Establishing SW1353 Chondrocytes as a Cellular Model of Chondrolysis

Osteoarthritis (OA) is the most common degenerative joint disease characterised by chondrocyte cell death. An in vitro model of chondrocyte cell death may facilitate drug discovery in OA management. In this study, the cytotoxicity and mode of cell death of SW1353 chondrocytes treated with 24 h of OA inducers, including interleukin-1β (IL-1β), hydrogen peroxide (H2O2) and monosodium iodoacetate (MIA), were investigated. The microscopic features, oxidative (isoprostane) and inflammatory markers (tumour necrosis factor-alpha; TNF-α) for control and treated cells were compared. Our results showed that 24 h of H2O2 and MIA caused oxidative stress and a concentration-dependent reduction of SW1353 cell viability without TNF-α level upregulation. H2O2 primarily induced chondrocyte apoptosis with the detection of blebbing formation, cell shrinkage and cellular debris. MIA induced S-phase arrest on chondrocytes with a reduced number of attached cells but without significant cell death. On the other hand, 24 h of IL-1β did not affect the cell morphology and viability of SW1353 cells, with a significant increase in intracellular TNF-α levels without inducing oxidative stress. In conclusion, each OA inducer exerts differential effects on SW1353 chondrocyte cell fate. IL-1β is suitable in the inflammatory study but not for chondrocyte cell death. H2O2 and MIA are suitable for inducing chondrocyte cell death and growth arrest, respectively.


Introduction
Osteoarthritis (OA) is a degenerative joint disease commonly found among the elderly [1]. The exact aetiology of OA has been unclear until now [2]. Several risk factors like age (>50 years), gender (female), obesity, diabetes, past joint trauma, strenuous exercise, and genetic susceptibility could increase the risk of OA [2,3]. Currently, there is no effective treatment that could delay, stop or reverse the cartilage destruction due to an incomplete understanding of the molecular mechanisms of OA initiation and progression [2,4]. The hypocellularity of articular cartilage due to chondrocyte cell death is one of the hallmarks of OA initiation [5]. In the early stage of OA, chondrocytes are damaged by mechanical stress, oxidative stress or inflammation [6]. They have limited capacity to recover from these assaults due to the avascular and alymphatic nature of articular cartilage [7,8]. The damaged chondrocytes may experience hypertrophy-like changes and lose the ability to synthesise proteoglycan [9]. Eventually, the cells will die due to an impaired reparative Life 2021, 11,272 3 of 17 tochondrial dysfunction [16,43,46]. MIA-mediated oxidative stress and energy depletion ultimately cause the death of primary articular chondrocytes [46] and chondrocyte cell lines [16,17,43]. Intra-articular injection of MIA causes the cytotoxicity in chondrocytes, leading to cartilage degradation and subchondral bone alterations, which mimics the pathogenesis of OA [12,43].
SW1353 chondrocytes are among the most established cell lines in in vitro OA studies. Despite the prevalent use of IL-1β, H 2 O 2 and MIA in SW1353 chondrocyte models, no direct comparison among these inducers of SW1353 cell death has been made. This research gap will impede the selection of an appropriate inducer in in vitro OA studies. So far, only a study by Martin et al. comparing the differential catabolic effects of IL-1β and H 2 O 2 on juvenile bovine chondrocytes is available [47]. The differential effects of these common OA inducers on chondrocyte cell death are largely unknown. Therefore, in this study, we investigated and compared the IL-1β-, H 2 O 2 -and MIA-induced cytotoxicity and mode of cell death in SW1353 chondrocytes. The involvement of oxidative stress and inflammation were also investigated. This is the first study to investigate the differential effects of these OA inducers. Such information would benefit researchers in selecting an appropriate agent for their in vitro OA models.

Materials
MIA was purchased from Sigma-Aldrich (St. Louis, MO, USA), dissolved in dimethyl sulfoxide (DMSO) as a 100 mM stock solution, aliquoted and kept at −80 • C until treatment. Concentrated 30% H 2 O 2 solution was purchased from Merck Millipore (Merck KGaA, Gernsheim, Germany) and kept at 4 • C until treatment. During treatment, the 30% H 2 O 2 solution was pre-diluted with sterile phosphate-buffered saline (PBS) in dark conditions to prepare a 20 mM H 2 O 2 solution, then continued to the second dilution step in culture media to prepare the treatment solution accordingly. Lyophilised recombinant human IL-1β with albumin (catalogue No. 8900SC) was purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA) and reconstituted according to the manufacturer's instructions. Briefly, 10 µg of IL-1β powder was dissolved with sterile PBS as a 50 µg/mL stock solution at room temperature with 30 min of solubilization and occasional gentle vortexing. The dissolved IL-1β was aliquoted and kept at −80 • C until treatment. The stock solutions of IL-1β, H 2 O 2 and MIA were always protected from light during preparation, storage and treatment. Other chemicals were purchased from Sigma-Aldrich unless stated otherwise. C and 5% carbon dioxide conditions. These cells were sub-cultured every 2-3 days to maintain optimal cell growth. The cells within the first 3 to 10 passages after being revived from cryopreservation were used in the experiment.

Cytotoxicity Assay
CellTiter 96 AQueous One solution (Promega, Fitchburg, WI, USA) containing 3-(4,5dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt (MTS) was used to determine the cytotoxicity of IL-1β, H 2 O 2 and MIA on SW1353 chondrocytes as previously described, with modifications [48,49]. Briefly, 100 µL of 5 × 10 4 cells/mL of SW1353 chondrocytes was seeded in a 96-well plate with complete media for 24 h. On the next day, all the media were then replaced with 100 µL of complete media with a series of concentration of either IL-1β, H 2 O 2 and MIA for another 24 h. DMSO (0.05%) was used as vehicle control (VC). At the end of the treatment, 20 µL of MTS reagent was added to each well and the plate was further incubated for another 1 h at 37 • C. The absorbance of each well was then measured at 490 nm using a Multiskan GO microplate reader (Thermo Fisher Scientific, Vantaa, Finland). The viability of cells was calculated as a percentage with the optical density (OD) value of the VC as the denominator. The values of the 25% and 50% maximal inhibitory concentration (IC 25 and IC 50 ) were determined for subsequent experiments.

Annexin V-Fluorescein Isothiocyanate/Propidium Iodide Dual-Labelling Assay
The mode of cell death, either via apoptosis or necrosis, was determined according to the plasma membrane externalization of phosphatidylserine and its integrity [50]. This assay was conducted in accord with a previous study [51]. Briefly, 3 mL of SW1353 cells (5 × 10 4 cells/mL) was seeded in a 6-well plate for 24 h, followed by treatment with H 2 O 2 and MIA at their IC 25 and IC 50 values, or IL-1β at 50 and 100 ng/mL for 24 h. Before collecting the cells, the microscopic images of cells were captured at 200× magnification using an Olympus CKX31 inverted microscope with an X-CAM alpha camera and DigiAcquis 2.0 acquisition software (Matrix Optics, Petaling Jaya, Malaysia) for morphological comparisons. All the treated cells, including the floated cells, were then collected, trypsinised and washed twice with ice-cold PBS at 200 g for 5 min. The cells were resuspended in 350 µL of Annexin V staining buffer (BD Bioscience, San Jose, CA, USA) and stained with 5 µL of Annexin V conjugated with fluorescein isothiocyanate (Annexin V-FITC) (BD Bioscience, San Jose, CA, USA) on ice in dark conditions for 15 min. A 10 µL solution of 50 µg/mL propidium iodide (PI) was then added to the cells and incubated for another 5 min. After the staining, another 150 µL of Annexin V staining buffer was added to the cells, and the entire content was transferred to a flow tube for analysis. At least 10,000 stained cells were recorded and analysed using a BD FACSVerse flow cytometer (BD Bioscience, San Jose, CA, USA). The cell population with negative staining for both markers represents the viable cell population, positive staining for Annexin V-FITC represents the early apoptotic population, positive staining for PI represents the necrotic population, whereas positive staining for both markers represents a late apoptotic population. Both early and late apoptotic populations are collectively reported as the apoptotic population in this study.

Cell Cycle Assay
The cell cycle assay was conducted in accord with a previous study [51]. Briefly, 3 mL of SW1353 cells (5 × 10 4 cells/mL) was seeded in a 6-well plate for 24 h, followed by treatment with the IC 50 of MIA for 24 h. A serum-free condition was used as a cell cycle arrest control. All the treated cells, including the floating cells, were then collected, trypsinised and washed twice with ice-cold PBS at 200 g for 5 min. The cells were then counted, and 1 × 10 5 cells were fixed with 1 mL of 70% ethanol drop by drop while being vortexed. After that, they were kept overnight. The fixed cells were washed again with icecold PBS at 400 g for 10 min. Subsequently, the cells were stained with 500 µL of PI/RNase staining solution (BD Bioscience, San Jose, CA, USA) for 15 min at room temperature. At least 12,000 stained cells were recorded, and the cell cycle distribution was further analysed using ModFit LT software (Verity Software House, Topsham, ME, USA) via Gaussian curve modelling estimation [52,53].

Lysate Preparation and Protein Determination
Briefly, 15.5 mL of SW1353 cells (5 × 10 4 cells/mL) was seeded in a 100 mm culture dish for 24 h, followed by treatment with the IC 25  were collected as lysates and kept at −80 • C until analysis. The protein concentrations of lysates were determined using Bio-Rad Protein Assay concentrated dye reagent (Bio-Rad Laboratories, Hercules, CA, USA) in microtiter format according to the manufacturer's instructions. Bovine serum albumin (0-2 mg/mL) was used as the protein standard. The absorbance of each well was measured at 595 nm using a Multiskan GO microplate reader (Thermo Fisher Scientific, Vantaa, Finland). All the protein concentrations of lysates (500 µL) were then diluted to 1 mg/mL for the subsequent quantitative detection of isoprostane and TNF-α levels.

Enzyme-Linked Immunosorbent Assay
Human 8-isoprostane F2-α (catalogue No. E4805Hu) and TNF-α (catalogue No. E-EL-H0109) were determined using the sandwich-ELISA kits from Bioassay Technology Laboratory (Shanghai, China) and Elabscience (Houston, TX, USA). The procedures were conducted according to the manufacturer's instructions [54,55]. For human 8-isoprostane F2-α determination, 40 µL of lysates (1 mg/mL) and 10 µL of human 8-isoprostane F2-α antibody were added in duplicate into a 96-well plate pre-coated with human 8-isoprostane F2-α antibody. A 50 µL aliquot of standard solutions (0-800 ng/L) was added separately into other wells. Next, 50 µL of streptavidin-horseradish peroxidase (HRP) solution was added into all wells (except the blank) and the plate was incubated at 37 • C for 1 h. After that, the plate was washed 5 times with a wash buffer and blot-dried with paper towels, followed by the addition of 100 µL of substrate Solution A and substrate Solution B in a 1:1 ratio. The plate was sealed and further incubated at 37 • C in the dark for another 10 min. Lastly, 50 µL of the stop solution was added into each well, and the absorbance of each well was measured at 450 nm by using a Multiskan GO microplate reader (Thermo Fisher Scientific, Vantaa, Finland). The human 8-isoprostane F2-α levels of the lysates were calculated based on the standard curve.
For human TNF-α quantification, 100 µL of lysates (1 mg/mL) was added in duplicate into a 96-well plate pre-coated with human TNF-α antibody. Similar amounts of standard solutions (0-500 pg/mL) were added separately into other wells. The plate was incubated at 37 • C for 90 min. All the content of wells was then removed and 100 µL of a biotinylated detection antibody working solution was added directly. The plate was further incubated at 37 • C for 1 h. After that, the plate was washed 3 times with a wash buffer and blot-dried with paper towels, then 100 µL of an HRP conjugate working solution was added. The plate was sealed and further incubated at 37 • C for another 30 min. After that, the plate was washed 5 times with a wash buffer and blot-dried with paper towels, then 90 µL of the substrate reagent was added. The plate was incubated at 37 • C for another 15 min. Lastly, 50 µL of the stop solution was added into each well and the absorbance of each well was measured at 450 nm by using a Multiskan GO microplate reader (Thermo Fisher Scientific, Vantaa, Finland). The human TNF-α levels of lysates were determined based on the standard curve.

Statistical Analysis
The data were collected from three independent experiments with the respective technical replicates. Statistical analysis was conducted using SPSS software for Windows, version 25 (IBM, Armonk, NY, USA). Normality tests confirmed the normal distribution of data. Mean differences among the study groups were analysed with one-way analysis of variance with Tukey's or Dunnett's test as post hoc analysis. A p-value of less than 0.05 was considered statistically significant.  was considered statistically significant. Figure 1 shows the cytotoxicity of IL-1β, MIA and H2O2 on SW1353 chondrocytes upon 24 h of treatment. MIA (25 and 50 μM) and H2O2 (50, 100 and 200 μM) were cytotoxic to SW1353 chondrocytes, with a concentration-dependent decline in cell viability (all p < 0.001). The IC25 and IC50 values for H2O2 treatment were 55 and 82.5 μM. The values were 22.5 and 31.5 μM for MIA treatment. IL-1β did not cause a significant decline in the viability of cells up to 100 ng/mL upon 24 h treatment (p > 0.05). Therefore, the highest and second-highest concentrations of IL-1β (50 and 100 ng/mL) were used in the subsequent experiment.  Figure 2 shows the morphology of SW1353 chondrocytes after 24 h of MIA, H2O2 and IL-1β treatment. SW1353 cells in the VC group demonstrated a typical spindle-shaped fibroblast-like morphology. Active and proliferating cells with a shiny appearance were also observed. Consistent with the cytotoxicity and mode of cell death   findings, IL-1β did not cause any morphological changes in SW1353 chondrocytes compared with VC. Upon MIA treatment, there was no obvious change in SW1353 cell morphology, but the number of attached cells was reduced in a concentration-dependent manner. On the other hand, SW1353 chondrocytes treated with H2O2 demonstrated typical apoptotic features, including blebbing formation, cell shrinkage and apoptotic body formation. Besides, H2O2-treated SW1353 cells exhibited elongated or epithelial-shaped cells with reduced cell density.

Mode of SW1353 Cell Death under MIA, H 2 O 2 and IL-1β Treatment
Plasma membrane phosphatidylserine externalization is one of the earliest signs of apoptosis [50,56]. The exposed phosphatidylserine on the cell surface can be bound with Annexin V in the presence of calcium ions [57]. In this study, an Annexin V-FITC/PI duallabelling assay was used to distinguish the mode of SW1353 cell death, either via apoptosis and/or necrosis. Figure  SW1353 cell death with a primarily increase in the apoptotic population (p < 0.001) and a slight increase in the necrotic population (p < 0.001). Parallel with the cytotoxicity data, IL-1β did not induce significant SW1353 cell death compared with the VC (p > 0.05). Surprisingly, MIA (IC 25 and IC 50 ) did not induce significant SW1353 cell death after 24 h of treatment (p > 0.05).

Mode of SW1353 Cell Death under MIA, H2O2 and IL-1β Treatment
Plasma membrane phosphatidylserine externalization is one of the earliest signs of apoptosis [50,56]. The exposed phosphatidylserine on the cell surface can be bound with Annexin V in the presence of calcium ions [57]. In this study, an Annexin V-FITC/PI dual-labelling assay was used to distinguish the mode of SW1353 cell death, either via apoptosis and/or necrosis. Figure 3 summarises the viable, apoptotic and necrotic cell populations with or without the respective treatments. H2O2 (both IC25 and IC50) significantly induced SW1353 cell death with a primarily increase in the apoptotic population (p < 0.001) and a slight increase in the necrotic population (p < 0.001). Parallel with the cytotoxicity data, IL-1β did not induce significant SW1353 cell death compared with the VC (p > 0.05). Surprisingly, MIA (IC25 and IC50) did not induce significant SW1353 cell death after 24 h of treatment (p > 0.05). Figure 3. Mode of SW1353 cell death under treatment with MIA, H2O2 and IL-1β. The IC25 and IC50 of MIA did not significantly induce apoptosis or necrosis in SW1353 chondrocytes. Therefore, MIA was cytostatic at these concentrations, whereby it arrested the growth of SW1353 chondrocytes but did not induce cell death. On the other hand, H2O2 (IC25 and IC50) significantly induced apoptosis in SW1353 chondrocytes. Additionally, H2O2 marginally but significantly increased necrosis in chondrocytes. IL-1β (50 and 100 ng/mL) did not significantly induce apoptosis or necrosis in SW1353 chondrocytes, parallel with the cytotoxicity result. The data were collected from three independent experiments. *** p < 0.001 against the VC. Figure 4 demonstrates the cell cycle analysis of SW1353 chondrocytes with VC, MIA (IC50) and serum-free media for 24 h. Serum deprivation is known to induce G1 arrest [58]. In this study, SW1353 cells with serum-free media showed a significant accumulation of the G1-phase cell population (p < 0.001), while the MIA treatment significantly increased the S-phase cell population (p < 0.05) with the lower G2-phase population, indicating a S-phase arrest.  Figure 4 demonstrates the cell cycle analysis of SW1353 chondrocytes with VC, MIA (IC 50 ) and serum-free media for 24 h. Serum deprivation is known to induce G1 arrest [58]. In this study, SW1353 cells with serum-free media showed a significant accumulation of the G1-phase cell population (p < 0.001), while the MIA treatment significantly increased the S-phase cell population (p < 0.05) with the lower G2-phase population, indicating a S-phase arrest.

8-Isoprostane F2-α and TNF-α Levels
8-isoprostane F2-α is one of the oxidative stress and lipid peroxidation markers, which is stably formed via non-enzymatic peroxidation of fatty acids, primarily arachidonic acid [59]. Figure 5 shows the 8-isoprostane F2-α and TNF-α levels of SW1353 chondrocytes after 24 h MIA, H 2 O 2 and IL-1β treatment. The 24 h treatment with MIA and H 2 O 2 significantly reduced the TNF-α levels (all p < 0.05) compared with the VC, while the IL-1β treatment significantly increased the TNF-α levels (all p < 0.001) compared with the VC in a concentration-dependent manner. On the other hand, SW1353 chondrocytes treated with the IC 25 and IC 50 of H 2 O 2 demonstrated significantly higher 8-isoprostane F2-α levels (all p < 0.05). A similar increasing trend was observed in MIA-treated SW1353 chondrocytes, but only the IC 50 of MIA treatment showed statistical significance compared with the VC (all p < 0.05). Neither 50 nor 100 ng/mL IL-1β significantly altered the isoprostane levels in SW1353 chondrocytes (p > 0.05), despite a significant induction of TNF-α levels.   pared with the VC in a concentration-dependent manner. On the other hand, SW1353 chondrocytes treated with the IC25 and IC50 of H2O2 demonstrated significantly higher 8-isoprostane F2-α levels (all p < 0.05). A similar increasing trend was observed in MIA-treated SW1353 chondrocytes, but only the IC50 of MIA treatment showed statistical significance compared with the VC (all p < 0.05). Neither 50 nor 100 ng/mL IL-1β significantly altered the isoprostane levels in SW1353 chondrocytes (p > 0.05), despite a significant induction of TNF-α levels.
(a) (b) Figure 5. The isoprostane and tumour necrosis factor-α (TNF-α) levels under MIA, H2O2 and IL-1β treatment. (a) H2O2 and MIA (IC50 only) induced oxidative stress, with significantly higher isoprostane levels compared with the VC. IL-1β did not cause any upregulation of isoprostane levels, indicating the absence of oxidative stress. (b) IL-1β induced inflammation with higher TNF-α levels. H2O2 and MIA did not induce inflammation, with low TNF-α levels. The data were collected from three independent experiments with two technical replicates. * p < 0.05 against the VC; ** p < 0.01 against the VC; *** p < 0.001 against the VC.

Discussion
In this study, 24 h of IL-1β was non-cytotoxic to SW1353 chondrocytes, without significant cell death up to 100 ng/mL. The cell morphology of IL-1β-treated SW1353 cells appeared similar to the control, consistent with the cytotoxicity and cell death findings. Besides, IL-1β-treated SW1353 cells were detected at a higher intracellular TNF-α level but insignificant changes to the 8-isoprostane F2-α (oxidative stress marker) level. On the other hand, 24 h of H 2 O 2 induced concentration-dependent cytotoxicity and cell death, primarily via apoptosis in SW1353 chondrocytes, with the typical apoptotic features including blebbing formation, cell shrinkage and cell debris formation. Besides, there was a slight but significant increase (3-5%) in the necrotic cell population upon H 2 O 2 treatment which might be due to the direct H 2 O 2 damage to the cell membrane [35]. H 2 O 2 also significantly increased the 8-isoprostane F2-α levels in SW1353 cells, which is coherent with its known oxidant activity [15,37,38]. However, H 2 O 2 reduced the intracellular TNF-α levels in SW1353 chondrocytes. On the other hand, 24 h of MIA decreased the viability of SW1353 cells in a concentration-dependent manner but without significant cell death. MIA-treated SW1353 chondrocytes were morphologically similar to the VC but with fewer attached cells. Subsequently, we demonstrated that MIA treatment caused S-phase cell cycle arrest on SW1353 cells, with a higher level of 8-isoprostane F2-α. DNA damage and GSH depletion have been reported to cause S-phase arrest for cell repair process [60]. Therefore, it is rational to speculate that MIA-mediated oxidative stress may induce oxidative DNA damage to SW1353 cells, which subsequently triggers cell cycle arrest but not cell death.
There was a significant increase in intracellular TNF-α levels in IL-1β-treated cells. Mechanistically, IL-1β activates the activator protein-1 and nuclear factor kappa-lightchain-enhancer of activated B cell signalling pathway, subsequently upregulating the expression of downstream cytokines, including IL-1β and TNF-α [80]. A 24 h of 10 ng/mL IL-1β treatment was reported to increase TNF-α levels by four-fold in serum-starved SW1353 chondrocytes [28]. Only a 1.7-to 3.2-fold increase in TNF-α levels upon 50 and 100 ng/mL of IL-1β treatment was observed in our study. As suggested earlier, serum deprivation may enhance the IL-1β signalling in SW1353 cells and greatly upregulated the TNF-α level, despite being at lower concentration. In terms of morphology, the IL-1βor lipopolysaccharide-stimulated primary human articular chondrocytes displayed no noticeable changes [81] or changed to a rounded shape with smaller cell size [75]. In both cases, no cell death features were observed. Similarly, we also demonstrated that IL-1β did not induce obvious morphological changes in SW1353 cells.
Oxidative stress occurs when cell antioxidant defences are overwhelmed by ROS accumulation [82]. Previous studies reported that IL-1β (40 ng/mL) could induce ROS production in chondrocytes [73]. However, a time-course analysis on the 10 ng/mL IL-1βtreated SW1353 chondrocytes revealed that ROS levels were significantly increased only during the first hour of treatment [83]. Subsequently, the ROS levels reverted to baseline when the intracellular superoxide dismutase level was adaptively upregulated due to redox signalling [83]. Moreover, a non-cytotoxic ROS level could act as a secondary messenger to trigger redox and cytokine signalling [84,85]. Only overwhelmingly high ROS levels could damage the cell antioxidant defence, resulting in oxidative damage and cell death [86]. In this study, IL-1β might upregulate an early but non-lethal ROS level as a secondary messenger to stimulate the cytokine signalling, without overwhelming the cell antioxidant defence mechanism. Therefore, cell cycle arrest or cell death events were not triggered. Analysis of the antioxidant levels and/or direct ROS measurement in a time-course manner will be helpful for better visualisation of the oxidative status upon IL-1β treatment.
Our study showed that H 2 O 2 and MIA were cytotoxic and cytostatic to SW1353 chondrocytes, respectively. These changes were accompanied by an increase in 8-isoprostane F2-α levels but a lower level of intracellular TNF-α. Previous studies showed that the IC 50 values of H 2 O 2 in SW1353 chondrocytes ranged from 0.5 to 1 mM upon 1 h to 24 h of treatment [15,[37][38][39], while IC 50 values around 5-8.26 µM for MIA were reported for SW1353 chondrocytes [16,43] and primary chondrocytes [17,46,87]. Our data demonstrated  50 values of H 2 O 2 and MIA were 82.5 µM (6-to 12-fold lower) and 31.5 µM (4-to 6-fold higher), respectively. Multiple factors, such as batch-to-batch variation of cells and reagents, and culturing and treatment conditions, may have contributed to this discrepancy. The differences in media composition [88] may affect the sensitivity of cells. The DMEM (ThermoFisher Scientific, catalogue No. 11965092) used in this study contains high glucose and 4 mM l-glutamine without sodium pyruvate. l-glutamine in the media could contribute to the higher IC 50 value of MIA because it could provide an alternative energy source to the cells when glycolysis is inhibited [89]. Additionally, the greater sensitivity of SW1353 cells towards H 2 O 2 in our study might be due to the absence of pyruvate in the culture media, where 0.5-1 mM pyruvate (the typical concentration in media) was shown to protect the cells from exogenous H 2 O 2 [90,91]. Besides, the IC 50 values of H 2 O 2 from the media without pyruvate was about five-fold lower than in those containing pyruvate [91], which is similar to our findings. Nevertheless, detailed information on the media composition used in previous studies [15][16][17][37][38][39]43,46,87] is not available. A further comparative study is required to confirm our speculation.
Both H 2 O 2 and MIA treatment significantly reduced the TNF-α levels in SW1353 chondrocytes despite higher 8-isoprostane F2-α levels. H 2 O 2 [15,37,38] and MIA [16,17,43,46,87,92] were consistently reported to induce oxidative stress in SW1353 chondrocytes and/or primary rat chondrocytes. However, few studies demonstrated the upregulation of intracellular TNF-α levels in H 2 O 2 -or MIA-treated SW1353 chondrocytes. The available studies showed that 24 h of H 2 O 2 (0.5-10 mM) increased the release of TNF-α in SW1353 chondrocytes [37] and CHON-001 chondrocytes [19]. A lower concentration of H 2 O 2 treatment (40 ng/mL or 1.17 µM) also increased the intracellular TNF-α levels in primary rat articular chondrocytes [93]. MIA was reported to upregulate the TNF-α mRNA level [43,92] and secretion of TNF-α [92] in SW1353 chondrocytes. The underlying reason for this discrepancy is unclear. We hypothesise that the induction of cytostasis and apoptosis by MIA and H 2 O 2 may interfere with the expression of TNF-α. Several proteinases, including caspases and calpains, are activated during apoptosis and they actively degrade the intracellular proteins and organelles [94]. Further studies are required to confirm the role of the cytostatic and cytotoxic effects of MIA and H 2 O 2 treatment in downregulating TNF-α levels.
There are several limitations to this study. The use of single oxidative and inflammatory markers renders the interpretation of data non-comprehensive. The status of other cytokines or mediators like IL-6 and MMPs may be useful in validating the effects of OA inducers. Besides, direct measurement of ROS species is more informative than an indirect indication from the oxidative markers, although the ROS are extremely reactive and transiently present, which may be difficult to visualize [95]. Time-course analysis of the ROS levels and inflammatory markers may help to identify the role of redox and cytokine signalling in mediating the effects of these inducers in chondrocytes. Moreover, the single time-point assessment used in this study may have missed out some critical changes in cell growth.

Conclusions
In summary, our study demonstrated that H 2 O 2 induced apoptosis while MIA induced a cytostatic effect on SW1353 cells, without a significant inflammation process. IL-1β is a suitable inducer of inflammation but not SW1353 cell death. Caution should be taken regarding the treatment conditions, especially the media composition and serum-free conditions in these in vitro OA models.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to institutional policy.