Interference with the contractile machinery of the fibroblastic chondrocyte cytoskeleton induces re-expression of the cartilage phenotype through involvement of PI3K, PKC and MAPKs

Abstract

Chondrocytes rapidly lose their phenotypic expression of collagen II and aggrecan when grown on 2D substrates. It has generally been observed that a fibroblastic morphology with strong actin–myosin contractility inhibits chondrogenesis, whereas chondrogenesis may be promoted by depolymerization of the stress fibers and/or disruption of the physical link between the actin stress fibers and the ECM, as is the case in 3D hydrogels. Here we studied the relationship between the actin–myosin cytoskeleton and expression of chondrogenic markers by culturing fibroblastic chondrocytes in the presence of cytochalasin D and staurosporine. Both drugs induced collagen II re-expression; however, renewed glycosaminoglycan synthesis could only be observed upon treatment with staurosporine. The chondrogenic effect of staurosporine was augmented when blebbistatin, an inhibitor of myosin/ actin contractility, was added to the staurosporine-stimulated cultures. Furthermore, in 3D alginate cultures, the amount of staurosporine required to induce chondrogenesis was much lower compared to 2D cultures (0.625 nM vs. 2.5 nM). Using a selection of specific signaling pathway inhibitors, it was found that PI3K-, PKC- and p38-MAPK pathways positively regulated chondrogenesis while the ERK- pathway was found to be a negative regulator in staurosporine-induced re-differentiation, whereas down-regulation of ILK by siRNA indicated that ILK is not determining for chondrocyte re- differentiation. Furthermore, staurosporine analog midostaurin displayed only a limited chondrogenic effect, suggesting that activation/deactivation of a specific set of key signaling molecules can control the expression of the chondrogenic phenotype. This study demonstrates the critical importance of mechanobiological factors in chondrogenesis suggesting that the architecture of the actin cytoskeleton and its contractility control key signaling molecules that determine whether the chondrocyte phenotype will be directed along a fibroblastic or chondrogenic path.

Introduction

One of the biggest concerns during in vitro culture expansion of cells is their inherent sensitivity to their microenvironment, leading to a rapid change in cell morphology and behavior upon cultivation on artificial substrates [1–4]. While primary chondro- cytes are typically round-shaped, they adopt an elongated, fibroblast-like morphology upon cultivation on 2D substrates such as tissue culture polystyrene (TCP), which is accompanied by drastic changes in the expression of specific marker proteins including the down-regulation of transcription factor SOX9 as well as collagen II (col2) with concomitant up-regulation of collagen I (col1) [5–7]. De-differentiation was shown to be reversible with fibroblastic chondrocytes re-expressing the chon- drogenic phenotype upon changing the cell shape by cultivation chondrogenic markers SOX9 and col2. Re-expression of the full chondrogenic phenotype, marked by glycosaminoglycan (GAG) synthesis, was only observed upon treatment with staurosporine. Inhibition of myosin/actin-contractility by blebbistatin further enhanced staurosporine-induced re-differentiation and cultiva- tion in 3D alginate beads effectively lowered the necessary dose of staurosporine to stimulate a chondrogenic response. Additional treatment of staurosporine-stimulated cultures with a set of specific inhibitors showed that PI3K, PKC and mitogen activated protein kinase (MAPK)-signaling pathways are involved and can be pharmacologically tuned to regulate the re-differentiation process. Considering the limited healing by current cartilage repair strategies, these findings could have important implica- tions for cell-based approaches like autologous chondrocyte implantation (ACI).

Interestingly, re-differentiation was observed with drugs that interfere with the actin cytoskeleton integrity by different means [17–19]. While members of the cytochalasin family bind to the barbed ends of actin filaments to prevent actin polymerization, ultimately causing filament disruption [17], the mode of action of staurosporine is less clear. Staurosporine is a very potent inhibitor of a broad range of protein kinases [19]. Being one of those targets, inhibition of LIM kinases by staurosporine has been suggested to over-activate actin depolymerizing factor (ADF)/ cofilin resulting in actin modification [18]. Importantly, both drugs were reported to disrupt the actin cytoskeleton without affecting microtubules or intermediate filaments [20,21].

Mechanical forces are well known factors influencing the composition and structure of cartilage [22], and the aforemen- tioned drug treatment and cultivation protocols to regulate chondrocyte behavior both interfere with mechanobiological aspects of the cells. The actin structure is directly linked to focal adhesions that contain a multitude of proteins such as protein kinase C (PKC) or phosphoinositide-3-kinase (PI3K), which show enzymatic activity in addition to being structural components of these complexes [23]. While changes in the cytoskeleton directly feed into the regulation of specific pathways that control re- differentiation, the underlying mechanisms are far from well understood. Yet, certain signaling molecules such as

PKC have been described to be involved in regulating chondrogenesis across a variety of cell types and cultivation protocols [24–26].
Here we asked the question whether the activation/deactivation of just one specific set of key signaling molecules can control the expression of the chondrogenic phenotype, and which pathways might be involved. To understand the relationship of actin cytoskeletal conformation and expression of chondrogenic mar- kers, we elucidated key signaling molecules involved in staurosporine-induced re-expression of the chondrogenic pheno- type. Bovine articular chondrocytes were serially passaged to achieve a de-differentiated phenotype and were then cultivated in medium supplemented with low concentrations of the pharma- cological agents cytochalasin D and staurosporine. Drug treat- ment resulted in actin disruption and induced re-expression of in agarose [8], alginate [9,10], high-density pellets [11,12], or on a hydrogel surface [7]. Treatment of fibroblastic chondrocytes with pharmacological agents (e.g. dihydrocytochalasin B (DHCB) and staurosporine) that interfere with the integrity of the actin cytoskeleton also resulted in re-expression of the chondrogenic phenotype [13–16].

Results

Col2 re-expression is dependent on disruption of the actin cytoskeleton in fibroblastic chondrocytes

To evaluate the influence of drug-induced disruption of the actin cytoskeleton on the process of re-differentiation, cytochalasin D and staurosporine were selected, as they affect either the actin cytoskeleton only or additionally a broad range of kinases, respectively. De-differentiated fibroblastic chondrocytes were cultivated on TCP in control medium or medium supplemented with increasing concentrations of these drugs and stained against actin and col2. De-differentiation was confirmed after serially passaging the chondrocytes, with cells acquiring a fibroblast like morphology already at passage 1 and being devoid of chondro- genic markers by passage 3 (Fig. S1).

Cultivation of fibroblastic chondrocytes in medium supplemen- ted with 0, 2.5, 5 or 10 nM staurosporine (Fig. 1A), showed minor changes of the actin cytoskeleton with only very few cells expressing col2 at 2.5 nM staurosporine. At 5 nM and 10 nM staurosporine, the cells showed a more compacted morphology with a severely compromised actin cytoskeleton. At both con- centrations, many cells showed high expression levels of col2, with the effect being more evident for 5 nM staurosporine. Gene expression analysis of cells treated with 5 nM staurosporine showed a 5.1-fold induction of SOX9 by day 1, which remained 5.6 to 9.9-fold higher in the subsequent 3 days (Fig. 1B). Col2 expression was increased 5-fold after 24 h and remained 123 to 533-fold higher in the following 3 days of cultivation. Col1 expression remained largely unaffected over the 4 days of cultivation showing only 1.2 to 1.6-fold higher expression com- pared to controls at the respective days.

Interestingly, col2 was predominantly found inside the cells upon treatment with staurosporine showing progressive extracellular loca- lization only after removal of the drug (Fig. 1C). Furthermore, staurosporine induced re-differentiation and the associated decrease in cell proliferation (Fig. S2) was found to be only of transient nature as after removal of the drug and a further subcultivation step, proliferation rate appeared normal again and only few cells showed col2 expression (data not shown).

When cytochalasin D was added to cells treated at final concentrations of 0.125, 0.25 or 0.5 mM and compared to control (Fig. 2A), chondrocytes also showed an increasingly disrupted actin cytoskeleton with increasing drug concentrations. The cells were devoid of col2 in control medium and expression levels were hardly detectable at 0.125 mM. Supplemented with 0.25 mM or 0.5 mM cytochalasin D, the majority of cells were positive for col2, being more prominent in cultures treated with 0.25 mM than with 0.5 mM cytochalasin D. Gene expression analysis of chon- drogenic markers during treatment with 0.25 mM cytochalasin D showed a 3.1-fold (po0.05) increase in SOX9 expression on day 1 which was down-regulated maximally 3-fold on the following days, as compared to the controls (Fig. 2B). Col2 expression was elevated 1.8 to 2.7-fold (po0.05 for day 2) in cultures treated with 0.25 mM cytochalasin D over the 4 days period in culture (Fig. 2B). Interestingly, expression levels of chondrogenic markers were significantly lower upon treatment with cytochalasin D compared to staurosporine. This might indicate a differential role of these two drugs on distinct kinases that regulate re-differentiation. Therefore, fibroblastic chondrocytes were subjected to stauros- porine analog midostaurin (PKC412), which has a similar, yet more restricted kinase interaction pattern and is less potent [27]. Probably the most intriguing difference between staurosporine and midostaurin is the lack of binding activity of latter drug to LIM domain kinase (LIMK) and myosin light chain kinase (MLCK), which play a role in actin cytoskeleton integrity and cell con- tractility, respectively. Interestingly, upon supplementation with midostaurin, the cells also showed a severely disrupted cell
morphology, but in contrast to staurosporine treatment, only very few cells expressed col2 at 2 mM and 4 mM midostaurin (Fig. 2C).

Treatment with staurosporine but not with cytochalasin D induces GAG synthesis

To assess the ability of cytochalasin D and staurosporine to stimulate expression of the full chondrogenic phenotype, fibro- blastic chondrocytes were subjected to increasing drug concen- trations and stained with alcian blue for GAG synthesis. Upon treatment with staurosporine, strong induction of GAG synthesis could be observed at 2.5 nM and higher (Fig. 3A). Assessment of GAG production per total DNA content showed increasing levels up to 10 nM staurosporine, which decreased again at 20 nM (Fig. 3B). Surprisingly, there was no pericellular localization of alcian blue positive GAGs with cytochalasin D treatment (Fig. 3C).

Staurosporine induced re-differentiation is enhanced by the inhibition of myosin II and by cultivation in 3D alginate beads

The drastic changes in the actin cytoskeletal conformation induced by cytochalasin D and staurosporine treatment inher- ently alter cell contractility. It is similarly influenced by 3D hydrogel culture, which disrupts the physical link between the actin stress fibers and the ECM. When fibroblastic chondrocytes are cultivated in 2.5 nM staurosporine, only few cells re-express col2. However, upon addition of 1.25 mM blebbistatin, a drug that binds myosin II in an actin-detached state [28] thus preventing the build-up of contractile forces, col2 expression was strongly enhanced (Fig. 4). When supplemented with 2.5 mM and 5 mM blebbistatin, induction of col2 expression was even stronger, having though a negative effect on cell numbers at the highest concentration. Notably, blebbistatin treatment alone was not sufficient to induce col2 re-expression (data not shown).

To evaluate a possible interplay between 3D culture and staur- osporine treatment, fibroblastic chondrocytes were cultivated in alginate beads in the presence of 0.625–5 nM staurosporine. In alginate culture, col2 levels were barely detectable by day 4 and showed moderate expression by day 14 (Fig. 5A). In contrast, upon addition of 0.625 nM staurosporine cells stained strongly for col2 already at day 4 and expression levels remained high by day 14. At the level of mRNA, compared to TCP, alginate culture resulted in a 3.9- and 3.4-fold induction of col2 expression by days 4 and 14, respectively. Upon addition of 0.625 nM staurosporine, col2 levels were elevated 10.6- and 28-fold at days 4 and 14, respectively (Fig. 5B).

Interestingly, higher drug concentrations showed decreased cell viability (Fig. S3) and did not show this synergistic effect on col2 levels (data not shown).

PI3K, PKC, PKA and p38-MAPK positively regulate staurosporine-induced re-expression of col2

To identify positive regulators in the staurosporine-induced re-differentiation process, specific pathway inhibitors were added to fibroblastic chondrocytes cultivated in medium containing 5 nM staurosporine (Fig. 6A). Assessing the involvement of phosphoinositide-3 kinase (PI3K), the cells were subjected to 10 mM LY294002, with cell numbers being decreased and col2 expression being greatly reduced to only very few cells showing a faint signal (marked with arrows). By increasing the inhibitor concentration to 20 mM, expression of col2 was completely abolished (Fig. 6A). To test for the relative involvement of protein kinase C (PKC), protein kinase A (PKA) or p38-mitogen activated protein kinase and (p38-MAPK), respec- tively, fibroblastic chondrocytes were subjected to PKC-inhibitor bisindolylmaleimide I (GF109203X), to PKA-inhibitor H89 or to p38-MAPK inhibitors SB202190. Treatment with 5 mM GF109203X had a profound negative effect on cell number and lead to marked changes in cell morphology causing a more flattened cell appearance with a much more prominent actin cytoskeleton. Col2 expression was completely abolished in the presence of 5 mM GF109203X and this inhibition could be observed at concentrations as low as 0.625 mM (Fig. 6A). When subjected to 5 mM PKA-inhibitor H89, both cell number and col2 expression were strongly reduced but even at 10 mM H89, some col2 expression could still be observed in single cells (Fig. 6A; col2-expressing cells marked with arrows). Cultivation in the presence of 20 mM p38-MAPK inhibitor SB202190 completely abolished col2 expression while both cell
numbers as well as cell morphology were not affected (Fig. 6A).

ILK and MEK/ERK negatively regulate staurosporine- induced re-expression of col2

To identify negative regulators of staurosporine-induced re-differ- entiation, fibroblastic chondrocytes were cultivated in medium supplemented with 2.5 nM staurosporine and subjected to specific inhibitors (Fig. 6B). The contribution of mitogen activated protein kinase kinase (MEK) was evaluated by adding 10 mM PD98059 to the cells. While this inhibitor had no observable effect on cell number, it greatly enhanced col2 expression. This response was even stronger at higher (20 mM and 40 mM) concentrations (Fig. 6B). Addition of the integrin linked kinase (ILK)-inhibitor QLT0267 at a concentration of 5 mM on the other hand noticeably reduced cell numbers and concomitantly strongly induced col2 expression. Interestingly, increasing the con-
centration to 10 mM did not stimulate col2 expression any further but resulted in drastically reduced cell numbers (Fig. 6B). Noteworthy however, col2 expression levels were higher for cells treated with 2.5 nM staurosporine in combination with 5 mM QLT0267 compared to 5 nM staurosporine alone (Fig. 7). Importantly, irrespective of the concentration, none of inhibitors alone were sufficient to induce re-expression of col2 (data not shown).

To elucidate whether a knockdown of ILK results in a similarly enhanced chondrocyte re-differentiation as was observed with QLT0267, fibroblastic chondrocytes were treated with silencing RNA (siRNA) against ILK alone or in combination with 2.5 nM staurosporine and 5 mM QLT0267 for 5 days (Fig. 7). Confirming the functionality of ILK siRNA, compared to control, expression levels of ILK mRNA were decreased by 34–72% upon treatment with ILK siRNA in all media conditions, but showed only minor variations (mostly upregulation) using mock siRNA (Fig. 7A). Down-regulation of ILK upon siRNA treatment could also be confirmed on the protein level using western blots, on which a 20% and 79% decrease with mock and ILK siRNA, respectively, was seen 2 days after transfection (Fig. 7C). In eminent contrast to the increased col2 expression upon treatment with QLT0267, com- pared to control, col2 levels were slightly reduced when cells were treated with ILK siRNA (Fig. 7B). Importantly however, there was no difference in col2 expression between ILK and mock siRNA treated cells. Furthermore, no synergistic effect with respect to col2 expression could be observed in ILK silenced cells cultivated in 2.5 nM staurosporine with or without QLT0267. Similar trends with albeit lower values were already observed at day 2 following ILK siRNA treatment (data not shown).

Discussion

Chondrocytes lose their characteristic chondrogenic phenotype upon cultivation on 2D substrates. While there is a wealth of knowledge on cues triggering de- and re-differentiation of chondrocytes [8,11,13], the underlying mechanisms still remain ill-defined.

Disrupted actin cytoskeleton conformation induces re-expression of collagen II

In agreement with previous studies [13,15], modification of the actin cytoskeleton with both cytochalasin D and staurosporine resulted in potent re-expression of col2 (Figs. 1 and 2), being much higher at both the mRNA and protein level for the latter drug. Interestingly, both agents displayed the highest expression levels upon treatment with intermediate drug concentrations that caused only a moderate degree of actin cytoskeleton re- arrangement. This “preference” for a modulation rather than a complete disruption of the cytoskeleton has been described pre- viously [14] and is in agreement with recent findings demonstrat- ing that the cortical actin ring is necessary for the re-expression of a chondrogenic phenotype [29]. Actin disruption with rather high doses of cytochalasin B and latrunculin has been shown to result in decreased pericellular matrix assembly and retention of the matrix [30]. Similarly, col2 was largely localized in the cytoplasm upon staurosporine treatment and only after removal of the drug it appeared extracellular (Fig. 1C). Taken together, these results show that actin disruption induces col2 re-expression in fibroblastic chondrocytes and that modification rather than complete disrup- tion of the actin cytoskeleton is preferential for eliciting this response.

Re-expression of the full chondrogenic phenotype is induced by staurosporine

The chondrogenic phenotype required for proper function in vivo is marked not only by the expression of markers such as col2, but more so by the synthesis of GAGs. Staurosporine has been shown to strongly induce the synthesis of both, col2 and GAGs [15] (Figs. 1 and 3), furthermore suggesting that under the conditions chosen in this study, probably only col2 secretion but not matrix assembly in general was impaired. Interestingly, there was no pericellular ring of alcian blue positive GAGs after cytochalasin D treatment (Fig. 3C). This is in contrast to previous studies, in which GAG synthesis was shown to increase upon treatment with cytochalasins B [13,31] and D [24,32]. In these previous studies however, a variety of cell types had been used and grown to confluence, which increases the number of cell–cell contacts and also changes paracrine signaling, inherently affecting a range of cell responses [33–35]. Our results indicate that modification of the actin cytoskeleton alone is not sufficient to induce expression of the full chondrogenic phenotype and that mechanisms or signaling pathways that regulate this progression are activated by staurosporine but are not sufficiently influenced by cytocha- lasin D. The limited degree of re-differentiation observed when using midostaurin, which has a restricted kinase interaction pattern compared to staurosporine, furthermore indicates that a very specific set of signaling molecules regulates expression of the chondrogenic phenotype.

Re-differentiation is regulated by cell contractility and synergistically induced by staurosporine treatment in 3D alginate culture

With the contractile actin cytoskeleton playing a major role in the regulation of the chondrogenic phenotype, its disruption inher- ently affects cell contractility. Inhibition of Rho dependent kinase (ROCK), a regulator of contractility and actin bundle formation, has been shown to induce expression of chondrogenic markers in primary chondrocytes [36] and to prevent their de-differentiation [37]. Contractile forces generated by chondrocytes have been demonstrated to increase during in vitro culture expansion and decrease upon application of staurosporine [38]. Furthermore, Lee et al. [16] demonstrated that chondrocytes cultivated in 3D collagen–GAG (CG) scaffolds exhibit higher GAG synthesis and decreased scaffold contraction upon staurosporine treatment. In agreement, we could show that when interfering with cell contractility via inhibition of myosin II, staurosporine-induced re- expression of col2 was strongly enhanced (Fig. 4). Importantly however, treatment of chondrocytes with blebbistatin alone was not sufficient to induce re-expression of the chondrogenic phe- notype, indicating that contractility acts as a secondary factor during re-differentiation.

Cultivation of chondrocytes in alginate beads, which affects both the actin cytoskeleton and cell contractility, is known to induce re-differentiation [9] and could also be observed in this study (Fig. 5). Much like the recently described synergistic effect of ECM stiffness and TGF-β administration on the expression of the chondrogenic phenotype [39], re-differentiation was greatly enhanced upon staurosporine addition to cells in alginate beads. Interestingly, the synergistic effect seen in our study could only be observed at relatively low levels (0.625 nM) of staurosporine (Fig. 5). This suggests that a soft and 3D environment not only induces re-differentiation but also renders the cells susceptible to additional stimulatory cues, likely by influencing similar signaling pathways. Contrasting our results however, supplementation of 5 nM staurosporine to chondrocytes grown on 3D-CG scaffolds has been described to strongly induce GAG-synthesis [16], which can probably be explained by differences in the cell state induced by the culture substrates, as cells acquire a more spread mor- phology in CG scaffolds [40] compared to the round shape in alginate cultures.

Re-differentiation is regulated by the PI3K/Akt pathway and influenced by ILK

Several signaling pathways that have previously been described to regulate the chondrogenic phenotype under various conditions were analyzed for their involvement in staurosporine-induced re-differentiation with the findings being discussed below and summarized in Fig. 8.
The PI3K/Akt pathway has been shown to be regulated by integrin and focal adhesion kinase (FAK) activation [41,42], as well as by changes in the actin cytoskeleton [43]. It is furthermore crucial for insulin-like growth factor-1 (IGF-1) induced chondro- genesis of mesenchymal cells [44,45], as well as for IGF-1 induced proteoglycan synthesis in human chondrocytes [25,46]. In agreement, we could show here that when PI3K was inhibited in staurospor- ine treated fibroblastic chondrocytes, re-expression of col2 was completely abolished. PI3K is furthermore an important regulator of ILK [47], which, localized to focal adhesions, has a dual function as a scaffold and signaling protein [48–50]. However, the kinase activity of ILK has recently been questioned [51], which is supported by results demonstrating that phosphorylation of Akt is unaffected in chondrocytes of ILK deficient mice [52]. ILK deficiency has been shown previously to result in a decrease in cell proliferation and cell spreading, but col2 levels were unaffected in vivo [52]. The results in this study similarly showed reduced cell numbers and spreading upon ILK-inhibition but in presence of staurosporine, col2 expression was enhanced (Fig. 6B). When using siRNA against ILK however, no such synergistic effect on col2 expression could be observed. Conversely, siRNA treatment resulted in slightly decreased col2 levels when compared to control (Fig. 7). While siRNA treatment itself [53] can be the cause for such a decrease it could potentially also be explained by the fact that ILK siRNA also affects other components of focal adhesions including ILK-binding partners PINCH and parvin [54]. However, col2 expres- sion levels were indifferent for both ILK and mock siRNA and since ILK siRNA treatment resulted in decreased ILK gene expression and protein levels, this indicates that ILK has no active role in the regulation of chondrocyte re-differentiation. These results therefore strongly suggest that QLT0267 exerts its positive effect via a different route not directly involving ILK. It was proposed previously that inhibition of ILK with QLT0267 influences the organization of the actin cytoskeleton via integrin signaling [53]. Therefore, the here described increased col2 expression upon ILK inhibition with QLT0267 indicates that this response occurs indirectly by affecting signaling pathways via changes in cell adhesion and the tension built by the actin cytoskeleton.

Re-differentiation is regulated by PKC, MEK/ERK and p38-MAPK

Another signaling pathway important for the regulation of the chondrogenic phenotype is PKC, which has been described to exert its action via ERK and p38-MAPK – two signaling molecules with opposing roles in chondrogenesis [55]. Using rat articular chondrocytes, PKC expression decreased and phosphorylation of ERK increased during de-differentiation being vice versa during re-differentiation [26]. In agreement with these studies, we could show that inhibition of PKC completely abrogated staurosporine- induced re-expression of col2 (Fig. 6A). It was suggested pre- viously, that PKC expression rather than its activity is critical for the chondrogenic phenotype [26]. While high concentrations of GF109203X likely inhibit also other kinases and not just PKC, decreasing the inhibitor concentration to only 0.625 mM still resulted in a complete loss of col2 re-expression (Fig. 6A), suggesting that both expression and activity of PKC are important. This could also explain the observation that expression of chondrogenic markers is highest at intermediate drug levels as with increasing concentrations of staurosporine, PKC is inhibited resulting in a decrease of the re-differentiation stimulus. Support- ing this, inhibition of PKA, which has been shown to inhibit chondrogenesis by modulating PKC [56], did not completely block but greatly reduced col2 expression in staurosporine treated cells (Fig. 6A).

When the phosphorylation of downstream target p38-MAPK was blocked, re-expression of col2 was abolished (Fig. 6A). Conversely, blocking phosphorylation of ERK via inhibition of MEK enhanced re-expression of col2 (Fig. 6B). This is in good agreement with previous studies that demonstrated the impor- tant role of these signaling molecules in chondrogenesis [55,57] and thus shows a similar importance of the respective pathways during staurosporine-induced re-differentiation.

In conclusion, our results highlight the importance of mechan- obiological factors in regulation of the chondrogenic phenotype, suggesting that the observed drug-induced re-differentiation occurs via the direct link of the actin cytoskeleton with the focal adhesion complex, translating changes in both actin structure and contractility into intracellular signaling events. With intermediate drug concentrations eliciting the strongest response, induction of re-differentiation apparently requires a specific degree of re-arrangement of the actin cytoskeleton for activation of the responsible signaling pathways. Expression of the full chondro- genic phenotype occurred only upon treatment with staurospor- ine, indicating that signaling pathways, which include the PI3K-, PKC-, ERK- and p38-MAPK-pathway, are not or not sufficiently influenced by the action of cytochalasin D. The concerted activa- tion of the very same signaling cascades might also regulate re-differentiation induced by 3D alginate culture as on the one hand, the actin cytoskeleton is drastically altered in 3D alginate compared to 2D TCP [58], and on the other hand substrate stiffness has been identified to directly influence MAPK signaling [59]. The finding, that col2 expression levels are higher in cells treated with 2.5 nM staurosporine in combination with 5 mM QLT0267 compared to 5 nM staurosporine alone further supports the hypothesis, that a concerted activation of specific signaling pathways is necessary as well as sufficient to control re-differentiation.

Implications for cartilage regeneration and tissue engineering

The findings from this study could have important implications for cartilage tissue engineering, as they show the possibility to control re-expression of the chondrogenic phenotype via the regulation of specific signaling pathways by drugs such as staurosporine as well as by tuning the mechanical properties of cell carriers. Optimally designed materials could thus support re- differentiation while helping to lower the drug doses necessary for eliciting a chondrogenic response. Performing such a low dose drug treatment with chondrocytes seeded on a carrier material prior to matrix assisted chondrocyte implantation could greatly improve the later performance of such constructs and could be more readily implemented in the clinical context compared to e.g. gene therapy approaches [60]. However, to control the overall cell response during re-differentiation, a much more detailed under- standing of the respective roles, contributions and interrelations of the individual pathways is required.

Materials and methods

Isolation of articular chondrocytes

Primary bovine articular chondrocytes were isolated from knees of 6 months old calves as described previously [61]. In brief, intact knees were opened under aseptic conditions; cartilage shavings were obtained from the joint surface and placed in isolation medium (DMEM supplemented with 1% antibiotic-antimycotic (15240, Invitrogen, Life Technologies Ltd., Paisley, UK) and 360 mg/ml L-glutamine). The specimens were washed twice with 0.9% sodium chloride supplemented with 100 mg/ml gentamycin, cut into small pieces and transferred into 2% pronase solution (6911, Sigma-Aldrich, St. Louis, MO, USA) for 4 h at 37 1C. The cartilage pieces were washed three times with isolation medium before incubation in 0.03% collagenase (C6885, Sigma-Aldrich, St. Louis, MO, USA) for 6 h at 37 1C with gentle stirring. Undigested matrix was then removed by filtration of the solution through a 120 mM-
followed by a 20 mM-cell strainer sieve before washing the cells three times with isolation medium. Primary chondrocytes were then either directly used for cell experiments or frozen at 1 to 5 × 106 cells/ml in FCS containing 10% DMSO for later use.

De-differentiation of primary chondrocytes

After isolation, freshly prepared or frozen primary articular chon- drocytes were seeded in expansion medium (DMEM supplemented 10% FCS, 1% penicillin–streptomycin–neomycin (PSN) and 50 mg/ml ascorbic acid) at a density of 10,000 cells/cm2. Medium was changed every 2–3 days and at indicated days cells were either fixed with 4% paraformaldehyde (PFA)/0.2% Triton X-100 (TX-100) for 8 min and stored in PBS until staining or total RNA was harvested using TRIzol (Invitrogen, Life Technologies Ltd., Paisley, UK) according to the manufacturers protocol.

Re-differentiation of fibroblastic chondrocytes

Fibroblastic chondrocytes (passage 4) were seeded at a density of 1000 cells/cm2 in medium without inhibitor or at 10,000 cells/ cm2 in medium supplemented with cytochalasin D (Sigma- Aldrich, St. Louis, MO, USA), and staurosporine (Sigma-Aldrich, St. Louis, MO, USA) at indicated concentrations. Alginate cultures were prepared by dispersing 1 × 106 cells/ml in 1.2% alginate (PRONOVA, NovoMatrix, Sandvika, Norway) in 0.15 M sodium chloride. Alginate beads were formed by dripping the mixture slowly into a stirred 102 mM calcium chloride solution for 10 min, washed three times with PBS before being transferred into medium containing well plates. Medium was changed 3 and 72 h after cell seeding. After 5 days of culture, cells were fixed with 4% PFA/0.2% TX-100 and stored in PBS until staining or total RNA was harvested using TRIzol. For alginate samples, cells were released from the gel with an alginate dissolution buffer (0.055 M sodium citrate/0.15 M sodium chloride, pH 6.8) for 20 min prior to isolation of RNA with TRIzol.

Signaling pathway inhibitor studies

Fibroblastic chondrocytes (passage 4) were seeded at a density of 1000 cells/cm2 in control medium or at 10,000 cells/cm2 in medium supplemented with 0.25 mM cytochalasin D, and 2.5 or 5 nM staurosporine, respectively. Media were further supplemen- ted with LY294002, SB202190, PD98059 (all Calbiochem, Merck Millipore, Darmstadt, Germany), QLT0267 (Valocor Therapeutics Inc., Vancouver, British Columbia, Canada), GF109203X (TOCRIS bioscience, Bristol, UK) or blebbistatin (Sigma-Aldrich, St. Louis, MO, USA) at indicated concentrations. Inhibitors were dissolved in DMSO and final concentration of DMSO in medium was kept below 0.1%, a concentration not changing basic metabolic activity and proliferation (data not shown). After 5 days of culture, cells were fixed with 4% PFA/0.2% TX-100 and stored in PBS until staining.

siRNA transfection

Short interfering RNAs (siRNA) against ILK (siILK 5′-AUG GGA CUC UGA ACA AAC ATT-3′) and non-targeting RNA (mock 5′-GAA CGA CGC CGU ACU CAU UTT-3′) were purchased from Microsynth (Balgach, Switzerland). Fibroblastic chondrocytes (passage 4) were seeded at a density of 30,000 cells/24-well in control medium. After 24 h cells were transfected with siRNA (25 nM final concentration) using the INTERFERINs transfection reagent (Polyplus) according to the manufacturer’s instructions. The cells were cultivated in indicated media compositions and were analyzed by RT-PCR after 5 days. siRNA knockdown (43718.5%) was calculated by normalizing the Cq values for ILK siRNA to Cq values of non-targeting (mock) siRNA.

For Western Blot analysis, cells were washed with PBS and lysed in RIPA buffer (20 mM Tris/HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% IGEPAL CA-630, 0.25% Deoxycholate, 0.1% SDS, 100 mM NaF). The protein concentration was determined by Pierce™ BCA protein assay kit (Thermo Scientific, 23225). 5 mg of protein was resolved on a 16% SDS-Page and transferred to a nitrocellulose membrane (iBlotR Transfer Stack, life technologies). The mem-
brane was blocked in 5% BSA–PBST-buffer (PBS without glucose with 0.1% Tween20). ILK was detected using primary antibody anti-Integrin linked ILK–IgG antibody [EP1593Y], 1:5000 (Abcam, ab52480) and secondary antibody Goat-anti-rabbit-IgC-HRP, 1:10000 (GE healthcare life science, RPN 4301). GAPDH, used as loading control, was detected by anti-GAPDH-IgC-HRP, 1:2000 (Abcam, ab9385). Proteins were visualized with an enhanced chemiluminescence detection reagent (Amersham ECL Detection Kit, GE Healthcare Life Sciences, RPN2135) and imaged with ChemiDoc-Its Imager (UVP) and VisonWorksLS Image Acquisition software (UVP). The protein ratios were quantified using the ImageJ software.

Quantitative RT-PCR

To investigate gene expression levels of SOX9 and col2 under the different conditions, total RNA was isolated from cultivated chondrocytes and RNA concentration was measured with a spectrophotometer (Nanodrop, Thermo Scientific, Waltham, MA, USA). cDNA was synthesized (protocol: 5 min at 25 1C, 30 min at 42 1C, and 5 min at 85 1C) from 500 ng total RNA using the iScripts reverse transcription system (BioRad, Hemel Hamstead, UK) on a thermal cycler (iCycler iQTM Real Time Detectino System, BioRad). The following primer pairs (purchased from Microsynth, Balgach, Switzerland) were used for amplification of chondrocyte specific markers SOX9, col2, col1 and normalizing standard GAPDH or RPL13. Col2_a: 5′ GGC CAG CGT CCC CAA GAA 3′ and Col2_b: 5′ AGC AGG CGC AGG AAG GTC AT 3′, SOX9_a: 5′ ACG CGG CCC CAG GAG AAC 3′ and SOX9_b: 5′ CGG ATG CAC ACG GGG AAC TT 3′, Col1_a: 5′ CAG CCG CTT CAC CTA CAG C 3′, Col1_b: 5′ TTT TGT ATT CAA TCA CTG TCT TGC C 3′, GAPDH_a: 5′ AGG CCA TCA CCA TCT TCC 3′ and GAPDH_b: 5′ TTC ACG CCC ATC ACA AAC 3′, RPL13_a: 5′ GCC AAG ATC CAC TAT CGG AAA 3′, RPL13_b: 5′ AGG ACC TCT GTG AAT TTG CC 3′, ILK_a: 5′ GAC ATG ATC GTGCCT ATC CTG 3′, ILK_b: 5′ GTC CTG ACA CCT CTG AAG TTC 3′. cDNA was diluted 1:5 prior to amplification of mRNA using the iQTM SYBRs Green system (BioRad). Reactions were performed in triplicate consisting of 5 ml diluted cDNA, 25 ml iQ SYBR Supermix, 0.1 ml of each primer (100 mM) and DNase-free water up to a final volume of 50 ml. RT-PCR reactions for siRNA experiments were performed in triplicate consisting of 5 ml diluted cDNA (synthe- sized from 200 ng total RNA), 6 ml iQ SYBR Supermix, 0.24 ml of each primer (10 mM) and DNase-free water up to a final volume of 12 ml. Amplification (protocol: 4 min activation at 95 1C followed by 40 cycles of 30 s denaturation at 95 1C, 30 s annealing and elongation at 60 1C and final cooling to 4 1C) was monitored and analyzed with the iCycler Software (BioRad). Expression levels of chondrocyte specific genes were calculated using the 2—ΔΔCT method [62] and were normalized to GAPDH (Figs. 1 and 2) or RPL13 (Fig. 5). Triplicate measurements were performed with indicated experimental replications.

Immunohistochemistry

For immunohistochemical stainings of cells cultivated under different conditions in 2D, antibodies/dyes were diluted in 1.5% skim milk/PBS and incubated at RT for 1 h. Unspecific binding was blocked by 1 h incubation with 5% goat serum and 1% fetal calf serum (FCS) in PBS. Col2 was stained using a monoclonal mouse anti-collagen II (ascites, 1:1000, cat.no II-II6B3 ascites, Develop- mental Studies Hybridoma Bank, DSHB, Iowa City, IA, USA)
followed by a goat α-mouse IgG Alexa Fluor 488 (1:400, A11029, Molecular Probes, Life Technologies Ltd., Paisley, UK) antibody. Actin and the nuclei were stained using Alexa546 conjugated phalloidin (1:40, B607, Molecular Probes, Life Technologies Ltd., Paisley, UK) and DAPI (4,6-diamidino-2-phenylindole, 1:1000, D9542, Sigma-Aldrich, St. Louis, MO, USA), respectively. For cells in 3D alginate beads, unspecific binding was blocked by 1 h incubation in 5% bovine serum albumin (BSA) and 3% FCS in PBS. After digestion in pronase (1 mg/ml) for 15 min, col2 was stained using a monoclonal mouse anti-collagen II (ascites, 1:50, cat.no CIIC1 ascites, Developmental Studies Hybridoma Bank, DSHB, Iowa City, IA, USA) in 1% BSA in PBS for 24 h, followed by a goat α-mouse IgG Alexa Fluor 488 (1:200, A11029, Molecular Probes, Life Technologies Ltd., Paisley, UK) antibody in 1% BSA for 1 h. All solutions contained 10 mM CaCl2. Samples were imaged on a fluorescent microscope (Axio Imager M.1, Carl Zeiss, Oberkochen, Germany) or a confocal laser scanning microscope (LSM 5, Carl Zeiss, Oberkochen, Germany) using filter sets corresponding to the fluorescence of interest.

Alcian blue staining/quantification

Alcian blue staining and quantification were performed as described previously [63]. In brief, fixed cells were stained with 1% alcian blue in 3% acetic acid (pH 2.5) for 30 min. Thereafter, cells were washed thoroughly with ddH2O and imaged by phase contrast microscopy (Nikon Instruments, Melville, NY, USA). The bound dye was solubi- lized in 500 ml DMSO on a gyratory shaker for 2 h and quantified by measuring the absorbance at 630 nM (BioTek ELx800, BioTek Instruments, Winooski, VT, USA). For quantification of DNA content, a Hoechst 33258 (Sigma-Aldrich, St. Louis, MO, USA) assay [64] was employed. In brief, cells were washed with PBS and stained with 500 ml of a 10 mg/ml Hoechst solution in TNE-buffer (10 mM Tris,1 mM EDTA, and 2 M sodium chloride) for 1 h before measuring fluorescence of the adherent cells at 460 nM.

Live/dead viability assay

Fibroblastic chondrocytes were cultivated for 3 days in control medium or medium supplemented with staurosporine at indi- cated concentrations before staining with the live/dead viability kit (L3224, Invitrogen, Life Technologies Ltd., Paisley, UK) accord- ing to the manufacturer’s instructions with minor modifications. In brief, medium was removed and cells were cultivated in fresh medium supplemented with 1.3 mM ethidium homodimer-1 and
2.6 mM calcein AM for 10 min at 37 1C. Subsequently, cells were imaged within 2–3 min. Control samples of dead cells were produced by treatment of cells with 0.2% digitonin in PBS for 5 min prior to incubation with dyes.

MTT- and DNA-assays

Cellular activity was assessed by a 3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyltetrazolium bromide (MTT) assay as it has been described previously [65]. In brief, cells were washed once with plain DMEM before adding 125 ml of an MTT solution (5 mg/ml) to each well for 1 h at 37 1C. The solution was then removed and intracellular MTT–formazan crystals were dissolved in 90% etha- nol for 10 min. Absorbance was measured at 550 nM and wells
without cells were used as blanks. For measurement of total DNA content as an index of cell proliferation a Hoechst 33258 assay was employed. In brief, medium was removed and cells were incubated with 250 ml ddH2O per well at 37 1C for 1 h before adding 250 ml/well of a 20 mg/ml Hoechst solution in TNE-buffer for 1 h. Fluorescence was measured at 460 nM, wells without cells were used as blanks and serial dilutions of calf thymus DNA (Sigma, D-3664) were used as standard.

Statistical analysis

Statistical analysis was performed using Student’s t-test. Results shown are triplicate measurements (mean7SD) obtained from three independent experiments. Differences were considered statistically significant at p-values below 0.05 (np40.05, nnp40.01).