin activation of cAMP signaling in differentiating osteoblasts on osteogenesis
Effects of PTH and forskolin activation of cAMP signaling in BMSCs on subsequent osteogenesis
To examine the effects of exposing MSCs to PTH on the subsequent osteoblast differentiation, freshly isolated BMSCs were exposed to PTH for five days. Before the induction of osteoblast differentiation, cells from bone marrows that had attached to culture dishes were allowed to differentiate later in the absence of PTH (Fig. 1a). We presumed that most of the attached cells were of mesenchymal origin and likely MSCs. BMSCs that were exposed to PTH were able to produce a 34 % increase in mineralization and a 27 % increase in the total level of ALP activity over vehicle-treated control cultures (Fig. 2b, c). Thus, treating BMSCs with PTH enhanced the ability of osteoblasts to mineralize and increased the number of ALP-positive cells.
Combined effects of forskolin activation of cAMP signaling in BMSCs and differentiating osteoblasts on osteogenesis
Effects of PTH and forskolin activation of cAMP signaling in BMSCs on subsequent adipogenesis
Next, we examined the effect of exposing BMSCs to PTH on subsequent adipogenesis. PTH treatment on BMSCs for five days actually produced a modest increase (27 %) in the subsequent adipogenesis as measured by Oil Red O staining (Fig. 7d, e). However, the total Oil Red O-positive colony number was not significantly affected (Fig. 7f). Unlike the effects of forskolin, PTH treatment on BMSCs did not reduce the subsequent adipogenesis.
Expression of early gene markers of osteogenesis and adipogenesis in MSCs following five-day forskolin treatment
Although forskolin treatment of BMSCs increased expression of genes associated with osteoblast differentiation, changes in the expression of these genes might only reflect the altered efficiency of osteoblast differentiation. Critical changes at the molecular level might have already occurred at earlier stages of differentiation process. Therefore, we examined changes in the expression of early marker genes before the induction of differentiation. Transcript levels of three early markers of osteogenesis (Runx2, Osterix, and ALP) and three early markers of adipogenesis (C/EBP, PPARγ, and ADRP) were measured after five-day forskolin or PTH treatment. Exposure to forskolin increased the ALP transcript level in MSCs by 17 fold over vehicle-treated controls (Fig. 5b). On the other hand, PPARγ was downregulated by more than forty percent (Fig. 5b). No significant changes in transcript levels were detected in other genes (data not shown). Five-day exposure to PTH had little impact on the expression of any of these early marker genes in MSCs (data not shown). Thus, forskolin was able to alter the gene expression of ALP and PPARγ prior to the start of differentiation process.
Canonical Wnt pathway as a mediator of PTH/cAMP action in MSCs
Precisely, how Gs-mediated cAMP signaling regulates MSC development remains controversial. In this paper, we explored the direct involvement of cAMP pathway in regulating MSC differentiation and demonstrated that stimulating cAMP signaling in MSCs enhanced osteogenic response, while continuous activation of cAMP signaling in differentiating osteoblasts negatively affected mineralization. Furthermore, stimulating cAMP signaling in MSCs favored osteogenic development over adipogenic development.
First, we used natural ligand, PTH, to activate the cAMP pathway in developing osteoblasts and MSCs. Using primary mouse BMSCs as the source of MSCs, we tested the effects of long-term exposure to PTH after initiation of osteogenesis. Long-term exposure to PTH during differentiation completely abolished the osteoblasts’ ability to mineralize, but not the number of ALP-positive colonies. On the other hand, treatment of BMSCs with PTH, followed by induction of osteogenesis, resulted in a modest increase in osteogenesis. As such, PTH produced varied effects on the mineralization potential of osteoblasts depending on the developmental stage of the cells.
PTH can also activate the MAPK cascade through Gi, as well as the phospholipase C pathway via Gq signaling. Therefore, forskolin was used to directly stimulate the production of cAMP to examine the specific role of cAMP signaling pathway in MSC differentiation. Stimulation of cAMP signaling by forskolin in differentiating osteoblasts for five days resulted in diminished osteogenesis, both in terms of total area of mineralization and number of mineralizing colonies. Thus, exposing differentiating osteoblasts to PTH or forskolin greatly reduced osteoblasts’ ability to mineralize.
Direct activation of cAMP signaling by forskolin in BMSCs yielded an osteogenic response that was opposite of activating cAMP signaling in differentiating osteoblasts. In addition to increased mineralization, there was an increase in the number of ALP-positive colonies. Dosage-dependent increase in the ratio of Von Kossa signal over ALP signal (data not shown) indicated that cAMP signaling in BMSCs conditioned the cells to more efficiently differentiate into osteoblasts, particularly their ability to mineralize. Elevated expression of osteoblast marker genes such as Runx2, Osterix, Collagen 1α1, Osteocalcin, and ALP at the late stages of differentiation also implied a more efficient osteoblast differentiation. As to why modest increase in gene expression of Collagen 1α1 and Osteocalcin could result in strong mineralization, one possibility is that early expression of ALP (before osteogenic induction) could act to increase local concentration of inorganic phosphate, a mineralization promoter, and to decrease the concentration of extracellular pyrophosphate, an inhibitor of mineral formation . Conversely, the effect of direct activation of cAMP signaling in BMSCs on adipogenesis was negative. Activation of cAMP signaling in mesenchymal cells, which were earlier in their development, might impart in them an ability to more efficiently differentiate into mature osteoblasts and less efficiently differentiate into mature adipocytes.
Activation of cAMP signaling by forskolin in BMSCs produced increased number of ALP-positive osteogenic colonies and decreased number of adipogenic colonies as measured by total Oil Red O staining. Expressions of early osteogenic and adipogenic marker genes were examined in MSCs which had been exposed to forskolin to determine changes in osteogenic or adipogenic potential of MSCs. ALP gene expression was greatly upregulated in MSCs that had been exposed to forskolin although the same treatment with PTH did not alter its expression. Conversely, PPARγ gene expression was downregulated in MSCs by forskolin treatment. Interestingly, the elevated ALP transcript level was not maintained once cells had been induced to differentiate into osteoblasts (Fig. 5, day 10) and recovered only in later stages of osteogenesis (Fig. 5, day 18). We do not know the biologic relevance of this transcriptional response. Perhaps, ALP transcriptional activation requires continuous presence of forskolin, thus it could be a direct target of cAMP signaling. Sustained elevated cAMP signaling in MSCs might have produced epigenetic changes that altered the transcriptional control of specific promoters. Recently, a PKA-dependent histone lysine demethylase complex had been characterized, suggesting that PKA signal could regulate histone methylation and gene transcription . This could be a potential mechanism that cAMP acts through to favor osteogenesis over adipogenesis in MSCs. Forskolin did increase PCNA transcript levels in MSCs (data not shown), suggesting changes in the makeup of MSC population. As such, activation of cAMP signaling in BMSCs could be both pro-osteogenic and mitogenic, resulting in more MSCs directed toward osteogenic lineage and less MSCs directed toward adipogenic lineage.
PTH can control Wnt signaling in osteoblast lineage cells via regulation of negative modulators such as Dkk1 and Sclerostin [41, 42]. There is also evidence that the effect of PTH on the canonical Wnt pathway occurs at least partially through the cAMP/PKA signaling by differentially regulating Wnt receptor complex proteins and Dkk1 . The role of canonical Wnt signaling in mesenchymal cell’s ability to differentiate into osteoblasts is unclear. Wnt stimulation had been reported to regulate osteogenesis in human MSCs both negatively and positively in a context-specific manner . Maturation state of the osteogenic cells also appeared to be important. Activation of canonical Wnt signal in undifferentiated MSCs and juvenile mouse calvarial osteoblasts inhibited osteogenesis, but strongly induced osteogenesis in mature calvarial osteoblasts [45, 46]. When we measured canonical Wnt target gene expression in MSCs exposed to either forskolin or PTH for 5 days before induction of osteoblast differentiation, only Axin2 gene expression was significantly upregulated by forskolin treatment. We also treated MSCs with Wnt3a for five days to make sure the gene expression of the selected Wnt target genes could be controlled by Wnt3a. As expected, Wnt3a was able to upregulate selected Wnt target genes. Contrary to enhanced osteogenic response produced by forskolin, osteogenic response followed by 5-day Wnt3a treatment was greatly reduced. Therefore, the effect of activation of cAMP signaling in MSCs on subsequent osteogenesis was not mediated through the canonical Wnt pathway.
Previously, several groups had investigated the role of cAMP in MSC linage specification. A high-throughput siRNA library screen conducted by Zhao and Ding  identified GNAS (Gαs) and ADCY8 (adenylate cyclase 8) as genes that suppressed osteogenic specification. The authors carried out a very similar experiment in which cells were subjected to specific siRNA knockdowns (initial 2–4 days) followed by osteogenic induction (additional 5–9 days) . Gene expression knockdown of these targets genes enhanced osteogenic maturation. Substituting siRNA with forskolin suppressed osteogenesis although a human MSC cell line (PT-2501) was used in their experiments . In contrast, de Boer and coworkers [28, 29] pretreated human MSCs derived from bone marrow with db-cAMP or forskolin in basic medium and was able to enhance in vivo bone formation after transplantation. Subsequently, they proceeded to test the same idea in the rodent models. Db-cAMP was applied to mouse MSCs in the presence of mineralization medium for 3 or 10 days. These treatments suppressed mineralization . As noted, db-cAMP was present in the mineralization medium in one case but not the other. It is important to consider the specific type of cell population in which the cAMP signal is activated, whether it is homogeneous MSCs, MSCs populated with other elements of the bone marrow, or MSCs that have been induced to undergo osteogenesis. In addition, dexamethasone was used in these experiments as one of the osteogenic inducers, while it was not used in our study. It is a synthetic member of the glucocorticoid class of steroid hormones and can exert its influence on lineage commitment through a cAMP/PKA-independent mechanism .
PTH and forskolin treatments on BMSCs sometimes produced different effects on subsequent MSC development. Stimulating cAMP signal in BMSCs by forskolin reduced the efficiency of adipogenesis, whereas PTH enhanced adipogenesis to a small degree. Forskolin activation of cAMP signaling in BMSCs favored osteogenesis over adipogenesis. Exposure to PTH promoted both. This is in contrast to the findings by Rickard et al.  where intermittent PTH treatment or a non-peptide agonist of PTH1 receptor suppressed adipocyte differentiation from human MSCs. However, human MSCs used in these experiments were relatively homogeneous. In contrast, BMSCs derived from mouse long bones are heterogeneous, and it is possible that cells other than MSCs contribute to the ability of PTH to promote adipogenesis. Furthermore, Rickard’s human MSCs were cultured and treated with PTH under mildly adipogenic conditions whereas we activated PTH and cAMP signaling before adipogenic induction. Finally, in our case, PTH and forskolin treatments were continuous for five days. In Richard’s experiments, intermittent PTH and forskolin treatments were one hour per day. Further studies are needed to define the precise basis for the differences between the present results and those of Rickard’s et al. .
PTH stimulates a diverse range of signaling pathways. It would be interesting to pharmacologically inhibit either cAMP/PKA or protein kinase C/MAPK pathway to discover the contribution of each signaling pathway to the effects of PTH on osteogenesis and adipogenesis. However, the different results produced by PTH and forskolin can also be attributed to the cell types which express PTH1R to receive the signal. Forskolin treatment of BMSCs activated cAMP pathway in all cell types in that population. PTH treatment activated cAMP pathway only in cells that express PTH1R. Therefore, without knowing the cell populations that express PTH receptors, it is difficult to interpret results of pharmacological manipulation of the downstream pathways of PTH signaling. BMSCs consist of a wide range of cell types, including cells of hematopoietic lineage. Studies using genetically modified mice provided evidence that PTH-mediated anabolic response required participation of cells of hematopoietic origin [48, 49, 50]. Osteopetrotic c-fos-deficient mice that have osteoclast defect were not responsive to PTH treatments . Perhaps it is more important to first identify the cell type(s) in which activated cAMP signaling can potentiate osteogenesis and to identify the GPCR(s) that activates cAMP signaling and can induce a more robust osteogenesis. Recent reports suggested that Gs- and Gq-coupled A2A and A2B adenosine receptors played a role in promoting proliferation and differentiation of bone marrow-derived mesenchymal stem cell [51, 52]. In fact, A2B adenosine receptor knockout mice exhibited lower bone density in the femur . Moreover, transfection of A2B receptors into murine osteoblast precursor cell line, 7F2, stimulated osteocalcin and ALP expression and inhibited subsequent adipogenesis .
In conclusion, activation of cAMP pathway in BMSCs conditioned MSCs to produce a more efficient osteogenic response, but a less efficient adipogenic response. On the other hand, sustained activation of cAMP pathway in differentiating osteoblasts inhibited mineralization. Cyclic AMP signaling in BMSCs may have a role in controlling the balance between adipogenesis and osteogenesis, partially by regulating lineage commitment of MSCs.
This work was supported by NIH Grant DK072071 and VA Merit Review Grant, which were administered by the Northern California Institute for Research and Education, and with resources of the Veterans Affairs Medical Center, San Francisco, CA. Robert A. Nissenson is a Senior Career Scientist of the Department of Veterans Affairs. We thank Susan Millard, Lalita Wattananchaya, and Liping Wang for their scientific input and assistance in various aspects of these studies.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.