Bone marrow stromal cells (BMSCs) are a multipotent adult stem cell type that can give rise to a variety of lineages, including osteoblasts, adipocytes, chondrocytes, and myocytes.1 Experimental evidence indicates that many of these differentiation programs are regulated in a reciprocal manner2, 3 and that the fate of BMSCs is a determinant of bone mass. For example, under normal conditions, the balance between adipogenic and osteoblastogenic differentiation of BMSCs favors osteoblast differentiation and thereby bone formation. However, in diseases such as osteoporosis, this homeostatic balance becomes disrupted, leading to a shift toward adipogenesis at the expense of osteoblastogenesis.4–6 Ultimately, this reduction in osteoblast number and consequent capacity for bone formation is believed to contribute to several diseases associated with chronic bone loss.2, 7, 8 Several transcription factors are now recognized as key determinants of osteoblastogenic or adipogenic differentiation of BMSCs, and changes in the expression and/or activity of these factors have been associated with bone formation or bone loss, respectively, in animal models. For example, increased expression of peroxisome proliferator-activated receptor γ (PPARγ), a master adipogenic transcription factor, is associated with bone loss during aging and osteoporosis.9–11 Conversely, expression of runt-related gene 2 (RUNX2), a transcription factor that promotes osteoblast differentiation, is generally associated with bone formation.12, 13 While there is much evidence in support of a reciprocal relationship between these transcription factors and BMSC osteoblastogenesis/adipogenesis, the mechanisms that ultimately govern the balance of the expression and activity of these factors remains poorly understood.
An important determinant of BMSC fate is believed to be extracellular signaling molecules that cause one lineage to be favored over another. For example, the hormones vitamin D and estrogen both inhibit adipogenesis and promote osteoblastogenesis in vitro and in vivo.14–16 In support of this idea, with aging and in patients with osteoporosis, the bone marrow milieu is altered such that BMSC differentiation favors the adipocyte over the osteoblast lineage.17 Therefore, identification of other factors that influence BMSC fate may provide us with potential therapeutic targets for conditions associated with low bone formation and high marrow adiposity. Experimental evidence indicates that adipocyte-derived signaling molecules (adipokines) may modulate osteoblast and/or adipocyte differentiation and metabolic function within bone.18 For example, leptin and adiponectin can induce osteoblastogenesis and bone growth while inhibiting bone marrow adipogenesis.19–23 Chemerin, a more recently identified adipokine, is highly expressed in adipose tissue in vivo and has been shown to promote adipogenesis in a 3T3-L1 fibroblast preadipocyte model.24, 25 Knockdown of chemerin expression or that of the cognate receptor chemokine-like-receptor-1 (CMKLR1) in preadipocytes by RNA interference almost completely abrogates adipogenesis along with a concomitant loss of adipogenic gene expression in response to hormonal inducers.25 Chemerin also serves as a ligand for chemokine (CC motif) receptor-like 2 (CCRL2),26 but the role of this receptor in adipocyte differentiation has not been studied.
Given the role of chemerin/CMKLR1 signaling in 3T3-L1 adipogenesis, we hypothesized that this signaling pathway also promotes adipocyte differentiation of osteoblast progenitor cells while negatively regulating the osteoblastogenic differentiation of these cells. To investigate this hypothesis, we examined the requirement of chemerin/CMKLR1 for adipocyte differentiation of a 7F2 preosteoblast cell line as well as primary BMSCs isolated from murine bone marrow.
All experiments using laboratory animals were performed according to the guidelines of Canadian Council on Animal Care and were approved by the Dalhousie University Committee on Laboratory Animals. C57/BL/6J mice were bred in-house in the Carleton Campus Animal Care Facility and were maintained at an ambient temperature of 20 to 22°C, relative humidity 18% to 22%, and a 12 hour of light/dark cycle. Free access to standard rodent chow and water was allowed for the duration of the study.
Murine preosteoblast 7F2 cells (ATCC, Rockville, MD, USA) were grown in Minimum Essential Medium alpha modification (α-MEM) supplemented with 1% L-glutamine and 10% fetal bovine serum (FBS) in T-75 tissue culture flasks. Cells were cultured in a humidified 5% CO2 incubator at a temperature of 37°C. When the cells reached approximately 75% confluence, they were trypsinized and seeded at a density of 1.5 × 104 cells/cm2 in 12-well plates for differentiation and gene knockdown experiments.
BMSCs were isolated from the bone marrow of 8-week-old C57/BL6 mice (20 to 25 g) of either sex based on the method of Meirelles and Nardi.27 On the day of isolation, animals were euthanized with an overdose of sodium pentobarbital (90 mg/kg i.p.). Both tibiae were collected and their marrow was flushed with α-MEM containing 1% L-glutamine and 10% FBS. The medium containing the crude bone marrow was passed through a 22-gauge needle to separate the cells, centrifuged at 2500 rpm, and resuspended in fresh medium. The cells were plated into a T-75 flask at a density of 1.5 × 106 cells/cm2 and cultured in a humidified 5% CO2 incubator at a temperature of 37°C. After 72 hours, the medium containing nonadherent hematopoietic cells was removed, and fresh medium was added. After a further 2 weeks of culture, rapidly proliferating adherent spindle-shaped fibroblastic cells formed the majority of the cell population. At this point, the cells were trypsinized using 0.05% trypsin-EDTA acid (Invitrogen, Burlington, Ontario, Canada) and subsequently passaged with a split ratio of 1:2. After a subsequent two passages, the cells were exclusively fibroblast-like and grew rapidly, enabling a 1:4 split ratio with every 5 days. The phenotype of these cells was confirmed by the presence of BMSC surface marker proteins (CD44 and CD49e) and the absence of the hematopoietic lineage marker CD45 by flow cytometric immunophenotyping (data not shown).
Murine colon adenocarcinoma MCA38 cells (a generous gift of Jonathan Blay, Dalhousie University) were maintained in Dulbecco's modified Eagle's Medium (DMEM) containing 10% FBS at 37°C in a humidified 5% CO2 incubator. When the cells reached approximately 75% confluence, they were trypsinized and seeded at a density of 1.5 × 104 cells/cm2 in 12-well plates for chemerin expression analysis experiments.
Adenovirus expressing shRNAs were generated as described previously.25 The following target sequences were used:
Chemerin (NM_027852), GGAGTTGCAATGCATTAAGAT
CMKLR1 (NM_008153), GGAAGATAACCTGCTTCAACA
CCRL2 (NM_017466), GGGAACATCTACTTCCTAAAC
In all cases, a loop sequence of CGAA was used. For expression of PPARγ, a cDNA encoding the complete murine coding sequence (NM_011146) was initially cloned into the pENTR4 entry vector (Invitrogen) and subsequently recombined with the pAD/CMV/V5-DEST vector using the ViraPower Adenoviral Gateway Expression Kit as per the manufacturer's (Invitrogen) instructions. Crude viral lysates were purified using the ViraBind Adenovirus Purification Kit (Cell Biolabs, Inc., San Diego, CA, USA), and titers were determined using the QuickTiter Adenovirus Titer ELISA Kit (Cell Biolabs Inc.). Primary mouse BMSCs or 7F2 cells were plated a minimum period of 12 hours before initiating transduction. For knockdown studies, viral transduction was carried out at an optimal dose of 20 (for BMSC) and 50 (for 7F2) multiplicity of infection (MOI; viruses/cell) in 0.5 mL transduction medium (serum-free α-MEM containing polybrene at 6 µg/mL). Transduction with the PPARγ adenovirus was performed similarly, but with an MOI of 20 for both BMSCs and MCA38 cells. After 24 hours, the transduction medium was replaced with complete medium, and the cells were allowed to grow for an additional 24 hours prior to initiating differentiation experiments or gene expression analysis.
Induction of adipogenic differentiation
Adipocyte differentiation of BMSCs was achieved based on established methods.28 The cells were treated for 2 days with induction medium followed by 2 days with maintenance medium for a total of three cycles, beyond which they were kept in maintenance medium. The induction medium was α-MEM supplemented with 1 µM rosiglitazone, 5 µg/mL insulin, 0.1 nM dexamethasone, 50 µg/mL ascorbic acid, 60 µM indomethacin, 7.5% FBS, and 2.5% lot-selected rabbit serum. The maintenance medium was α-MEM containing 1 µM rosiglitazone, 5 µg/mL insulin, 7.5% FBS, and 2.5% rabbit serum. Adipocyte differentiation of 7F2 cells was initiated using an adipogenic medium of α-MEM supplemented with 5 µg/mL insulin, 0.1 nM dexamethasone, 50 µg/mL ascorbic acid, 60 µM indomethacin, and 10% FBS. After 3 days, the cells were changed to a medium that was identical to the adipogenic medium but lacking indomethacin. Lipid accumulation was stained and quantified by oil red O staining, as described previously.25 Macroscopic images of the plates were obtained using a desktop computer scanner, and microscopic images were taken using an inverted light microscope.
Assessment of cell proliferation
Proliferation of BMSCs and 7F2 cells in response to serum or the adipogenic differentiation cocktail was evaluated using a modification of established methods.29, 30 The cells were plated at a density of 1.5 × 104 cells/cm2 in a 12-well plate in a total volume of 1 mL/well. Twelve hours after plating, the cells were transduced with the indicated adenoviruses for 24 hours in serum-free medium. Following this, the cells were rinsed once with medium containing 10% FBS, and complete growth medium containing 10% FBS was added to assess the basal proliferation rates. In order to assess the postconfluent proliferation before commitment to adipocytes, the adipogenic differentiation cocktail was added after reaching confluence (required an additional 36 hours in growth medium). To measure cell proliferation, a thymidine uptake assay was performed at 12 hour intervals for a total of 3 days. Radioactive [methyl-3H]thymidine (Amersham Biosciences Inc., Baie d'Urfe, Quebec, Canada) at a final concentration of 1 µCi/mL was added to the medium, and after 15 minutes, the plates were placed on ice and the medium was aspirated. The wells were washed twice with ice-cold PBS, and 0.5 mL 10% trichloroacetic acid (TCA) was added to precipitate the DNA for 1 hour at 4°C. After aspiration of the TCA and rinsing with absolute ethyl alcohol, the plates were allowed to air dry. The final precipitate of the cellular macromolecules was dissolved in 500 µL 0.1N NaOH/1% sodium dodecyl sulfate (SDS), and the radioactivity of 100 µL of this solution was assessed using a Beckman LS 6500 Multi-Purpose Scintillation Counter (Brea, CA, USA).
Induction of osteoblastogenic differentiation
Osteoblastogenic,1 combined adipogenic and osteoblastogenic,31 or interconversion of adipogenic to osteoblastogenic differentiation32 of BMSCs was performed based on established procedures. For osteoblastogenesis, the cells were treated with a medium consisting of α-MEM supplemented 10% FBS, 50 µg/m ascorbic acid, and 10 mM β-glycerophosphate for 14 days. For the combined induction, adipogenic medium (induction or maintenance, as described earlier) was mixed in a ratio of 1:1 with the osteoblastogenic medium described earlier. For interconversion experiments, the BMSCs were incubated with adipogenic medium for an initial period of 3 days and then were switched to the osteoblastogenic medium for a subsequent 11 days. Mineralization was assessed by alizarin red S staining, and quantification was according to established protocols.33
SDS-PAGE and Western blotting of chemerin protein in tissue culture medium were performed as described previously25 with the modification that immunodetection of chemerin protein was achieved using a goat antichemerin (1:1000) antibody (R&D Systems Inc., Minneapolis, MN, USA).
Gene expression analysis
Primers for quantitative real-time PCR (qPCR) were designed using the primerbank software algorithm (http://pga.mgh.harvard.edu/primerbank/), and whenever possible, primers were designed to be exon-spanning. The primer sequences used and amplicon sizes were: mChemerin (NM_027852), forward: TACAGGTGGCTCTGGAGGAGTTC, reverse: CTTCTCCCGTTTGGTTTGATTG, 195 bp; mCMKLR1 (NM_008153), forward: CGAGTTCTCAAACCCTGAAGTCGC, reverse: CAAGTCCACAAAGTAGCCAAAGCC, 222 bp; mCCRL2 (NM_017466), forward: CTCTGCTTGTCCTCGTGCTT, reverse: GCCCACTGTTGTCCAGGTAG, 209 bp; mAdiponectin (NM_009605), forward: AGCCGCTTATATGTATCGCTCA, reverse: TGCCGTCATAATGATTCTGTTGG, 118 bp; mPPARγ (NM_011146), forward TCGCTGATGCACTGCCTATG; reverse: GAGAGGTCCACAGAGCTGATT, 102 bp; mAKP2 (NM_007431), forward: AGGGCAATGAGGTCACATCC, reverse: GCATCTCGTTATCCGAGTACCAG, 150 bp; mCOL1A2 (NM_007743), forward: CTGGAACAAATGGGCTCACTG, reverse: CAGGCTCACCAACAAGTCCTC, 147 bp; mOsterix (NM_130458), forward: ATGGCTCGTGGTACAAGGC, reverse: GCAAAGTCAGATGGGTAAGTAGG, 205 bp; mCD36 (NM_007643), forward: GAACCACTGCTTTCAAAAACTGG, reverse: GTCCTGAGTTATATTTTCCTTGG, 178 bp; and mCyclophilinA (X52803), forward: GAGCTGTTTGCAGACAAAGTTC, reverse: CCCTGGCACATGAATCCTGG, 124 bp.
Total RNA was isolated using an RNeasy Plus Minikit (Qiagen) according to the manufacturer's instructions, and 0.5 µg of the RNA was reverse-transcribed using Affinityscript reverse transcriptase (Stratagene, Cedar Creek, TX, USA) according to the manufacturer's instructions. Then 1 µL of each reaction was used as a template for qPCR amplification using Brilliant SYBR Green QPCR Master Mix (Stratagene) in a total volume of 20 µL. An MX3000p thermocycler (Stratagene) was used for amplification according to the following cycling conditions: initial denaturation at 95°C for 10 minutes, followed by 40 cycles of 20 seconds at 95°C, 18 seconds at 60°C, and 30 seconds at 72°C. PCR product size and the presence of a single amplicon were verified by electrophoresis on 2.5% agarose gels. The Ct values for cyclophilin A were used to normalize the expression level of the gene of interest using the ΔΔCt method.34
Results are expressed as mean ± SEM. All data were analyzed using the Student's t test (two-tailed), and p < .05 was considered statistically significant.
Chemerin expression during 7F2 adipocyte differentiation
The murine bone marrow–derived preosteoblast cell line 7F2 was chosen as an initial experimental model for these studies. This cell line provided a homogeneous population of cells that exhibit some phenotypic markers characteristic of osteoblasts. However, unlike other terminally differentiated osteoblasts (e.g., 5H6), 7F2 cells retain the ability to undergo adipocyte differentiation in response to appropriate chemical inducers.35 To verify adipogenic differentiation of 7F2 cells, qPCR analysis was used to measure mRNA levels for the adipocyte marker genes PPARγ and Adiponectin. As shown in Fig. 1(A, B), PPARγ and adiponectin mRNA levels increased (approximately 500- and 600-fold, respectively) with increasing time of exposure to the adipogenic media. Adipogenesis also was verified by an increase in intracellular lipid accumulation, as measured by oil red O staining (data not shown). Similar to the adipocyte markers, chemerin mRNA expression increased significantly with adipocyte differentiation of 7F2 cells. Specifically, by day 10, chemerin mRNA levels were increased 600-fold compared with undifferentiated cells (see Fig. 1C). This coincided in both magnitude and temporal pattern with the adipocyte marker genes and is consistent with previous findings25 in 3T3-L1 preadipocytes. Beginning at day 3 of adipogenic differentiation, a 16-kDa protein corresponding to the active form of chemerin was detected by Western blotting of 7F2-conditioned adipocyte media (see Fig. 1D). In contrast, cells maintained for identical periods in normal growth medium did not exhibit detectable chemerin secretion at any time point (see Fig. 1D). In accordance with the mRNA levels, chemerin secretion increased with adipogenic differentiation. In contrast to chemerin, mRNA levels for the chemerin receptor CMKLR1 were significantly decreased initially with adipogenesis but recovered to a level comparable with that of undifferentiated cells by day 12 (see Fig. 1E). There appeared to be an increasing trend of mRNA levels for CCRL2, but the changes were not statistically significant (see Fig. 1F). Interestingly, by day 3 of adipogenic differentiation, a significant decrease in mRNA levels for two established osteoblast markers, collagen type Iα2 (COL1A2) and alkaline phosphatase (ALP), was observed (see Fig. 1G, H). While COL1A2 mRNA levels remained lower than in undifferentiated cells throughout adipocyte differentiation, mRNA levels for ALP were restored with advancing day of differentiation (from day 5 onward) back to the levels observed prior to adipogenic induction.
Knockdown of chemerin, CMKLR1 and CCRL2 in 7F2 cells
To test the functional significance of chemerin, CMKLR1, and CCRL2 expression to 7F2 adipogenesis, adenoviral vectors expressing shRNA targeting Chemerin, CMKLR1, CCRL2, or LacZ (a control for nonspecific effects of viral transduction and shRNA expression) were developed. Using this approach to reduce chemerin, CMKLR1, and CCRL2 mRNA levels, 7F2 cells were transduced for 24 hours prior to addition of adipogenic medium to the cells. At days 3 and 14 of differentiation, oil red O staining of neutral lipids and mRNA expression of two adipocyte marker genes (PPARγ and Adiponectin) were significantly reduced with chemerin or CMKLR1 shRNA compared with LacZ- and vehicle-treated controls (Fig. 2A–C). CCRL2 knockdown resulted in an approximately twofold increase of PPARγ mRNA levels, but only at day 14 of differentiation (see Fig. 2B). Consistent with efficient knockdown, more than 90% reduction of chemerin, CMKLR1, or CCRL2 mRNA levels was observed in cells transduced with adenovirus expressing chemerin, CMKLR1, or CCRL2 shRNA, respectively, at days 3 and 14 of differentiation (see Fig. 2D–F). Both LacZ- and vehicle-treated controls exhibited the expected increase in chemerin expression (see Fig. 2D). CMKLR1 shRNA caused a decrease in chemerin mRNA expression, presumably as a consequence of the abrogation of adipocyte differentiation. Interestingly, transduction with adenovirus expressing chemerin shRNA significantly increased the mRNA levels of CMKLR1 and CCRL2 during adipocyte differentiation (see Fig. 2E, F), suggesting a possible compensatory response to the forced reduction of chemerin. Conversely, CCRL2 knockdown resulted in an approximate twofold induction of chemerin mRNA levels, but only at day 14 of differentiation (see Fig. 2D). Intriguingly, the mRNA levels of osteoblast marker genes ALP and COL1A2 were increased approximately fivefold by chemerin or CMKLR1 knockdown versus vehicle or LacZ control groups during adipocyte differentiation (see Fig. 2G, H). Consistent with these findings, mRNA levels for the osteoblastogenic transcription factor osterix (OSX) also was increased five- to tenfold by transduction with chemerin or CMKLR1 shRNA adenovirus in comparison with the respective vehicle controls at both time points (days 3 and 14) of examination (see Fig. 2I).
Chemerin, CMKLR1, and CCRL2 expression during BMSC adipocyte differentiation
To test whether the effects of chemerin and CMKLR1 knockdown in the 7F2 preosteoblast cell line translated to a more immature bone precursor, we next examined BMSCs isolated from murine bone marrow. After 14 days of exposure to the adipogenic medium, the BMSCs exhibited a robust adipocyte differentiation, as evidenced by the presence of large lipid vesicles (Fig. 3A), strong oil red O staining (data not shown), and increases in the mRNA level of the adipocyte marker genes PPARγ (200-fold) and adiponectin (>100,000-fold) (see Fig. 3B, C). Similar to 7F2 cells, adipocyte differentiation was associated with a marked increase (approximately 10,000-fold) in chemerin mRNA levels (see Fig. 3D). Consistent with this finding, a 16-kD protein corresponding to active chemerin was identified in the medium of differentiating cells as early as day 3 of differentiation and persisted for up to 28 days (see Fig. 3G). In contrast, cells maintained for identical periods in normal growth medium did not exhibit detectable chemerin secretion at any time point (see Fig. 3G). Similar to observations in 7F2 cells, BMSC adipogenic differentiation was associated with a progressive decrease in CMKLR1 mRNA levels (see Fig. 3E). While there was an initial small (20%) but significant decrease in CCRL2 mRNA levels on day 3 of differentiation, in general, CCRL2 expression was largely unchanged with progressive adipocyte differentiation of BMSCs (see Fig. 3F). As with adipocyte differentiation of 7F2 cells, a significant decrease in mRNA levels for the osteoblast marker genes ALP and COL1A2 was observed throughout BMSC adipogenesis (see Fig. 3H, I).
Knockdown of chemerin, CMKLR1, and CCRL2 in BMSCs
To explore the functional significance of chemerin signaling to BMSC adipogenesis, we transduced cells with adenovirus expressing chemerin, CMKLR1, CCRL2 or LacZ shRNAs prior to inducing adipogenic differentiation. Consistent with 7F2 cells, knockdown of chemerin or CMKLR1 expression reduced lipid accumulation by approximately 75% compared with vehicle treatment in BMSCs induced to undergo adipocyte differentiation (Fig. 4A, B). Knockdown of CCRL2 expression also inhibited lipid accumulation, albeit to a lesser extent (25% loss) and only at day 14. In some models, adipogenic stimuli induce one to two rounds of postconfluent clonal expansion that is critical for adipogenic differentiation.36, 37 To determine if chemerin/CMKLR1 signaling was involved in this process, we examined BMSC proliferation immediately after exposure to adipogenic medium. As shown in Fig. 4C, vehicle-treated cells exhibited a clear increase in cell proliferation 12 to 72 hours after exposure to adipogenic medium. In contrast, knockdown of chemerin, CMKLR1, or CCRL2 expression almost completely abrogated this postconfluent clonal expansion. These findings were consistent with experiments using 7F2 cells, which also exhibited a significant decrease in postconfluent clonal proliferation with chemerin or CMKLR1 knockdown (data not shown). Since proliferation is an important aspect of the maintenance and self-renewal of BMSCs, preconfluent proliferation also was examined in the presence of normal growth medium. Similar to the clonal expansion in response to adipogenic stimuli, preconfluent proliferation of BMSCs was significantly reduced by knockdown of chemerin, CMKLR1, or CCRL2 (see Fig. 4D).
Consistent with inhibition of adipogenesis, when compared with vehicle-treated cells, chemerin or CMKLR1 shRNA expression decreased (>90% loss) the mRNA levels of the adipocyte marker genes PPARγ and adiponectin at days 3 and 14 of differentiation (Fig. 5A, B). CCRL2 knockdown also resulted in a greater than 50% loss of PPARγ mRNA levels, but only at day 14 of differentiation (see Fig. 5A). Similar to 7F2 cells, greater than 90% reduction of mRNA levels for the respective target genes for each adenoviral-expressed shRNA was achieved (see Fig. 5C–E). Also consistent with previous results, knockdown of each target was associated with effects on other components of the signaling pathway. For example, chemerin knockdown was associated with fivefold increased levels of CMKLR1 mRNA (see Fig. 5D), and CMKLR1 knockdown was associated with greater than 90% loss of chemerin mRNA levels (see Fig. 5C) compared with vehicle-treated cells.
Osteoblastogenic gene expression and mineralization of BMSCs
Consistent with the effects seen in 7F2 cells, knockdown of chemerin or CMKLR1 expression prior to adipogenic differentiation of BMSCs resulted in a 2.5- to 5-fold increase of the mRNA levels of the osteoblast markers genes ALP, COL1A2, and OSX compared with vehicle-treated controls (see Fig. 5F–H). To determine if these changes in marker gene expression associated with chemerin or CMKLR1 knockdown were associated with phenotypic activity characteristic of osteoblasts, three different differentiation protocols were used: osteoblastogenic medium for 14 days, mixed osteoblastogenic-adipogenic medium for 14 days, or an interconversion protocol where BMSCs were first treated with adipogenic medium for 3 days, followed by replacement with osteoblastogenic medium for 11 days. In all protocols, mineralization was measured using alizarin red S staining. A consistent finding with all differentiation protocols was an increase in the total extent of mineralization when chemerin or CMKLR1 knockdown was performed prior to differentiation (Fig. 6). When the combined osteoblastogenic-adipogenic protocol was used, compared with the complete osteoblastogenic protocol, there was a twofold decrease in total extent of mineralization accompanied by the presence of cells containing large lipid vesicles (see Fig. 6A–D). Chemerin or CMKLR1 knockdown significantly increased the mineralization (>25% increase) compared with the vehicle-treated controls (see Fig. 6C, D). Furthermore, with chemerin or CMKLR1 knockdown, cells containing large lipid vesicles were essentially absent and were replaced by mineralized cells (see Fig. 6C). Similar results were observed in the adipogenic-to-osteoblastogenic interconversion protocol, but the greatest improvement in total mineralization (two- to threefold increase) in response to chemerin or CMKLR1 knockdown was seen when the cells were driven to interconvert to osteoblasts from adipocytes (see Fig. 6E, F).
Regulation of chemerin by PPARγ and its rescue effects on chemerin or CMKLR1 knockdown during adipogenic differentiation of BMSCs
Given the critical importance of PPARγ as a regulator of gene expression during adipogenic differentiation, the impact of this transcription factor on chemerin expression was tested. Chemerin expression was increased significantly by treatment with the PPARγ ligand rosiglitazone (threefold), forced expression of PPARγ by adenoviral gene delivery (threefold), or a combination of the two (>50-fold) in BMSCs for 4 days (Fig. 7A). Since PPARγ activation and expression in BMSCs causes these cells to undergo adipogenic differentiation (data not shown) and chemerin expression is normally increased with adipogenic differentiation, the same experiment was performed with a nonadipogenic cell type to delineate the generalized adipogenic effects of PPARγ from a selective regulatory role of PPARγ on chemerin expression. As shown in Fig. 7B, rosiglitazone treatment, forced expression of PPARγ by adenoviral gene delivery, or a combination of the two for 4 days, significantly increased chemerin mRNA levels (>10-fold, >20-fold, and 100-fold, respectively) in murine colon adenocarcinoma MCA38 cells. For comparison, expression of the established PPARγ target gene, CD36, in response to rosiglitazone treatment and/or forced expression of PPARγ was qualitatively similar to that of chemerin in both cell types (see Fig. 7C, D). Functionally, forced expression of PPARγ partially rescued the loss of adipocyte marker gene expression (50% to 100% rescue of adiponectin with 10- to 50-fold expression of PPARγ) caused by chemerin or CMKLR1 knockdown in BMSCs (see Fig. 7E, F). Interestingly, forced expression of PPARγ also was capable to a limited extent of restoring the adipogenic induction of chemerin (25% to 50% rescue) even in the presence of chemerin knockdown (see Fig. 7G). In contrast, forced expression of PPARγ resulted in a significant reversal of the increased levels of mRNA for the osteoblastogenic gene osterix in response to chemerin and/or CMKLR1 knockdown coincident with adipogenic differentiation of BMSCs (see Fig. 7H).
For this study we proposed that the novel adipokine chemerin and the cognate receptor CMKLR1 are positive regulators of adipocyte differentiation of bone marrow–derived osteoblast precursor cells. Following from this, we proposed that interference with chemerin/CMKLR1 signaling not only would block adipogenesis in these cells but also would promote osteoblastogenesis. We also tested the role of CCRL2, a nonsignaling receptor for chemerin,26 in these differentiation processes. Consistent with other murine and human cell models of adipogenesis.24, 25, 38, 39chemerin mRNA levels and secretion of chemerin protein were increased dramatically with adipocyte differentiation of both 7F2 cells and primary bone marrow–derived BMSCs. Of note, when compared with 7F2 cells, the fold induction for chemerin was markedly higher in BMSCs as consequence of the much lower basal expression on day 0 in the latter cell type (data not shown). This observation is consistent with the ability of 7F2 cells to undergo adipocyte differentiation more readily than BMSCs and without the requirement of additional components in the differentiation medium such as rabbit serum or rosiglitazone. In contrast to previous studies, adipocyte differentiation of 7F2 cells or BMSCs was associated with a general decline in CMKLR1 mRNA levels. At present, it is unknown whether the decline in CMKLR1 expression equates with a loss of signaling function. However, it is worth noting that chemerin/CMKLR1 signaling appears to be most important early in adipogenesis because knockdown of chemerin or CMKLR1 in 3T3-L1-derived adipocytes after 4 days of differentiation did not affect subsequent adipocyte maturation, nor did it result in a significant loss of adipocyte phenotype.25 CCRL2 was recently reported26 to function as a mast-cell-expressed “silent” receptor that does not undergo internalization or function as a signal-transduction protein in response to binding with chemerin. Instead, CCRL2 is believed to contribute to inflammation by binding and increasing the concentration of chemerin at sites of inflammation. There have been no previous reports regarding the expression of this receptor during adipogenesis. In this study, expression of CCRL2 remained relatively constant throughout adipocyte differentiation of 7F2 cell or BMSCs.
This study demonstrates for the first time that chemerin and CMKLR1 are required for adipocyte differentiation of bone marrow–derived cells such as 7F2 cells and BMSCs. This is consistent with shRNA knockdown experiments in 3T3-L1 adipocytes and argues for the functional conservation of chemerin/CMKLR1 signaling as a generalized requirement for adipogenesis. While various studies have linked CMKLR1 activation to increased cell calcium levels40–42 and extracellular signal-regulated kinase (ERK) phosphorylation25, 38, 41 to decreased cell cAMP levels41 and modulation of inflammatory gene expression,43 very little is known regarding the intracellular signaling pathways coupled to CMKLR1. Thus, at present, the mechanism underlying the adipogenic effect of chemerin/CMKLR1 signaling remains undefined. However, given the critical importance of PPARγ for adipocyte differentiation, it is likely that the lack of early induction of this transcription factor in response to adipogenic stimuli with chemerin or CMKLR1 knockdown is a major factor. This is consistent with previous findings demonstrating that early expression of PPARγ is required to induce the adipogenic differentiation program44, 45 and is supported by our observation that while CCRL2 knockdown caused a significant decrease in PPARγ expression in BMSCs at day 14 of adipocyte differentiation, there was no significant effect on PPARγ expression at day 3 and no overt effect on adipocyte differentiation. However, while not affecting commitment to the adipogenic lineage, this decrease in PPARγ expression later in the adipogenic program may have affected terminal differentiation, as suggested by reduced lipid accumulation and adipogenic gene expression on day 14 compared with controls. Indeed, experimental evidence supports that PPARγ is required to maintain the mature adipocyte phenotype for the life span of the cell.46 However, this result diverges from the findings with 7F2 cells, where CCRL2 knockdown resulted in increased expression of chemerin and PPARγ, albeit with mild effects and observed only at day 14. One remarkably consistent finding between these cell types is that the expression of chemerin and PPARγ exhibits a close temporal association during adipocyte differentiation. This association also was reflected with CCRL2 knockdown in either cell type. For example, while CCRL2 knockdown in BMSCs decreased PPARγ mRNA levels on day 14 of adipocyte differentiation, decreased chemerin mRNA levels also were observed. In a related but opposite fashion, CCRL2 knockdown in 7F2 cells increased PPARγ and chemerin mRNA levels on day 14 of adipocyte differentiation.
In contrast to the effects on lipid accumulation and adipocyte target gene expression, chemerin, CMKLR1, and CCRL2 knockdown were similarly effective at inhibiting the early postconfluent clonal expansion of BMSCs in response to adipogenic stimuli. One explanation for these results is that CCRL2 may serve early in adipogenesis to concentrate the smaller amounts of chemerin secreted by the precursor cells into the tissue culture mediim. As adipocyte differentiation proceeds and increasing amounts of chemerin are secreted into the medium, this function becomes dispensable. Although much of the evidence concerning the requirement of clonal expansion for adipogenic differentiation is controversial and cell-type-dependent,47 our data are in accordance with the prevailing view that although clonal expansion does occur in murine and human BMSCs, it is not required for full adipocyte differentiation. A chemerin-concentrating function for CCRL2 is also consistent with the negative effect of CCRL2 knockdown on the preconfluent proliferation of BMSCs in the presence of normal growth medium. Under these conditions, where chemerin secretion is low, it would be expected that any concentrating function of CCRL2 would be most important.
In addition to a critical role for chemerin/CMKLR1 signaling in adipocyte differentiation of bone marrow–derived 7F2 cells and BMSCs, the data from this study indicate that this signaling pathway is a negative regulator of osteoblastogenesis. This is supported by the observation that chemerin or CMKLR1 knockdown in 7F2 cells was associated with the maintenance and/or upregulation of expression of genes associated with the osteoblast phenotype even in the presence of adipogenic stimuli. Similarly, CMKLR1 knockdown and, to some extent, chemerin knockdown preserved osteoblast gene expression in BMSCs subjected to adipogenic stimuli. Consistent with the minimal effects of CCRL2 knockdown on adipogenesis, reduction of the expression of this receptor prior to adipocyte differentiation was not associated with any significant effects on osteoblast gene expression in BMSCs. In the context of an osteoblastogenic medium, the increase in alizarin red S staining with chemerin or CMKLR1 knockdown reflected improved mineralization and was consistent with a negative effect of chemerin/CMKLR1 signaling on BMSC osteoblastogenesis. While not truly physiologic, the mixed adipogenic-osteoblastogenic medium is more likely representative of the conditions to which BMSCs are exposed in vivo. This condition also may better reflect the loss of osteoblastogenesis concurrent with stimulation of adipogenesis that may contribute to osteopenia induced by corticosteroids or aging. Under this condition, the concurrent loss of adipocyte formation and increase in osteoblast mineralization with knockdown indicated that decreasing the expression of chemerin or CMKLR1 in BMSCs shifts the balance between these differentiation programs to favor osteoblastogenesis.8, 17, 48 However, it is important to note that any reduction in mineralization observed with the combination of adipogenic-osteoblastogenic cocktails could be due to the combined effects of cells being driven to adipocyte lineage and as well as a direct inhibitory effect of components of the adipogenic cocktail such as dexamethasone. Thus there is the potential for an unforeseen interaction between the components of the adipogenic medium and the effects of the gene knockdown. Although such an interaction cannot be distinguished with the present experimental design, comparison with the appropriate controls (i.e., vehicle and LacZ shRNA) provides convincing evidence of the important role for chemerin/CMKLR1 signaling as a determinant of differentiation in the context of a mixed adipocyte-osteoblast differentiation mediun.
Experimental evidence indicates that bone marrow–derived adipocytes remain plastic and can be dedifferentiated back into fibroblast-like stem cells that can be induced subsequently to differentiate to osteoblasts.32 The increased mineralization with chemerin or CMKLR1 knockdown during adipogenic-to-osteoblastogenic interconversion of cells suggests that loss of chemerin/CMKLR1 signaling results in greater plasticity and greater retention of osteoblastogenic potential in cells that are partially committed to the adipocyte lineage. The broader significance of this finding is that in addition to stem cell precursors and preosteoblasts, a resident population of preadipocytes within bone marrow may be an additional source of cells that, under appropriate circumstances (e.g., inhibition of chemerin/CMKLR1 signaling) can be recruited to generate mature bone-forming osteoblasts. Given that knockdown of chemerin or CMKLR1 in BMSCs resulted in increased mineralization in the interconversion protocol, it is likely that this signaling pathway acts not only to promote adipogenesis but also to actively suppress osteoblastogenesis. Of note, we observed that chemerin or CMKLR1 knockdown in BMSCs resulted in a loss of PPARγ expression 3 days after the induction of adipogenesis (see Fig. 5A), a period identical to that used in the interconversion protocol. Since increased PPARγ activation and expression is known to suppress osteoblastogenesis,9–11 the enhanced mineralization observed in response to chemerin or CMKLR1 knockdown in the context of the osteoblastogenic differentiation protocols may be a consequence of the direct effects of loss of this signaling pathway and/or loss of PPARγ signaling. Regardless of the basis of the effect, collectively these results suggest that the changes induced by chemerin or CMKLR1 knockdown on mineralization are due to a combination of promoting osteoblast differentiation and inhibiting adipocyte differentiation. Intriguingly, the significant increase of cell mineralization with chemerin or CMKLR1 knockdown, despite a decrease in the cells' ability to proliferate, suggests a particularly effective induction of differentiation in a smaller subpopulation of cells.
The critical role of PPARγ in the regulation of gene expression during adipocyte differentiation and the close temporal relationship between PPARγ and chemerin expression during adipogenesis were strongly suggestive of a regulatory and functional relationship between these two genes. Given that ligand activation of PPARγ and/or forced expression of PPARγ markedly increased chemerin mRNA levels in both BMSCs and nonadipogenic MCA38 cells, it is likely that the induction of chemerin is not a generalized consequence of adipogenesis but rather reflects a direct or indirect action of PPARγ on the promoter of the chemerin gene. Functionally, the partial rescue of adipogenesis by forced expression of PPARγ in BMSCs, coincident with knockdown of chemerin or CMKLR1 expression, suggests that PPARγ may either activate adipogenic events downstream of chemerin signaling or act on adipogenic signaling pathways distinct from that of chemerin/CMKLR1. This partial rescue of adipocyte differentiation also resulted in increased (versus forced expression of GFP) chemerin mRNA levels, even in the presence of chemerin shRNA. At present, it is unclear if this reflects an obligatory role for chemerin in the PPARγ rescue or if it simply reflects the positive regulation of chemerin gene expression by this transcription factor. Interestingly, forced expression of PPARγ blocked the induction of the osteoblastogenic transcription factor osterix that was seen with chemerin or CMKLR1 knockdown. This further reinforces the functional interaction between PPARγ and chemerin in the context of both adipogenesis and osteoblastogenesis. Given this interaction and experimental evidence linking PPARγ with bone loss associated with aging and osteoporosis,9–11 it is also tempting to speculate that chemerin/CMKLR1 signaling may be similarly implicated. In support of this, BMSC chemerin mRNA and plasma chemerin levels are elevated (approximately two- and threefold, respectively) in mice older than 12 months of age compared with those 2 months of age (data not shown). Consistent with these data, a number of human studies have reported a positive correlation between circulating chemerin levels and age.24, 49, 50 While there is no doubt that additional study is needed, these findings do suggest the possibility of a role for chemerin in age-related bone loss and targeting chemerin/CMKLR1 signaling as a therapeutic approach in disorders of bone loss.
In summary, we have provided evidence that chemerin acting through the CMKLR1 receptor plays a critical role in promoting the adipogenic differentiation of bone precursor cells while negatively regulating osteoblast differentiation. Notably, the interaction between PPARγ and chemerin established in this study provides additional mechanistic insight into the role of chemerin in the cross-talk between adipogenic and osteoblastogenic differentiation in bone marrow. Importantly, we also have shown that under conditions mimicking the bone marrow milieu in vivo, chemerin signaling promotes adipocyte function while impairing bone mineralization. We have further demonstrated that chemerin signaling is involved in the proliferation of bone precursor cells. Taken together, these data point to the possibility of targeting bone marrow BMSCs as well as adipocytes as a means to promote bone formation in disease conditions associated with bone loss.
The authors state that they have no conflicts of interest.
This work was supported by an operating grant from the Canadian Institutes of Health Research. SM and AAR are recipients of scholarship support from the Nova Scotia Health Research Foundation.