Glycogen Synthase Kinase-3β Inhibition Links Mitochondrial Dysfunction, Extracellular Matrix Remodelling and Terminal Differentiation in Chondrocytes

Following inflammatory stimuli, GSK3 inhibition functions as a hub with pleiotropic effects leading to cartilage degradation. However, little is known about the effects triggered by its direct inhibition as well as the effects on mitochondrial pathology, that contributes to osteoarthritis pathogenesis. To this aim we assessed the molecular mechanisms triggered by GSK3β inactivating stimuli on 3-D (micromass) cultures of human articular chondrocytes. Stimuli were delivered either at micromass seeding (long term) or after maturation (short term) to explore “late” effects on terminal differentiation or “early” mitochondrial effects, respectively. GSK3β inhibition significantly enhanced mitochondrial oxidative stress and damage and endochondral ossification based on increased nuclear translocation of Runx-2 and β-catenin, calcium deposition, cell death and enhanced remodelling of the extracellular matrix as demonstrated by the increased collagenolytic activity of supernatants, despite unmodified (MMP-1) or even reduced (MMP-13) collagenase gene/protein expression. Molecular dissection of the underlying mechanisms showed that GSK3β inhibition achieved with pharmacological/silencing strategies impacted on the control of collagenolytic activity, via both decreased inhibition (reduced TIMP-3) and increased activation (increased MMP-10 and MMP-14). To conclude, the inhibition of GSK3β enhances terminal differentiation via concerted effects on ECM and therefore its activity represents a tool to keep articular cartilage homeostasis.

. Insets with sample details and original western blots showing expression of GSK3α and β phosphorylated and non phosphorylated forms in chondrocytes cultured in monolayer (A) and micromasses (B). To detect GSK3α and β we exploited western blot analysis using different antibodies (detailed in the paper and in the table) and conditions to maximize protein resolution and separation in the 50-60 kDa range (NuPAGE Novex 4-12% Bis-Tris Gel run with 1x MES running buffer). Moreover, we used Carestream Molecular Imaging Software to accurately assess the molecular weight of the bands exploiting a proper molecular weight marker (Novex Sharp Pre-stained Protein Standard, https://www.thermofisher.com/order/catalog/product/LC5800). To assess the molecular weight of western blot stained bands the pre-stained bands of the marker were highlighted by mean of a Glow Writer pen (http://divbio.com/glow-writerpen.aspx). Images show pictures of the pre-stained molecular weight markers run in each gel along with their corresponding chemiluminescent pattern. We carried out an accurate automatic assessment of the molecular weight of GSK3α and β unphosphorylated and phosphorylated forms and found the following:

GSK3beta
(n=12) The finding of GSK3β being the prevalent form expressed is in keeping with the information available in Proteinatlas.org: the expression of GSK3β (http://www.proteinatlas.org/ENSG00000082701-GSK3B/cell#human) is much more constitutive than that of GSK3α (http://www.proteinatlas.org/ENSG00000105723-GSK3A/cell). #7291 antibody allows to appreciate both the phosphorylated and the non-phosphorylated form, with the former migrating slightly slower than the other.

B.
Probing with the validated Santa Cruz Antibody #7291 (http://www.proteinatlas.org/ENSG00000082701-GSK3B/antibody) indicated that GSK3β is the only GSK3 isoform detectable in chondrocytes cultured in micromasses in our experimental settings using cells of four different patients. The upper blot reports results obtained with samples of 1 week micromasses established with cells derived from three different patients, besides one sample of monolayer chondrocytes. The lower blot shows data obtained with micromasses established with cells of another patient following GSK3β inhibiting stimuli. The relative expression of GSK3β is much higher in monolayer compared to micromass chondrocytes. The prevalent form is the non phosphorylated GSK3β, that is also the only one evident in monolayer. No bands are appreciable running at about 60 kDa, the molecular weight corresponding to GSK3α in our conditions. In the following western blot the phosphorylation of both GSK3β and GSK3α after GSK3 inactivating stimuli was investigated using the Cell Signaling antibody #9331. Shown are the results obtained with samples at 1, 2 and 3 weeks maturation. The pattern of phosphorylation obtained in 1 week micromasses indicated that, in this setting best mimicking that of healthy articular chondrocytes, only the GSK3β isoform undergoes phosphorylation. All the experiments (but that shown in Figure 5A) presented in the manuscript have been carried out in 1 week micromasses. GSK3α undergoes phosphorylation only at 3 weeks maturation, but still, the phosphorylation of GSK3β is much stronger. Collectively these results emphasize the role of GSK3 inactivation in terminal differentiation and the major role of GSK3β.
In both A and B the black solid vertical lines separate bands that were not adjacent as derived from the different full blots shown in the lower part of the figure.
The dotted black vertical lines delineate the markers that were run in each gel and highlighted by mean of a Glow Writer pen before luminescent detection of the signals. All molecular weight markers run adjacent to the bands shown beside them, but after transfer and before membrane probing were cut and re-aligned for the detection.

S3
Supplementary Fig. S3. "Long term" treatment with either 5mM LiCl or 10µM SB216763 increased nuclear translocation of active β-catenin as assessed by confocal analysis of the staining intensity in the nuclear area as mapped by the nuclear probe Sybr Green. Graphs report the mean nuclear intensity of immunofluorescent staining in several 60x fields taken from sections of micromasses at 1 week of culture. The means are relative to several hundreds (n = 2137, NS; 1441, LiCl and 2043, SB216763 of different cell nuclei as evaluated by the Student's t-test for independent samples. ***P < 0.001 Supplementary Fig. S4. LiCl but not SB216763 treatment led to a statistically significant increased calcium deposition. The picture reports data derived from 7 experiments with 1 week micromasses established with cell of different patients (n=7, *P < 0.05). Calcium content was quantified by Quantichrom DICA-500 assay. Given the high variability of the data, the Wilcoxon signed rank test was employed to compare each GSK3β inhibiting treatment with the basal condition.  Figure 5A of the main manuscript. C: Full blots used to derive the phosphoGSK3 α and β results shown in Figure 5A of the main manuscript obtained with micromass lysates at 1-2-3 weeks in unstimulated condition or stimulated with either LiCl or SB216763 or insulin. The dashed rectangles indicate the bands included in the crops shown in D. D: Assembly of the crops of the full blots shown in C, indicated by the dashed rectangles. The black solid rectangles indicate the bands used to setup phosphoGSK3 α and β figure in Figure 5A of the main manuscript. Therefore, in figure 5A of the main manuscript black lines highlight splicing between bands that were not adjacent in the full blot. E: Full blot corresponding to the results of phosphoGSK3β shown in Figure 5B of the main manuscript along with its loading control and obtained with micromass lysates at 1 week in unstimulated condition or stimulated with either LiCl or SB216763 or insulin. The dashed rectangle indicates the bands included in the crop used in figure 5B of the main manuscript. The arrow indicates the cropped bands in supplementary figure 5F. F: Crop of the original phosphoGSK3β results shown in E as indicated by the arrow along with its corresponding GAPDH loading control. The black solid rectangle indicates the phosphoGSK3β/GAPDH results shown in Figure 5B of the main manuscript.