Curcumin mediated suppression of nuclear factor-κB promotes chondrogenic differentiation of mesenchymal stem cells in a high-density co-culture microenvironment

Polyclonal anti-collagen type II antibody (PAB746), monoclonal anti-adult cartilage-specific proteoglycan antibody (MAB2015), anti-β1-integrin antibody (MAB1965), and alkaline phosphatase linked sheep anti-mouse (AP303A) and sheep anti-rabbit secondary antibodies (AP304A) for immunoblotting and immuno-electron labelling were purchased from Chemicon International, Inc. (Temecula, CA, USA). Monoclonal anti-β-Actin (A4700) was purchased from Sigma, St. Louis, MO, USA). Monoclonal anti-Sox-9 was purchased from Acris Antibodies GmbH, Hiddenhausen, Germany. Monoclonal anti-phospho-p42/p44 ERK1/2 antibody (610032) and polyclonal anti-Shc antibody (610082) were purchased from BD (BD Biosciences, Erembodegem, Belgium). Polyclonal anti-active caspase-3 (AF835) was purchased from R&D Systems (Abingdon, UK). Antibodies against phospho-specific IκBα (Ser 32/36) and against anti-phospho-specific p65(NF-κB)/(Ser536) were obtained from Cell Technology (Beverly, MA, USA). Curcumin with a purity > 95% was purchased from Indsaff (Punjab, India). This commercial source of curcumin contains three major components: Diferuloylmethane (the most abundant and active component of turmeric) (82%) and its derivatives demethoxycurcumin (15%) and bisdemethoxycurcumin (3%), together referred to as curcuminoids [9, 11‐16]. Curcumin was dissolved in dimethylsulfoxide (DMSO) as a stock concentration of 500 μM and stored at -80°C. Serial dilutions were prepared in culture medium.

Mesenchymal stem cells (MSCs) were isolated from canine adipose tissue biopsies and primary canine chondrocytes were isolated from cartilage from the femoral head. Samples were obtained during total hip replacement surgery with fully-informed owner consent and ethical project approval from the ethical review committee of the Ludwig-Maximilians-University, Munich, Germany. Chondrocytes and MSCs used in co-culture experiments were always from the same animal. In total, the experiments were performed three times and samples from three different donors were used. Donor ages ranged from five to seven years.

Briefly, for MSCs isolation, adipose tissue was cut into small pieces and digested with collagenase 0.2% in Ham's-F12 in a water bath at 37°C for two hours. Digested adipose tissue was centrifuged at 1,000 g for five minutes and the pellet was resuspended in cell culture medium consisting of DMEM/Ham's-F12 1:1, 10% FCS, 1% partricin solution, 1% penicillin/streptomycin solution (10 000 IU/10 000 IU), 75 μg/ml ascorbic acid, 1% essential amino acids and 1% Glutamine, all obtained from Seromed (Munich, Germany) in a T75 cell culture flask and incubated at 37°C/5%CO₂, 95% humidity. After four days, non-adherent cells were discarded by washing with Hank's salt solution. The medium was changed three times per week. Adherent cells were split following formation of fibroblast-like cell colonies and upon reaching 60 to 70% confluence, and were sub-cultured until the third or fourth passage was achieved. As there are no definitive MSC specific cellular markers, we identified them by their ability to adhere to tissue culture plastic in vitro, through their multilineage differentiation potential in vitro and through a combination of expression and lack of defined markers (CD105⁺, CD90⁺, CD45^-, CD34^-) [19‐22].

For chondrocyte isolation the cartilage sample was sliced into 1 to 2 mm thick slices and incubated first with pronase (2%/Hams-F12) (Roche Diagnostics, Mannheim, Germany), followed by collagenase incubation (0.2%/Ham's-F12) (Sigma) in a shaking water bath at 37°C. The digested sample was centrifuged at 1,000 g for five minutes and cells plated at 1 × 10⁶ cells per T75 flask at 37°C/5%CO₂. The first medium change was performed after 24 hours, and the following medium changes three times per week. Chondrocytes were split at approximately 70% confluency and passaged twice.

Three-dimensional high-density cultures at the air-liquid interface were prepared as previously described [23]. Cells were centrifuged at 1,000 g for five minutes and around one million cells (approximately 8 μl) from the cell pellet were pipetted directly onto a nitrocellulose filter on a steel grid. This model allows the cells to aggregate, forming a distinct pellet, which was examined after 14 days.

High-density culture pellets either consisted only of MSCs or primary chondrocytes, or a mixture of MSCs and primary chondrocytes (1:1) (co-culture). In all experiments, cultures and co-cultures were either incubated with cell culture medium (10% FCS) or with a chondrogenic induction medium as described by Pittenger et al. [24] consisting of DMEM base medium, D-(+)-glucose 0.35 g/100 ml (Sigma, Cat No. G7021), ITS+ 1 liquid media supplement (10 μg/ml insulin, 5.5 μg/ml transferrin, 5 ng/ml selenium, 0.5 mg/ml bovine albumin, 4.7 μg/ml linoleic acid (Sigma, Cat. No. I-2521), 0.1 mM ascorbate-2-phosphate (Sigma, Cat. No. A-8960), 10^-7 M dexamethasone (Sigma, Cat. No. D-8893), penicillin/streptomycin solution (10,000 IU/10,000 IU/100 ml). 10 ng/ml hTGFβ1 (Acris Antibodies GmbH, Hiddenhausen, Germany) were added fresh to the medium before each medium change. Furthermore, some cultures and co-cultures were then incubated with one of the following treatments: curcumin only; pre-stimulated with curcumin for four hours in suspension and then transferred to high-density culture; 10 ng/ml IL-1β; curcumin and IL-1β; pre-stimulated with curcumin for four hours in suspension and then brought into high-density culture and stimulated with IL-1β; or pre-stimulated with curcumin for four hours in suspension and then brought into high-density culture and stimulated with IL-1β and curcumin. Medium changes were made every three days.

To examine in more detail the interaction between curcumin and IL-1β in MSCs and the pathological pathways involved, monolayer cultures of MSCs were evaluated. First, MSCs were cultured with various concentrations of curcumin (0, 0.5, 1, 2 and 5 μM) for four hours, followed by 24 hours 10 ng/ml IL-1β stimulation. Second, MSC cultures were pre-treated for four hours with 5 μM curcumin followed by one hour 10 ng/ml IL-1β stimulation. Whole cell lysates, cytoplasmic extracts and nuclear extracts were taken at various time points and evaluated with Western blotting.

Transmission electron microscopy was performed as previously described [25]. High-density co-cultures were fixed for one hour in Karnovsky-fixative fixative and post-fixed in 1% OsO₄ solution. After dehydration, pellets were embedded in Epon, ultrathin cuts were made on a Reichert-Ultracut E and contrasted with a mixture of 2% uranyl acetate/lead citrate. A transmission electron microscope (TEM 10, Zeiss, Jena, Germany) was used to examine the co-cultures. To quantify apoptosis, the number of cells exhibiting typical morphological features of apoptotic cell death was determined by scoring 100 cells from 30 different microscopic fields per culture and the number of apoptotic cells expressed as an indicator of MSC culture degradation.

For Western blotting, total cell proteins were either extracted from the cell cultures with lysis buffer on ice for 30 minutes or nuclear extracts and cytoplasmic extracts prepared as previously described [9]. Total protein content was measured with the bicinchoninic acid system (Uptima, France) using bovine serum albumin as a standard. Samples were further reduced with 2-mercaptoethanol and total protein concentrations adjusted. Proteins (500 ng per lane total protein) were separated with SDS-PAGE under reducing conditions on 5%, 7%, 10% or 12% gels and then blotted onto nitrocellulose membranes using a trans blot apparatus (Bio-Rad, Munich, Germany). After blocking for two hours in 5% (w/v) skimmed milk powder in phosphate buffered saline (PBS)/0.1% Tween 20, membranes were incubated with the primary antibodies (overnight, 4°C), followed by incubation with the alkaline phosphatase conjugated secondary antibodies for two hours at room temperature. Finally, specific antigen-antibody complexes were detected using nitroblue tetrazolium and 5-bromo-4-chloro-3-indoylphosphate (p-toluidine salt; Pierce, Rockford, IL, USA) as substrates for alkaline phosphatase. Specific binding was quantified by densitometry using "Quantity One" (Bio-Rad Laboratories Inc. Munich, Germany).

In order to demonstrate that the cells isolated from the canine adipose tissue (Figure 1A) are indeed mesenchymal stem cells (MSCs) we differentiated them to adipocytes, osteoblasts and chondrocytes (Figure 1B, C). After three weeks' treatment with the adipogenic induction medium, the cells contained abundant amounts of vacuoles and Oil Red O staining for fat revealed that these vacuoles contained neutral lipids (B). After three weeks culture time with the osteogenic induction medium the cells changed to a more polygonal appearance, formed nodules and were stained positive with von Kossa stain for mineral deposition (C). Alcian blue staining after 14 days in culture revealed a high content of cartilage specific proteoglycans in induced cultures (D). Additionally, the isolated MSCs showed a strong positive signal for the stem cell specific markers CD90⁺ and CD105⁺ (Figure 1E, F). In contrast to this they were clearly labelled negative for the hematopoietic stem cell markers CD45^- and CD34^- (Figure 1G, H).

To study the effects of curcumin on MSCs after 14 days of cultivation in three-dimensional high-density culture, ultrastructural evaluations were performed (Figure 2A). In control cultures, MSCs did not survive, but underwent apoptosis or necrosis and mainly cell debris was observed (a). Treatment of the cultures with the chondrogenic induction medium stimulated chondrogenesis (b). The formation of cartilage nodules consisting of large rounded viable cells (containing large quantities of endoplasmic reticulum, mitochondria and other cellular organelles) embedded in a fine structured, highly organised ECM was observed. Treatment of MSCs with curcumin alone did not stimulate chondrogenesis and, as in control cultures, mainly cell debris was observed (c, e). It did not make a difference whether cultures were either pre-treated for four hours with curcumin (c) or pre-treated four hours with curcumin followed by treatment with curcumin over the entire culture period (e). In contrast, in cultures treated with curcumin and the chondrogenic induction medium (d, f) chondrogenesis was observed. Chondrogenesis was similar in cultures that were either pre-treated for four hours with curcumin (d) or cultures that were pre-treated four hours with curcumin followed by treatment with curcumin over the entire culture period (f).

Western blotting was performed to confirm these results (Figure 2B). Whole cell lysates were resolved by SDS-PAGE electrophoresis and blotted onto nitrocellulose. The membranes were probed with antibodies against cartilage specific proteoglycans (CSPG), collagen type II, β1-integrins, Shc, activated extracellular regulated kinases 1 and 2 (ERK 1/2) and Sox-9.

In agreement with the ultrastructural findings, MSC cultures treated with the specific chondrogenic induction medium alone or in combination with curcumin produced high amounts of CSPGs, collagen type II and β1-integrin. Here, four-hour curcumin pre-treatment followed by treatment with the chondrogenic induction medium was as effective in inducing the production of CSPGs, collagen type II and β1-integrin as a four-hour curcumin pre-treatment followed by treatment with curcumin and the chondrogenic induction medium over the entire culture period. Further, in these cultures the chondrogenic signalling cascade was activated with high expression of Shc, activated ERK 1/2 and Sox-9. In contrast to this, cartilage specific matrix components and members of the chondrogenic signalling cascade, were not expressed in untreated cultures or cultures incubated only with curcumin. Again, it did not make a difference whether cultures were pre-treated for four hours with curcumin or pre-treated four hours with curcumin followed by treatment with curcumin over the entire culture period.

It has been reported that IL-1β inhibits chondrocyte proliferation and induces apoptosis [26, 27]. We therefore evaluated the effects of curcumin on MSCs stimulated with IL-1β and/or the chondrogenic induction medium.

Ultrastructural evaluation revealed that stimulation of MSCs with IL-1β either alone (Figure 3A-a), in combination with the chondrogenic induction medium (b) or in combination with four-hour curcumin pre-treatment (c) resulted in apoptosis. However, treatment with IL-1β, curcumin and the chondrogenic induction medium lead to induction of chondrogenesis (d, e) with formation of cartilage nodules containing viable, rounded cells that were embedded in a cartilage specific matrix. In these cultures, a four-hour curcumin pre-treatment (d) was as effective in inhibiting IL-1β induced apoptosis as a four-hour curcumin pre-treatment followed by treatment with curcumin over the entire culture period (e).

Apoptosis was further quantified as described in Materials and Methods. The data shown in Figure 3B demonstrate a significantly increased number of apoptotic cells in cultures stimulated with IL-1β either alone, in combination with the chondrogenic induction medium or in combination with a four-hour curcumin pre-treatment. As shown at the ultrastructural level, the number of apoptotic cells significantly decreased in MSC cultures treated with IL-1β, curcumin and the chondrogenic induction medium (Figure 3B). In these cultures the number of apoptotic cells was similar between cultures that were either pre-treated for four hours with curcumin or pre-treated four hours with curcumin followed by curcumin treatment over the entire culture period.

To support these ultrastructural findings, Western blotting was performed by probing whole cell lysates with antibodies against CSPGs, collagen type II, β1-integrin, Shc, activated ERK 1/2 and Sox-9 (Figure 3C). Stimulation of MSC cultures with IL-1β either alone, in combination with the chondrogenic induction medium or with a four-hour curcumin pre-treatment did not lead to production of CSPGs, collagen type II, β1-integrin and activation of Shc, ERK 1/2 and Sox-9. In contrast, production of CSPGs, collagen type II and β1-integrin was upregulated and Shc, activated ERK 1/2 and the chondrogenic specific transcription factor Sox-9 was highly expressed in MSC cultures treated with IL-1β, curcumin and the chondrogenic induction medium. Underlining the ultrastructural findings, it did not make a difference whether cultures were either pre-treated for four hours with curcumin or pre-treated four hours with curcumin followed by curcumin treatment over the entire culture period.

Further, to demonstrate the influence of curcumin on the induction of the apoptotic signalling cascade by IL-1β in MSCs, cultures were evaluated for activated caspase-3 and the marker of inflammation and prostaglandin production COX-2 (Figure 3D). Production of activated caspase-3 and COX-2 expression was prominent in all MSC cultures stimulated with IL-1β alone and was blocked in all cultures treated with IL-1β and curcumin. Here, a four-hour pre-treatment with curcumin was as effective in inhibiting the IL-1β induced apoptotic signalling cascade as a four-hour pre-treatment with curcumin followed by curcumin treatment over the entire culture period.

As demonstrated above, MSC cultures treated with a chondro-inductive medium undergo chondrogenesis despite the presence of IL-1β if the apoptotic and inflammatory cascades induced by IL-1β are inhibited by curcumin. Next we evaluated whether the same effect can be observed in a co-culture model of MSCs and primary chondrocytes.

Ultrastructural evaluation demonstrated cellular debris in untreated MSC cultures (Figure 4A-a), and development of cartilage nodules with rounded, viable cells embedded in a highly organised, fine structured ECM in untreated primary chondrocyte cultures (b) and in untreated co-cultures (c). Stimulation of co-cultures with IL-1β resulted in cellular degradation (d). Curcumin treatment of the co-cultures, either as a four-hour pre-treatment (e) or over the entire culture period (f) did not impede chondrogenesis. Combinational treatment of co-cultures with IL-1β and curcumin suppressed the adverse effects of IL-1β on chondrogenesis (g, h). Here, inhibition of IL-1β induced apoptosis was as effectively blocked by a four-hour curcumin treatment (g) as by curcumin treatment for the entire culture period (h).

Western blotting using antibodies against CSPGs, collagen type II, β1-integrin, Shc, activated ERK 1/2 and Sox-9 was performed to evaluate induction of chondrogenesis in the co-cultures on a molecular level (Figure 4B, C). Production of cartilage matrix specific markers and of chondrogenic signalling pathway members was slight in untreated MSC cultures and high in untreated primary chondrocyte cultures and untreated co-cultures (Figure 4B, C). Stimulation of co-cultures with IL-1β alone inhibited production and expression of CSPGs, collagen type II, β1-integrin, Shc, activated ERK 1/2 and Sox-9. In contrast, co-cultures stimulated with IL-1β and curcumin produced high levels of chondrogenic matrix specific markers (Figure 4B) and chondrogenic signalling pathway proteins (Figure 4C). Stimulation of chondrogenesis was similar between cultures that were either pre-treated for four hours with curcumin or were pre-treated four hours with curcumin followed by curcumin treatment over the entire culture period and comparable to untreated primary chondrocyte cultures and untreated co-cultures.

To further demonstrate the influence of curcumin on the induction of the apoptotic signalling cascade by IL-1β in co-cultures, cultures were evaluated for activated caspase-3 and the marker of inflammation and prostaglandin production COX-2 (Figure 4D). Activated caspase-3 and COX-2 were highly expressed in untreated MSC cultures but were not expressed in untreated primary chondrocyte cultures and untreated co-cultures. Treatment of the co-cultures with IL-1β alone led to high production of activated caspase-3 and COX-2. In contrast, in co-cultures stimulated with both IL-1β and curcumin neither activated caspase-3 nor COX-2 was detected. A four-hour pre-treatment of curcumin or curcumin treatment over the entire culture period both effectively inhibited IL-1β induced activation of caspase-3 and COX-2 production. These results confirm the ultrastructural findings described above and demonstrate that curcumin inhibits IL-1β induced apoptotic and inflammatory signalling pathways, promoting co-culture induced chondrogenesis.

Further experiments were carried out to evaluate the interaction between curcumin and IL-1β in MSCs. These experiments demonstrated that curcumin suppressed IL-1β induced activation of apoptotic and inflammatory pathways in a concentration (Figure 5) and time (Figure 6) dependent manner.

As shown in Figure 5A, treatment with as little as 0.5 μM of curcumin over the entire culture period was sufficient to significantly suppress IL-1β induced activation of caspase-3 and COX-2 production in MSCs. Pro-caspase-3 production remained unaffected.

The IL-1β-activated transcription factor nuclear factor-κB (NF-κB) plays an essential role in mediating inflammatory and apoptotic processes in chondrocytes and it is known that in chondrocytes, curcumin is able to suppress NF-κB [9, 14, 28]. To evaluate whether curcumin influences IL-1β-induced NF-κB in MSCs, we investigated the NF-κB signalling pathway. As demonstrated in Figure 5B, curcumin suppressed IL-1β-induced activation of Iκ-Bα in MSCs. This correlated clearly with decreased NF-κB translocation to the nucleus. Inhibition of Iκ-Bα phosphorylation as well as NF-κB translocation to the nucleus was evident when curcumin was included at a concentration of 0.5 μM. Higher concentrations of curcumin completely blocked IL-1β-induced activation of Iκ-Bα and NF-κB translocation to the nucleus (Figure 5B).

Further, pre-treatment of MSC cultures with 5 μM curcumin also inhibited IL-1β-induced activation of caspase-3 and COX-2 expression in a time dependent manner (Figure 6A). After 60 minutes incubation time, activation of caspase-3 and COX-2 production was completely suppressed. Pre-treatment of MSC cultures with 5 μM curcumin also inhibited activation of Iκ-Bα and NF-κB translocation to the nucleus in a time dependent fashion (Figure 6B). In contrast, in IL-1β treated control cultures, activated caspase-3, higher expression of COX-2, activated Iκ-Bα and higher concentrations of NF-κB in the nucleus were observed.