Background
Bone Morphogenetic Proteins (BMPs) are critical in the formation of cartilage and bone. Osteogenic BMPs, such as BMP-2, and -7 are widely recognized to promote an osteogenic response [
1]. Transforming Growth Factor-β (TGF-β) belongs to the same superfamily as the BMPs, although its role in bone is less clear. Both BMPs and TGF-β bind to related Type I receptors (also known as Activin Receptor-like Kinases or ALKs) and Type II receptors and activate downstream SMAD signaling pathways [
2‐
4]. The osteogenic BMPs can bind to ALKs 1/2/3/6 and BMPRII or ActRII to induce phosphorylation of the receptor-regulated SMADs (R-SMADs) 1/5/8. In the case of TGF-β and the non-osteogenic BMPs, ligand binding to receptors such as ALKs 4/5/7 and TβRII induces phosphorylation of R-SMADs 2/3.
There is conflicting evidence on the effects of TGF-β signaling on bone formation. TGF-β isoforms are robustly expressed during the early stages of bone healing [
5,
6], and exogenous TGF-β has been purported to augment bone markers in cultured human osteoblasts [
7] and can lead to improvements in bone repair in orthopaedic animal models [
8‐
10]. However, in cultured murine cell lines, TGF-β acting through SMAD3 was reported to antagonize osteogenesis [
11,
12] and comparable findings were found in human mesenchymal stem cells [
13]. Further work suggests that exogenous TGF-β can delay osteogenesis in favor of chondrogenesis [
14]. In addition to direct effects on osteogenic differentiation, TGF-β may also lead to increased fibrosis. In rodent distraction osteogenesis and fracture models, TGF-β1 and TGF-β2 treatment (respectively) did not lead to improved outcomes but did result in increased fibrous and cartilage tissue [
15,
16]. In these studies, inflammation and edema were also reported as unfavorable side-effects.
TGF-β signaling has also been linked to other fibrotic conditions, such as the genetic disorder Marfan syndrome. Animal models with aberrant TGF-β signaling have been successfully treated with TGF-β neutralizing antibody or with Losartan, a small-molecule angiotensin II AT
1 receptor blocker (ARB) [
17‐
19]. ARBs are now under trial for Marfans syndrome [
20], and may be applicable for other TGF-β related disorders. However, the affects of ARBs on TGF-β protein expression are indirect and do not appear to translate to bone [
21], thus making these agents less attractive for orthopaedic applications. In contrast, a novel synthetic compound, SB431542, has been shown to rapidly and selectively inhibit ALK-4/5/7 but not ALK-2/3/6 kinase activity [
22]. This enables the blockade of the classical TGF-β-SMAD2/3 signaling pathway whilst allowing osteogenic BMP-SMAD1/5/8 signaling.
In a seminal study by Maeda
et al. (2004), SB431542 repression of TGF-β signaling was found to enhance osteoblastic differentiation in BMP-2 treated C2C12 myoblasts [
23]. Osteoblastic differentiation and matrix mineralization were also increased in cultured human mesenchymal stem cells. Based on these
in vitro findings, we speculated that this compound may also be able to positively influence bone formation or healing. As a putative anti-fibrotic agent, SB431542 could have additional benefits in the context of orthopaedic repair where fibrosis is problematic.
In this study we have used both in vitro and in vivo methods suitable for the rapid screening of compounds specifically for orthopaedic applications. These assays represent a systematic approach that can be readily applied to other putative pro-osteogenic agents. In cell culture experiments, we treated the MC3T3-E1 pre-osteoblast cell line with purified recombinant BMP-2, purified TGF-β1, and the TGF-β receptor inhibitor SB431542, individually and in combination. Outcome measures included alkaline phosphatase (AP) and mineralization staining, osteogenic gene expression, and activation of downstream SMAD signaling pathways. Next, we attempted to translate the effects of TGF-β inhibition using animal models. This included a marrow ablation model (where intramedullary reaming produces bone formation over a 10-day period via intramembranous ossification), and BMP-2 implantation (where ectopic bone nodules are induced in muscle over 3 weeks via endochondral ossification). This study design represents a straightforward methodology for testing prospective orthopaedic agents.
Methods
Cell culture methods
MC3T3-E1 pre-osteoblasts were grown in α-MEM media containing 10% FBS (Invitrogen, Carlsbad, CA, USA). Passage number 20 cells were used, and cultured for no more than 2 weeks prior to initiating differentiation. Osteogenic differentiation was instigated by supplementing media with 50 mg/L ascorbic acid and 10 mM β-glycerophosphate (Sigma Aldrich, St Louis, MO, USA). All culture media contained 2 mM L-glutamine, and antibiotics (100 units/ml penicillin and 0.1 mg/ml streptomycin) (Invitrogen). Cultures were grown in 37°C incubators at 5% CO2 with media changes every 2-3 days. For staining experiments cells were plated in 48-well plates at 5 × 104 cells/well. For cDNA or protein collection experiments cells were seeded in 6-well plates at 2 × 105 cells/well. Cells were plated overnight and were sub-confluent prior to the addition of drugs or recombinant proteins.
Recombinant proteins and drugs
Recombinant human BMP-2 (rhBMP-2) (Medtronic Australasia, North Ryde, NSW, Australia) was solubilized in sterile water at stock concentrations of 100 μg/ml. Transforming growth factor-beta 1 (TGF-β1) was purchased from Sigma Aldrich (T1654), and reconstituted at 1 μg/ml in filtered 0.1% BSA in 4 mM HCl. The ALK-4/5/7 inhibitor SB431542 was purchased from Sigma Aldrich (S4317) and solubilized in dimethylsulfoxide (DMSO) at stock concentration of 10 mM. For
in vitro experiments, all wells received the same volume of DMSO (<0.1%) to avoid the confounding effects of DMSO on osteogenic differentiation [
24]. PTH
(1-34) peptide was purchased from Auspep (Parkville, VIC, USA).
Alkaline phosphatase (AP) and Alizarin Red S staining
Prior to staining, all cells were washed with PBS and fixed with 4% paraformaldehyde (PFA). AP staining was performed using a solution containing 0.5 mg/ml naphthol AS-BI phosphate (Sigma Aldrich), 0.5 mg/ml Fast Blue (Sigma Aldrich), 10% N,N-dimethylformamide (Sigma Aldrich), 0.5% MgCl
2 in 0.1 M Tris (pH 9.4) for 10 min at room temperature [
25]. Mineralization of calcium deposits was assessed by Alizarin Red S staining (40 mM, pH 4.2) (LabChem, Pittsburgh, PA, USA) for 10 min at room temperature, followed by multiple washes with distilled water to remove background staining [
26].
Total RNA was isolated using TRIZOL reagent (Invitrogen). Equivalent amounts of total RNA were used to synthesize cDNA by reverse-transcription using Superscript III Reverse Transcriptase (Invitrogen). All samples were amplified using the SYBR Green PCR reagent kit (Integrated Sciences, Chatswood, NSW, Australia). PCR was performed on the Rotor-Gene 3000 (Corbett Life Science, Sydney, NSW, Australia). PCR primer sequences and amplification conditions are as previously published [
27]. PCR reactions were performed in quadruplicate and normalized to the housekeeping
Gapdh. Data are presented as mean fold induction relative to untreated cells, with the standard deviation shown in parentheses.
Western blotting
Cell extracts were harvested in SDS sample buffer, denatured at 95°C for 10 min, separated by electrophoresis on a 1% SDS-PAGE gel, and then transferred by semi-dry electroblotting to PVDF membrane (Millipore, Billerica, MA, USA) [
27]. Immunoblotting was performed with anti-Smad1 and anti-Smad2 (Zymed/Invitrogen), anti-pSmad1 and anti-pSmad2 (Cell Signaling Technology, Boston, MA, USA), and α-Tubulin (Sigma-Aldrich) antibodies. Horseradish peroxidase-conjugated donkey anti-rabbit IgG and sheep anti-mouse IgG secondary antibodies were used (GE Healthcare, Little Chalfont, Buckinghamshire, UK). Chemiluminescence reagents (New England Biolabs, Ipswich, MA, USA) were used to visualize the blots on X-ray film.
Surgical models
Experiments using mice were approved by the SWAHS Animal Ethics Committee (Protocol 5031). In all studies, 8-10 week old C57BL6/J mice were used and anesthesia was induced with inhaled isofluorane. Pain was managed using buprenorphine (0.05-0.1 mg/kg) subcutaneously (s.c.) postoperatively (then every 12 hours as required). Dehydration was managed by saline injection as required.
Marrow Ablation Model
New intramedullary bone was induced by femoral reaming as previously described in rats [
28] and mice [
29]. An incision was made on the right hind leg lateral to the knee and a 25G needle pushed through the cortex of the distal femur between the intercondyles and unilaterally up the medullary canal to loosen the marrow. This needle was removed and 0.1 ml of saline was injected into the cavity to flush out the marrow using a 30G needle. The wound was closed using 5-0 nylon suture (Ethicon). Mice were dosed s.c. daily with SB431542 in DMSO from the day of surgery until sacrifice at day 10. Control groups received an equivalent volume of vehicle (DMSO) or the bone inducing agent 25 μg/kg/day PTH
1-34 (Auspep). In an initial study, 1 mg/kg/day SB43152, DMSO control, and PTH
1-34 groups were examined (N = 6). In a second study, a wider dose range was used of 0.1 mg/kg/day, 1 mg/kg/day, and 10 mg/kg/day SB431542 (N = 6) with DMSO controls (N = 6) and a nominal number of repeated PTH
(1-34) controls (N = 2).
BMP-2 Implantation Model
New bone formation in a quadriceps muscle pouch was assayed after 3 weeks following the implantation of a BMP-2 containing pellet. Pellets containing 20 μg recombinant human BMP-2 (Medronic Australasia) and 0 μg, 38 μg, or 192 μg SB431542 (Sigma-Aldrich) were manufactured by solubilizing Poly-(D, L-lactic acid) polymer (viscosity 0.55-0.75, average M
w 75000-12000, Sigma-Aldrich) and drugs in ethyl acetate (analytical reagent grade, Chem Supply, Gillman, SA, Australia) and evaporating the solvent under vacuum. The BMP-2 dose was previously determined empirically using this polymer delivery system and the ratio of BMP-2/SB431542 was based on
in vitro efficacy. Pellets were molded by compression in a modified 0.3 ml syringe. Pellets were surgically implanted by a surgeon experienced in mice using published methods [
25]. Group sizes of N = 8 were used.
Radiographic analysis
The bone formed in both models was examined at the experimental endpoint using a digital X-ray machine (Faxitron X-ray Corp., Wheeling, IL, United States).
To quantify the total bone mineral content (BMC) in the distal region of reamed and unreamed femora, peripheral Quantitative Computed Tomography (pQCT) and analysis was performed using a XCT-960A scanner (Stratec Medizintechnik, Pforzheim, Germany). A region spanning the midshaft to the distal femur was selected after a pre-scan and measurements were taken from ten 2 × 5 × 0.2 mm slices using a red collimation mask. These values were averaged and for each mouse the reaming-induced bone was normalized to the non-operated limb.
To quantify the total bone volume (BV, mm3) in each BMP-induced bone pellet, micro Computed Tomography (microCT) scanning was performed using SkyScan 1174 compact microCT scanner (SkyScan, Kontich, Belgium). The X-ray source voltage was set at 50 kV and beam current set at 800 μA. To minimize the beam hardening artifacts a 0.5 mm aluminum X-ray beam filter was used to attenuate soft X-rays at the source. Samples were scanned at 8.7 μm resolution and 0.4° angular increments. Acquisition time was approximately 30 minutes/specimen. BV analysis was performed using CTAnalyser software, version 1.9.2.3 (SkyScan). A global grey value threshold representing new bone formation (47-255) was defined and applied to all samples. Total bone volume was calculated for the entire pellet. Representative three-dimensional bone pellets were reconstructed with transaxial slices (100 slices) of the pellet's mid-section, using CTVol Realistic Visualisation software version 2.1.0.0 (SkyScan).
For statistical analysis of in vivo data, group sizes <10 necessitated stringent non-parametric statistical tests. Kruskal Wallis and Mann Whitney U tests were performed using SPSS Statistics version 17 (SPSS Inc., Chicago, IL, United States).
Histological Analysis
Samples were harvested and fixed in 4% PFA and stored in 70% ethanol. Bone was decalcified at room temperature on a shaker in 0.34 M EDTA (pH 8.0) containing 0.5% PFA for 3 days, followed by 38 days in 0.34 M EDTA (pH 8.0) with changes every 3-4 days. Samples were embedded in paraffin blocks and 5 mm-thick sections cut. Sections were stained with Picrosirius red and Alcian blue for bone and cartilage.
Discussion
The study by Maeda
et al. (2004) used an
in vitro system to characterize the response of myogenic progenitors to BMP signaling and treatment with SB431542 [
23]. They showed that SB431542 enhanced the effects of BMP-2 on osteogenesis and that this was associated with increased SMAD1 signaling and decreased SMAD2 signaling. Our data using the MC3T3-E1 cell line supports a pro-osteogenic effect of SB431542 on pre-osteoblasts, even in the absence of exogenous BMP-2. In terms of a screening system for novel compounds, the MC3T3-E1 system is rapid, low-cost, and suitable for generating rapid dose response curves. Due to limitations with this cell line [
30,
31], prospective agents should also be trialed on primary mesenchymal stem cells, however in the case of SB431542 this data was already available [
23]. Maeda
et al also examined the expression of I-SMADs (inhibitory SMADs), which are downstream negative regulators of R-SMAD signaling, and showed a suppression of SMAD6 and SMAD7 by SB431542 with prolonged treatment [
23]. While I-SMADs represent potentially important modifiers of R-SMAD signaling, they are transcriptionally regulated by and secondary to the initial R-SMAD response. Our
in vitro data indicates that BMP and TGF-β signals can modulate R-SMAD signaling in a non-canonical fashion. Specifically, ALK-4/5/7 inhibition led to increases in pSMAD1 levels and BMP-2 treatment led to a reduction in pSMAD2 levels (Figure
1).
In this study we have also employed two rapid surgical models to screen for pro-osteogenic effects in a bone formation/bone repair context. The first was a marrow ablation model previously described in the context of biglycan null mice that show decreased bone formation following reaming [
29]. We adopted a high-resolution multi-slice pQCT scanning approach for quantification, which was found to give more accurate results than individual sections that were more susceptible to positional effects. We confirmed that reaming consistently induced new bone formation in all groups. PTH
(1-34) administration further increased the amount of bone on the reamed side, validating the model system. In contrast, SB431542 did not produce any substantive pro-osteogenic effect in reamed bones or in non-operated limbs.
The second model was a BMP-2 intramuscular implantation model [
25], which contains an endochondral bone formation component. Again, no significant increase was observed in bone formation with SB431542 treatment, rather a trend was seen towards reduced bone with local dosing (Figure
3).
The lack of a beneficial effect of the TGF-β inhibitor SB431542 in the
in vivo models may be due to several reasons. One possibility was an inappropriate dose selection, although higher doses were likely to non-specifically affect other receptors. In a previously published study, a single dose of 0.2 mg/kg was used to affect metabolic changes in rats [
32], indicating that our dose range of up to 10 mg/kg/day should be capable of producing significant physiological effects in mice. This SB431542 compound has also been successfully used in organ culture experiments to produce developmental effects [
33]. Nevertheless, the specificity and/or bioavailability of SB431542 may be suboptimal for
in vivo studies, and there certainly exists the potential for more specific inhibitor compounds to produce improved results.
An alternative explanation for the disparity between
in vitro and
in vivo results may be due to the fundamental differences between the techniques and outcome measures in the different systems. Cell culture models focus primarily on the process of cell differentiation, generally on committed bone cells [
11‐
14,
23]. In contrast, surgical models also incorporate elements of osteoprogenitor recruitment and proliferation. In the context of TGF-β, this may be critical as TGF-β release has been recently shown to play a major role in the recruitment of osteoprogenitors for bone homeostasis [
34]. Thus our study may highlight a basic limitation of
in vitro systems and stress the utility of expediting screens with surgical models such as the marrow ablation or BMP-2 implantation model.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AS conceptualized and organized the study. AM performed cell culture and histology analyses with the aid of RL, SK, and MMD. LP and KM performed surgical experiments in mice. NYCY performed quantative microCT analysis with support by PAB and AJR. DGL was the senior researcher who guided the research study. The study was primarily written by AS and DGL, although all authors read and approved the manuscript.