TGFβ1-induced SMAD2/3 and SMAD1/5 phosphorylation are both ALK5-kinase-dependent in primary chondrocytes and mediated by TAK1 kinase activity

Primary bovine chondrocytes were isolated from the metacarpophalangeal joint of cows (2–5 years old) obtained from a slaughterhouse within 3 hours post mortem. Chondrocytes were obtained by incubating cartilage slices overnight in Collagenase B (for details see [21]). Chondrocytes were seeded in DMEM/F12 1:1 containing 10% fetal calf serum (Thermo Scientific UK) at a density of 0.5 × 10⁵ cells per cm² in 6-well plates (Greiner Bio-one International, the Netherlands) for protein studies, or in 24-well plates (Bio-one International, the Netherlands) for mRNA experiments. Cells were cultured for 1 week at 37 °C and 5% CO₂ before the start of the experiments. For explant studies, four 7-mm² explants per condition were pooled in 1 ml of medium. Both monolayer and explant studies were repeated in different donors.

Adenoviruses were used to induce expression of constitutively active (ca)ALK1 [4], constitutively active ALK5 [4] and the CAGA₁₂-luciferace reporter construct [22] in primary chondrocytes. Cells were rinsed twice with saline, and adenoviruses were added in a multiplicity of infection of 200 for 3 h at 37 °C in a minimal volume of DMEM/F12 1:1. Hereafter, cells were washed twice with saline, and DMEM/F12 1:1 was added. Cells were used 48 h after transfection.

Before stimulation, cells were deprived of serum for 24 h and thereafter stimulated with 1 ng/ml recombinant human TGFβ1 (Biolegend, the Netherlands) for 2 or 24 h. In experiments where inhibitors were used, DMSO was used as vehicle control. To block TAK1 activity, we used (5Z)-7-Oxozeaenol [17] (Tocris Bioscience) in a concentration of 0.5 μM. To inhibit ALK5 kinase, we used SB-505124 [15] (Sigma Aldrich) in a concentration of 5 μM. For inhibition of ALK1 kinase, LDN-193189 [14] (Axon Medchem) was used in a concentration of 0.05 μM. This concentration of LDN-193189 is well above its reported half maximal inhibitory concentration (IC₅₀), 0.8 nM for ALK1, but far below its IC₅₀ for ALK5 of 350 nM [16, 23]. Cells were pre-incubated with the inhibitors for 1 h prior to addition of TGF-β1. Either 2 or 24 h after addition TGF-β1, medium was removed and TRI-reagent was added for RNA isolation.

To isolate proteins from monolayer cultures, cells were lysed on ice using lysis buffer (Cell signaling, USA) containing a protease inhibitor cocktail (complete, Roche Diagnostics, Germany) and subsequently sonicated on ice using a Bioruptor (Diagenode, USA). With a BCA-assay (Thermo Scientific, USA) protein concentration was measured and after addition of Laemmli sample buffer, samples were boiled for 5 minutes at 99 °C.

To isolate proteins from explants, four 7-mm² explants were homogenized using a Mikro-dismembrator (B. Braun, Germany) and dissolved in 1 ml ice cold radioimmunoprecipitation assay (RIPA) buffer with added 1 mM Na₃VO₄ and protease inhibitor cocktail (Roche Diagnostics, Germany). After 1 h incubation on a roller bench at 4 °C, samples were spun down for 3 minutes at 10⁴ × g and the pellet was discarded. Cetylpyridinium chloride was added up to a concentration of 1% (m/v) and samples were incubated on a roller bench for 1 h at 4 °C. Hereafter, samples were spun down twice for 15 minutes at 10⁴ × g at 4 °C and the pellet was discarded. Using 10 kDa centrifugal filter units (Millipore, USA) the supernatant was concentrated to a volume of 50 μl. Subsequently, proteins were precipitated by addition of 950 μl of 20% m/v trichloroacetic acid and 0.1% m/v dithiothreitol in aceton at -20 °C and spinning samples down for 15 minutes at 6700 × g at 4 °C. Next, the pellets were washed three times with 0.1% m/v dithiothreitol in aceton at -20 °C. Finally, the pellets were dried under vacuum for 20 minutes and dissolved in 100 μl 1% m/v sodium dodecyl sulfate in 100 mM tris(hydroxymethyl)aminomethane in H₂O pH 9.0.

Proteins were seperated on a 7.5% Bisacrylamide gel, and transferred to a nitrocellulose membrane using wet transfer (Towbin buffer, 2.5 h 275 mA at 4 °C). After overnight incubation at 4 °C with 1:1000 polyclonal Rabbit anti P-Smad1/5 (S463/465)/Smad8 (S426/428) (Cell signaling, USA) or anti P-Smad2 (S465/467) (Cell signaling, USA), membranes were incubated with 1:1500 polyclonal Goat anti Rabbit labeled with horseradish peroxidase (HRP) (DAKO, Belgium) for 1 h. Hereafter, enhanced chemiluminescence using ECL plus (GE Healthcare, UK) was used to visualize the proteins. To visualize overexpression of constitutively active ALKs, a rabbit polyclonal antibody directed against their internal HA tag was used: HA-probe Antibody (Y-11) (Santa Cruz, USA) (1:1000). As loading control either Gapdh was stained with an anti-Gapdh mouse mAb (1G5) (Sigma Aldrich, Germany) (1:10 000) in combination with IRDye 680RD Donkey anti mouse (1:10 000) (Licor, USA) using the Odyssey detection system (Licor, USA) or Vinculin with a rabbit pAb (H300) (Santa Cruz, USA) (1:1000) in combination with HRP-labeled Goat-anti Rabbit (1:2000) (DAKO, Denmark) using ECL. Finally, blots were quantified using ImageJ.

TRI-reagent (Sigma-Aldrich, Germany) was used for RNA isolation according to the manufacturer’s protocol (for details see [21]). RNA concentration was measured using a Nanodrop photospectrometer (Thermo Scientific, USA). Per sample, 1 μg of RNA was treated with DNAse (Life Technologies, USA), which was subsequently inactivated at 65 ° C with 1 μl 25 mM EDTA (Life Technologies, USA). To perform reverse transcriptase reaction, 1.9 μl ultra pure water, 2.4 μl 10 × DNAse buffer, 2.0 μl 0.1 M DTT, 0.8 μl 25 mM dNTP, 0.4 μg oligo dT primer, 1 μl 200 U/μl M-MLV Reverse transcriptase (all Life Technologies, USA) and 0.5 μl 40 U/μl RNAsin (Promega, the Netherlands) was added, and samples were incubated for 5 minutes at 25 °C, 60 minutes at 39 °C, and 5 minutes at 95 °C using a thermo cycler. The obtained cDNA was diluted 10 times in ultra pure water, and gene expression was measured using 1 μM of validated cDNA-specific primers (Biolegio, the Netherlands; see Table 1) in a quantitative real-time polymerase chain reaction (qPCR) using SYBR green master mix (Applied Biosystems). The following protocol was used: after 10 minutes at 95 °C, 40 cycles of 15 sec at 95 °C and 1 minute at 60 °C each were run. Hereafter a melting curve was made to verify gene-specific amplification. For calculations of the -ΔCt, two reference genes were used: glyceraldehyde 3-phosphate dehydrogenase and ribosomal protein S14.

All quantitative data are expressed as a mean of multiple repeats ± SD. For every analysis data were checked for normality using the Shapiro-Wilk test. One-way analysis of variance (ANOVA) with Tukey multiple comparison post-hoc test was used to determine the significance. The statistical analyses were performed using Graphpad Prism 5.0 software.

To begin, we confirmed in primary bovine chondrocytes the expression of the receptors that TGFβ can use to induce Smad2/3 phosphorylation i.e. ALK5 or Smad1/5 phosphorylation, i.e. ALK1, ALK2 and ALK3 [3, 4]. All four receptors were readily detected using qPCR (Additional file 1: Figure S1), and indeed confirmed that stimulation of chondrocytes with TGFβ1 results in both Smad2 and Smad1/5 phosphorylation (Fig. 1a).

Subsequently, we analyzed the effects of the ALK4/5/7-kinase inhibitor SB-505124, the ALK1/2/3-kinase inhibitor LDN-193189 and the TAK1-kinase inhibitor (5Z)-7-Oxozeaenol on TGFβ1-induced R-Smad phosphorylation (Fig. 1a, b and c). Strikingly, addition of 5 μM SB-505124 not only inhibited Smad2 phosphorylation, but also pSmad1/5 phosphorylation. Remarkably, treatment with 0.5 μM (5Z)-7-Oxozeaenol also resulted in inhibition of TGFβ1-induced Smad1/5 and Smad2 phosphorylation. Noteworthy, addition of 0.05 μM LDN-193189 did not affect TGFβ1-induced Smad1/5 phosphorylation but did significantly enhance TGFβ-induced pSmad2.

To investigate if the striking effect of SB-505124 on Smad1/5 phosphorylation was a dose-dependent effect, we performed a dose-response experiment (Fig. 1d and e). This experiment showed that at all tested doses (0.1, 0.5, 1, 5, 10 and 25 μM), SB-505124 inhibited TGFβ1-induced Smad1/5 phosphorylation, and that Smad1/5 phosphorylation was inhibited more potently than that of Smad2. Therefore, to address if ALK1 inhibition was a possible off-target effect of SB-505124, this compound was added to cells over-expressing constitutively active (ca)ALK1 or caALK5. This experiment showed that SB-505124 did not affect caALK1-induced Smad1/5 phosphorylation (Fig. 2a and b), but did inhibit caALK5-induced Smad2 phosphorylation (Fig. 2a and b). Therefore, SB-505124 is specific for ALK5 kinase and does not inhibit ALK1 kinase. The same experimental setup was used to address the specificity of (5Z)-7-Oxozeaenol, and no significant inhibitory effect of this compound on either caALK1 or caALK5 function was detected, making it unlikely that either the ALK1 or ALK5 kinase domain is an off-target of (5Z)-7-Oxozeaenol.

To investigate if the lack of effect of LDN-193189 on TGFβ1-induced Smad1/5 was maybe dose-dependent, we added a 10 times higher dose of LDN-193189 (0.5 μM) to the cells, but were still unable to inhibit TGFβ1-induced Smad1/5 phosphorylation (Fig. 2c). Therefore we sought to validate the function of LDN-193189. To do this, we added 0.05 μM LDN-193189 to chondrocytes over-expressing caALK1 and observed that this blocked caALK1-induced Smad1/5 phosphorylation (Fig. 2a and b), Furthermore, the effect of 0.05 μM LDN-193189 on bone morphogenic protein (BMP) signaling was also investigated (Fig. 2d) by stimulating primary chondrocytes for 1 h with either the ALK2/3 ligands BMP2 or BMP7 [24] or the ALK1/2 ligand BMP9 [25, 26]. We observed that BMP-induced Smad1/5 phosphorylation was clearly inhibited by 0.05 μM LDN-193189. Together, these experiments strongly confirm the function of 0.05 μM LDN-193189 in primary chondrocytes.

In conclusion, in primary chondrocytes, TGFβ1-induced Smad1/5 phosphorylation is unaffected by LDN-193189, whereas both Smad2/3 and Smad1/5 phosphorylation are inhibited by SB-505124 and (5Z)-7-Oxozeaenol.

Next, we analyzed the effects of TGFβ1 signaling inhibitors on TGFβ1-induced Smad-dependent gene expression. For this, bId1 and bSerpine1 were used as markers of pSmad1/5 and Smad3 signaling, respectively [4, 27] [22, 28]. Expression of bId1 was profoundly upregulated by stimulation with 1 ng/ml of TGFβ1 (Fig. 3a). In accordance with our western blot data on pSmad1/5, expression of bId1 was inhibited by both SB-505124 and (5Z)-7-Oxozeaenol, and was unaffected by LDN-193189. Expression of bSerpine1 was upregulated by 1 ng/ml of TGFβ1 and fully inhibited by SB-505124, but was only partly inibited by (5Z)-7-Oxozeaenol (Fig. 3b). To confirm these effects on pSmad3-dependent gene expression, we used the CAGA₁₂-luc construct, which produces luciferase specifically in response to pSmad3. After 8 h, TGFβ-induced CAGA₁₂-luc activity was significantly inhibited by (5Z)-7-Oxozeaenol and SB-505124, showing that pSmad3-signaling is indeed lowered in these conditions. When the effect of the inhibitors on multiple TGFβ1-induced (Fig. 3c), genes like bSmad7, bMmp3, bCol2a1, bNgf, bAlk5, bJunb, bFN1 and bTgfb1 was measured, none of these genes was significantly downregulated by LDN-193189, unlike SB-505124, which profoundly downregulated bSmad, bNgf, bAlk5, bJunb, bFN1 and bTgfb1, or unlike treatment with (5Z)-7-Oxozeaenol, which significantly lowered bNgf, bTgfb1 and bMmp3 expression (Fig. 3d). Because a role for pSmad1/5 in the regulation of e.g. Smad7, Junb and Col2a1 has been established [29‐31], these data indicate that the ALK5 kinase inhibitor SB-505124 inhibits gene expression downstream of both pSmad2/3 and pSmad1/5, whereas (5Z)-7-Oxozeaenol does so to a lesser extent and in a more limited amount of genes. No effect of LDN-193189 was observed on TGFβ1-induced gene expression.

As TGFβ1 signaling is known to be affected by cellular context, we sought to validate our observations in monolayer chondrocytes in intact cartilage ex vivo. Therefore we stimulated cartilage explants with 1 ng/ml TGFβ1 with or without inhibitors, and investigated Smad phosphorylation. Explants were stimulated for 2 h to allow for diffusion of TGFβ1 through the cartilage matrix. In explants, 1 ng/ml TGFβ1 induced pSmad2, but Smad1/5 phosphorylation was more difficult to detect (Fig. 4a), in contrast to in monolayer cultures. Therefore BMP9 was used as positive control, which, in comparison to TGFβ, resulted in far stronger induction of pSmad1/5 (for quantification see Additional file 2: Figure S3). A smaller amount of pSmad1/5 was detected in explants treated with TGFβ1 and SB-505124, (5Z)-7-Oxozeaenol or LDN-193189 compared to TGFβ1-treated samples alone. TGFβ-induced phosphorylation of Smad2 was clearly inhibited by SB-505124 and partly by (5Z)-7-Oxozeaenol (Fig. 4a), but on gene expression bSerpine1 was only significantly inhibited by SB-505124 and not by (5Z)-7-Oxozeaenol (Fig. 4b). In explants, expression of bId1 could not be used as a read-out for pSmad1/5 signaling, because TGFβ1 downregulated its expression, in contrast to its induction in monolayer experiments. This is a well-known Smad3-dependent cytostatic effect of TGFβ1 [32], and further indicates that TGFβ1 signaling differs between chondrocytes in monolayer and chondrocytes in intact cartilage. Notably, LDN-193189 did downregulate bId1 expression, even in absence of TGFβ1 (Additional file 3: Figure S2), demonstrating the bioactivity of this compound in cartilage explants. Again, when multiple, TGFβ1-induced, Smad-dependent, genes were measured (Fig. 4c) SB-505124 inhibited TGFβ1-induced expression of bSmad7, bNgf, bFn1 and bTgfb1, and (5Z)-7-Oxozeaenol inhibited TGFβ1-induced expression of bFn1, bTgfb1, whereas no effect of LDN-193189 was observed. Therefore, SB-505124 also inhibits both TGFβ1-induced Smad1/5 and Smad2 phosphorylation and gene expression in intact cartilage, whereas LDN-193189 does not. Furthermore, (5Z)-7-Oxozeaenol partly inhibited both TGFβ1-induced Smad1/5 and Smad2 phosphorylation.