Cell Culture
Normal human articular chondrocytes were obtained from a twenty-one-year-old donor during previous studies26. Normal bovine articular chondrocytes and synovial cells were obtained from the stifle joints of three-week-old calves provided by an abattoir. The human and bovine chondrocytes used in the present study were cultured in monolayer for less than four weeks. MG-63 (a human osteosarcoma cell line) and C3H10T1/2 (a mouse embryo mesenchymal cell line) were obtained from the American Type Culture Collection (Manassas, Virginia). All of the cells were cultured with Dulbecco's Modified Eagle's Medium/Ham's F-12 (DMEM/F12) medium containing 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen, Carlsbad, California) in a 5% CO2-humidified incubator at 37°C. Human and bovine articular chondrocytes were used to test the toxicity to the cells in articular cartilage; bovine synovial cells were used to test the toxicity to synovium; MG-63 cells were used to test the toxicity to the bone cells; and C3H10T1/2 cells were used to test the toxicity to fibroblasts.
Treatments of Cells in Monolayer Culture
Human and bovine articular chondrocytes were plated in a twelve-well plate at a density of 200,000 cells per well in 1 mL of culture medium. After twenty-four hours, the cells were treated with 0.3 mL of phosphate-buffered saline solution (control group) or drug solutions (treatment groups) by siphoning off the old medium and refilling with treatment solutions. For time-dependent assessment, the cells were treated with 6 mg/mL Celestone (the original concentration of corticosteroids) for 0.5, one, three, four, five, fifteen, and thirty minutes. For dosage-dependent assessment, the cells were treated with Celestone at 6, 0.6, 0.1, 0.01, 0.001, and 0.0001 mg/mL (diluted in phosphate-buffered saline solution) for thirty minutes. For investigation of the effects of the components of Celestone, the individual components (including betamethasone acetate, betamethasone sodium phosphate [=97% powder], and benzalkonium chloride [50% in water]) were purchased from Sigma-Aldrich (St. Louis, Missouri) and were made into solutions in ethanol (for betamethasone acetate) or phosphate-buffered saline solution (for betamethasone sodium phosphate or benzalkonium chloride). The cells were treated for thirty minutes with betamethasone acetate or betamethasone sodium phosphate at 0.2, 0.6, 1, 3, and 6 mg/mL or with benzalkonium chloride at 4, 8, 10, 20, and 200 µg/mL. The controls were either phosphate-buffered saline solution (for betamethasone sodium phosphate or benzalkonium chloride) or phosphate-buffered saline solution containing 2% ethanol (for betamethasone acetate). The maximum dosages of each treatment were chosen on the basis of the maximum concentrations of total corticosteroids or benzalkonium chloride in Celestone. After treatment, the cells were washed two times with phosphate-buffered saline solution, were maintained in culture medium, and were incubated at 37°C. The cells were observed under a light microscope immediately after treatment and were followed for seven days.
Treatment of Cells in Suspension
Approximately 0.3 million cells in phosphate-buffered saline solution suspension were placed into 12 × 75-mm plastic nonpermeable test tubes. The cells were kept in suspension by means of gentle vortexing once every five minutes. Each control or treatment group had three tubes. The control group was treated with phosphate-buffered saline solution only. The treatment groups were treated according to the following protocol.
Human and bovine articular chondrocytes were treated with betamethasone sodium phosphate only (1, 3, or 6 mg/mL), benzalkonium chloride only (4, 8, 10, 13.3, 20, or 200 µg/mL [of note, 50, 100 and 150 µg/mL had effects similar to 20 or 200 µg/mL]), a combination of betamethasone sodium phosphate (1, 3, or 6 mg/mL) and benzalkonium chloride (200 µg/mL), or benzalkonium chloride only (13.3, 20, or 200 µg/mL) but with 10% fetal bovine serum in cell suspension.
Bovine synovial cells were treated with benzalkonium chloride only (8, 10, 20, or 200 µg/mL).
MG-63 and C3H10T1/2 cells were treated with benzalkonium chloride only (4, 8, 10, 13.3, 20, or 200 µg/mL) or benzalkonium chloride only (13.3, 20, or 200 µg/mL) but with 10% fetal bovine serum in cell suspension.
All of the cells were treated for thirty minutes at room temperature in the test tubes. Immediately after treatment, the cells were washed with phosphate-buffered saline solution once and were processed for flow cytometry analysis.
Flow Cytometry Analysis of Cell Viability
The control and treated cells were incubated in 0.3 mL of staining solution containing 0.8-µM ethidium homodimer-1 and 0.8-µM calcein AM in phosphate-buffered saline solution for twenty minutes in the dark at room temperature. The reagents were from the Live/Dead Viability/Cytotoxicity Kit (Molecular Probes, Eugene, Oregon). Ethidium homodimer-1 enters dead cells with damaged membranes and produces a bright red fluorescence on binding to nucleic acid, whereas it cannot enter live cells with intact plasma membrane. Calcein AM enters both live and dead cells, but only live cells have the intracellular esterase activity that changes the nonfluorescent calcein AM into intense green fluorescent calcein. Therefore, the dead cells present red fluorescence and the live cells present green fluorescence27, which can be discriminated by means of flow cytometry analysis. The percentages of live and dead cells were obtained by counting 10,000 cells with use of a Life Science Research II (BD LSR II) analyzer (BD Biosciences, San Jose, California).
Treatment of Fresh Bovine Fetlock Joints
Eight fetlock joints (four pairs) were obtained from approximately-two-month-old calves provided by a local abattoir and were studied within three hours after the time of death. The right or left joint of each pair was randomly selected for phosphate-buffered saline solution or Celestone treatment. Fetlock joints were exposed aseptically and were incubated with phosphate-buffered saline solution or Celestone suspension in 100-mm dishes for thirty minutes. After two washes (ten minutes each) in phosphate-buffered saline solution, the control or treated cartilage was sampled by means of a 5-mm skin punch. The superficial layer (approximately 100 µm thick) of each cartilage cylinder was cut with use of a blade. These cartilage discs were stained with 0.8-µM ethidium homodimer-1 and 0.8-µM calcein AM in phosphate-buffered saline solution for twenty minutes in the dark at room temperature. The cartilage discs were photographed under a confocal microscope (Leica TCS SP2; Leica Microsystems, Exton, Pennsylvania) with a 10× objective lens.
Statistical Analysis
The means and standard deviations of triplicate samples per group were compared between the phosphate-buffered saline solution control group and each treatment group with use of the two-tailed Student t test. The level of significance was set at p < 0.05. The experiments were repeated at least twice, and representative sets of data were shown. Error bars represented the standard deviations.
Source of Funding
There was no external funding source for this study.
In the monolayer culture experiments, we found that Celestone caused death of all of the cells (Fig. 1). The cells were instantly broken into pieces and mostly detached from the culture plate after as short as thirty seconds of exposure. Neither betamethasone sodium phosphate alone nor betamethasone acetate alone caused any cell death. Benzalkonium chloride at dosages of 20 and 200 µg/mL caused death of all of the cells. After the cells were cultured for three additional days, the phosphate-buffered saline solution-treated control cells and the betamethasone sodium phosphate-treated cells continued to grow and fill the entire wells of the culture plate, whereas there was no increase of cell number in the cells treated with Celestone and benzalkonium chloride (Fig. 1; Day-1 panels compared with Day-4 panels). The Live/Dead staining assay confirmed that the benzalkonium chloride-treated cells were dead (Fig. 2). Flow cytometry analysis clearly separated the live cells (Quadrant 4) from the dead cells (Quadrant 1). A few cells fell into Quadrant 2 (representing cells that were in the early phase of death, which had some remaining esterase activity but with broken membranes) and Quadrant 3 (representing cells with intact membrane but not stained with calcein).
In the studies on human articular chondrocytes (Fig. 3), betamethasone sodium phosphate at the three dosages (1, 3, and 6 mg/mL) did not cause significant cell death in comparison with the phosphate-buffered saline solution control group (p = 0.674, 0.327, and 0.287, respectively). In contrast, although benzalkonium chloride at dosages of 4 and 8 µg/mL did not cause significant cell death (p = 0.079 and 0.500, respectively, in comparison with the phosphate-buffered saline solution control group), benzalkonium chloride at a dosage of 10 µg/mL caused a cell death rate of 10.6% (95% confidence interval, 8.2% to 13.0%) (p = 0.009 in comparison with the phosphate-buffered saline solution control group); benzalkonium chloride at a dosage of 13.3 µg/mL caused a cell death rate of 21.0% (95% confidence interval, 10.1% to 16.5%) (p = 0.003 in comparison with the phosphate-buffered saline solution control group); and benzalkonium chloride at dosages of 20 and 200 µg/mL caused a cell death rate of 99.3% (95% confidence interval, 99.0% to 99.6%) and 99.4% (95% confidence interval, 99.2% to 99.6%), respectively (p < 0.001 and p < 0.001, respectively, in comparison with the phosphate-buffered saline solution control group). When betamethasone sodium phosphate at the three dosages (1, 3, and 6 mg/mL) was combined with benzalkonium chloride at a dosage of 200 µg/mL, 99.3% (95% confidence interval, 98.6% to 100.0%), 99.4% (95% confidence interval, 99.1% to 99.7%), and 97.9% (95% confidence interval, 97.8% to 98.0%), respectively, of the cells were dead after treatment (p < 0.001, p < 0.001, and p < 0.001, respectively, in comparison with the phosphate-buffered saline solution control group). When 10% fetal bovine serum was included in the treatment, benzalkonium chloride at dosages of 13.3 and 20 µg/mL only caused cell death rates of 7.8% (95% confidence interval, 2.1% to 13.5%) and 2.2% (95% confidence interval, 2.0% to 2.4%), respectively, which were not significant in comparison with the rate of cell death in the phosphate-buffered saline solution control group (p = 0.149 and 0.181, respectively). However, in the presence of 10% fetal bovine serum, benzalkonium chloride at a dosage of 200 µg/mL caused a cell death rate of 99.5% (95% confidence interval, 99.5% to 99.5%) (p < 0.001 in comparison with the phosphate-buffered saline solution control group).
In the studies on bovine articular chondrocytes (Fig. 4), betamethasone sodium phosphate at the three dosages (1, 3, and 6 mg/mL) did not cause significant rates of cell death, compared with the phosphate-buffered saline solution control group. In contrast, benzalkonium chloride caused cell death in a dose-dependent manner. When betamethasone sodium phosphate at the three dosages (1, 3, and 6 mg/mL) was combined with benzalkonium chloride at a dosage of 200 µg/mL, almost all of the cells were dead after treatment. While 10% fetal bovine serum blocked the toxicity of benzalkonium chloride at dosages of 13.3 and 20 µg/mL, it had no effect in preventing cell death induced by benzalkonium chloride at a dosage of 200 µg/mL.
In the studies on bovine synovial cells (Fig. 5), benzalkonium chloride at a dosage of 8 µg/mL caused a cell death rate of 27.1% (95% confidence interval, 22.6% to 31.6%) (p = 0.013) and benzalkonium chloride at dosages of 10, 20, and 200 µg/mL caused a cell death rate of 99.9% (95% confidence interval, 99.8% to 100.0%), 99.8% (95% confidence interval, 99.7% to 99.9%), and 99.8% (95% confidence interval, 99.6% to 100.0%), respectively (p < 0.001, p < 0.001, and p < 0.001, respectively).
In the studies on mouse fibroblasts (C3H10T1/2 cells) (Fig. 6, A), benzalkonium chloride at a dosage of 4 µg/mL caused a cell death rate of 2.7% (95% confidence interval, 1.2% to 4.2%) (p = 0.105); benzalkonium chloride at a dosage of 8 µg/mL caused a cell death rate of 32.0% (95% confidence interval, 3.3% to 60.7%) (p = 0.032); benzalkonium chloride at a dosage of 10 µg/mL caused a cell death rate of 25.5% (95% confidence interval, 2.1% to 48.9%) (p = 0.036); benzalkonium chloride at a dosage of 13.3 µg/mL caused a cell death rate of 45.6% (95% confidence interval, 7.2% to 84.0%) (p = 0.024); and benzalkonium chloride at dosages of 20 and 200 µg/mL caused a cell death rate of 98.3% (95% confidence interval, 94.9% to 100.0%) and 99.9% (95% confidence interval, 99.8% to 100.0%), respectively (p < 0.001 and p < 0.001, respectively). When 10% fetal bovine serum was included in the treatment, benzalkonium chloride at dosages of 13.3 and 20 µg/mL caused cell death rates of 8.2% (95% confidence interval, 0.4% to 16.0%) and 14.9% (95% confidence interval, 0.0% to 30.0%), respectively, which were not significant (p = 0.066 and 0.058, respectively). However, in the presence of 10% fetal bovine serum, benzalkonium chloride at a dosage of 200 µg/mL caused a cell death rate of 99.9% (95% confidence interval, 99.8% to 100.0%) (p < 0.001).
In the studies of human osteosarcoma MG-63 cells (Fig. 6, B), we found that benzalkonium chloride killed MG-63 cells in a dose-dependent manner. While 10% fetal bovine serum blocked the toxicity of benzalkonium chloride at dosages of 13.3 and 20 µg/mL, it had no effect in preventing cell death induced by benzalkonium chloride at a dosage of 200 µg/mL.
In the studies of fresh bovine fetlock joints, we found more dead chondrocytes in the Celestone-treated joints than in the phosphate-buffered saline solution control group.
Recently, Seshadri et al.23 reported that Depo-Medrol caused a cell death rate of about 60% when bovine articular chondrocytes were treated with the drug for fifteen minutes. In the present study, we found that Celestone caused instant cell death when the articular chondrocytes were exposed to the drug for as briefly as thirty seconds. We suspected that Celestone's hypo-osmolality (175 mOsm/kg)28 might be the cause of instant cell death. The physiological osmolality for cells ranges from 280 to 320 mOsm/kg (when the concentration is very low, osmolarity and osmolality are considered equivalent)29. It has been demonstrated that hypo-osmolality (an osmolality of <280 mOsm/kg) contributes to chondrocyte death in wounded articular cartilage30,31. However, the hypo-osmolarity of Celestone may be unimportant in an in vivo application of Celestone in a human joint because Celestone (1 to 2 mL) is often mixed with 5 mL of 1% lidocaine (or 0.25% bupivacaine) for injection. Once injected, the drugs mix with 2 to 3 mL of synovial fluid that typically exists in the joint32. The local anesthetics and synovial fluid have physiological osmolarity. Therefore, the hypo-osmolarity of Celestone may be corrected to almost a physiological level on injection. To simulate the Celestone concentration in the in vivo injection, we mixed 1 mL of Celestone with 9 mL of phosphate buffered saline solution (287 mOsm/kg). We found that this diluted Celestone (0.6 mg/mL) still caused death of chondrocytes. Then, we speculated that certain chemical components in the Celestone might be cytotoxic.
We ruled out the possibility that dibasic sodium phosphate, monobasic sodium phosphate, and edetate disodium are the cause of cell death because these chemicals at their concentrations in Celestone are often used in solutions for cell culture (such as in phosphate-buffered saline solution or trypsin solution) and have been proved not to be cytotoxic33. We individually tested the remaining three chemicals, including the two corticosteroids (betamethasone acetate and betamethasone sodium phosphate) and the one preservative (benzalkonium chloride). Our results demonstrated that neither betamethasone acetate nor betamethasone sodium phosphate at a dosage of as high as 6 mg/mL (two times the concentration in Celestone) caused death of human or bovine articular chondrocytes. In contrast, benzalkonium chloride caused significant cell death starting at a dosage (10 µg/mL) as low as one-twentieth of the concentration in Celestone. The cell death was dose-dependent. At higher dosages such as 20 or 200 µg/mL of benzalkonium chloride, >99% of the cells were dead after treatment. Consequently, when betamethasone sodium phosphate was combined with 200 µg/mL of benzalkonium chloride, >99% of the cells were dead after treatment, which could be attributed to benzalkonium chloride because betamethasone sodium phosphate did not cause cell death by itself.
Benzalkonium chloride is used as a foaming, cleansing, and bactericidal agent at concentrations of as high as 5.0%. It has been found by the American College of Toxicology to be safe for dermatologic and ocular use at concentrations of as high as 0.1%34. Many drug formulas contain benzalkonium chloride; for example, Ventolin (albuterol) respirator solution (Salbutamol Inhalation; Allen & Hanburys, Uxbridge, United Kingdom) contains 10 µg/mL benzalkonium chloride, betamethasone sodium phosphate 0.1% eye drops (UCB Pharma, Dunstable, United Kingdom) contain 20 µg/mL benzalkonium chloride, and 2% cromoglycate nasal spray (Vividrin Nasal Spray; Pharma-Global, Sandycove, Dublin, Ireland) contains 10 µg/mL benzalkonium chloride. The cytotoxicity of benzalkonium chloride is due to the disruption of intermolecular interactions, which causes dissociation of cellular membrane bilayers, leading to leakage of cellular contents and cell death35. Two studies by Berg et al. demonstrated that benzalkonium chloride was a toxic component of nasal sprays in both in vitro and in vivo experiment models, which suggested that the preservative not be used36,37. The toxicity of benzalkonium chloride on articular chondrocytes has never been reported previously, to our knowledge. The present study showed that benzalkonium chloride was toxic not only to chondrocytes but also to synovial cells. Furthermore, benzalkonium chloride also caused death of osteoblast-like cells (MG-63) and fibroblasts (C3H10T1/2). We found that 10% fetal bovine serum could block the toxicity of low dosages of benzalkonium chloride, but not the dosage used in Celestone (200 µg/mL). This finding implies that the synovial fluid may not be able to neutralize Celestone's toxicity. However, when Celestone is mixed with local anesthetics for intra-articular injection in patients, the benzalkonium chloride concentration is diluted; thus, its toxicity may be neutralized to a certain extent, depending on the volume of injection.
It is worth pointing out that not all intra-articular corticosteroid injections contain benzalkonium chloride as preservative. For example, myristyl gamma-picolinium chloride is used in Depo-Medrol, whereas benzyl alcohol is used in other injections28. Both benzalkonium chloride and myristyl gamma-picolinium chloride belong to the quaternary ammonium group of cationic surface-acting agents, which may share the same mechanism of cytotoxicity, given that both Celestone and Depo-Medrol caused death of articular chondrocytes. However, the toxicity of myristyl gamma-picolinium chloride was not investigated either in the previous study23 or in our current study because it is not commercially available.
The limitation of the present study is that the cultured cells were treated in vitro—a condition not equivalent to the joint in vivo, where the synovial fluid and cartilage matrix may protect the chondrocytes. It has been shown that high molecular weight hyaluronan—a major component of cartilage matrix—decreases the toxicity of benzalkonium chloride to human epithelial cells35. In fact, when we soaked the fresh bovine fetlock joints in Celestone suspension, we observed that there were more dead chondrocytes in comparison with the phosphate-buffered saline solution control group. However, the majority of the chondrocytes were still alive, suggesting that the chondrocytes in cartilage were more resistant to benzalkonium chloride toxicity. Celestone (1 to 2 mL) is often mixed with 5 mL of local anesthetics for injection in patients; once injected, the drugs mix with 2 to 3 mL of synovial fluid that exists in the joint32. Therefore, Celestone (and benzalkonium chloride) is diluted five to ten times; that is, the benzalkonium chloride concentration in an injected joint is estimated to be about 20 to 40 µg/mL (the benzalkonium chloride concentration in Celestone suspension is 200 µg/mL). These concentrations are still toxic to in vitro cultured cells, but the actual toxicity in a joint may not be as severe as in cultured cells.
In conclusion, we have shown that Celestone caused death of human and bovine articular chondrocytes as well as bovine synovial cells, possibly because of its preservative benzalkonium chloride. The corticosteroids per se did not cause significant chondrocyte death under the conditions tested. These findings should stimulate an in vivo animal study to test the degree to which the changes seen in vitro also occur in vivo before any recommendations are made with regard to the clinical use of Celestone.
Note: The authors thank Mary Price (Senior Research Scientist, Cell Analysis Core Facility, Tulane Cancer Center) for help in flow cytometry analysis, Wei Huang and Dr. Guoyong Wang (Tulane University) for help in confocal microscopy, and Donna M. Watkins and Rita Richardson (Tulane University) for help in obtaining bovine joints and laboratory assistance.