Osteosarcoma is the most common bone sarcoma in children and adolescents. Although the prognosis for these patients has improved substantially, primarily because of the use of adjuvant chemotherapy, 20 to 40 per cent of patients still die as a result of metastasis of the tumor29. The main cause of therapeutic failure may be the resistance of the tumor cells to drug treatment. More than 80 per cent of all patients who have osteosarcoma are thought to have undetectable micrometastatic disease at the time of diagnosis20. Therefore, the current treatment of human osteosarcoma consists of adjuvant treatment of micrometastases with multi-drug chemotherapy, as well as operative removal of the primary tumor.
Several factors are related to the drug resistance of tumor cells; the most common mechanism in cancers in humans may be decreased accumulation of drugs caused by the membrane glycoprotein P-glycoprotein18,23,25. The term P-glycoprotein was chosen because initially it was believed to decrease the permeability of the cell membrane to drugs. Subsequently, it was determined that P-glycoprotein is an adenosine triphosphate-dependent drug-efflux pump that extrudes a variety of structurally unrelated drugs from the cell and is a mechanism of multidrug resistance10,22. Overexpression of P-glycoprotein has been widely observed in various cancers19,35. Moreover, recent investigations have demonstrated a substantial relationship between the expression of P-glycoprotein and the clinical outcome in some cancers, including neuroblastoma5,8, soft-tissue sarcoma6, and osteosarcoma7. The successful treatment of drug resistance in osteosarcoma may result in an improved prognosis.
Although several experimental models have been developed to study the biological characteristics or the treatment of osteosarcoma, no suitable models have been established to investigate drug-resistant osteosarcoma, to our knowledge. In this paper, we describe two newly developed multidrug-resistant murine osteosarcoma cell lines that overexpress P-glycoprotein. These cell lines may provide valuable in vitro and in vivo models that are applicable to the development of treatment protocols for multidrug-resistant osteosarcoma.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Department of Orthopaedic Surgery, Kohoku General Hospital, 1221 Kuroda, Kinomoto-cho, Ika-gun 529-04, Shiga Prefecture, Japan.
‡Department of Orthopaedic Surgery, Massachusetts General Hospital, Boston, Massachusetts 02114.
§Department of Orthopaedic Surgery, Kyoto Prefectural University of Medicine, Kyoto 602, Japan.
Establishment of Cell Lines
The parent cell line, designated MOS, was established from the murine osteosarcoma model developed at Massachusetts General Hospital (MGH-OGS); this model has been shown to be similar to human osteosarcoma with regard to histological characteristics, growth pattern, and response to chemotherapy2,11. The drug-resistant subclones, MOS/ADR1 and MOS/ADR2, were established by single-cell culture after the MOS cells were exposed to six pulse, stepwise increments of Adriamycin (doxorubicin) at concentrations ranging from 0.01 to one microgram per milliliter. For each pulse treatment, cells growing exponentially in six-well plates were exposed to Adriamycin for three days. The treated cells then were washed thoroughly with phosphate-buffered saline solution and grown in Adriamycin-free medium until confluent; they were subjected to another pulse treatment after the cells had been obtained by exposure to trypsin. After the sixth treatment with Adriamycin, single-cell clones were obtained with use of a limiting dilution method. The plates were inspected with a microscope on a daily basis to identify wells that contained a single cell. In a preliminary survey to select the drug-resistant clones, the intracellular accumulation of Adriamycin and the staining of the fifteen cloned cell lines for P-glycoprotein were compared with that in the parent cells with the methods to be described. All of the cell lines were stored in liquid nitrogen, and the experiments were performed within a month after the cells had been thawed. The cell lines were maintained in Dulbecco modified Eagle medium supplemented with fifteen-millimolar HEPES buffer, 10 per cent fetal calf serum, and an antibiotic solution of penicillin (100 units per milliliter) and streptomycin (fifty micrograms per milliliter) and were stored at 37 degrees Celsius in a humidified incubator containing 5 per cent carbon dioxide. Two subclones were selected for the present study. All of the experiments described were carried out during the exponential growth phase.
Staining for P-Glycoprotein
Cells grown on coverslips were washed with phosphate-buffered saline solution, fixed with acetone for thirty minutes at room temperature, and stained with the indirect immunofluorescence method. The primary monoclonal antibody to P-glycoprotein, C219 (five micrograms per milliliter; Centocor Diagnostics, Malvern, Pennsylvania), was applied for twenty hours at 4 degrees Celsius. The cells were washed five times with phosphate-buffered saline solution, incubated at room temperature for one hour with fluorescein isothiocyanate-conjugated F(ab')2 goat anti-mouse IgG (thirty-five micrograms per milliliter; Caltag Laboratories, San Francisco, California), and washed again with phosphate-buffered saline solution. The coverslips were mounted on glass slides with the cells facing the slides, and the immunofluorescence of P-glycoprotein was examined with a fluorescence microscope.
Assay for the Accumulation of Adriamycin
Cells grown on coverslips were incubated for one hour with Adriamycin (ten micrograms per milliliter), washed with phosphate-buffered saline solution, and fixed with buffered formalin for twenty minutes at room temperature. The coverslips were mounted on glass slides, and intracellular Adriamycin, which fluoresces red on excitation with green light, was measured by an epi-illumination cytofluorometer (Microphot with dual photometers; Nikon, Melville, New York).
Efflux of Adriamycin
Cells grown on coverslips were incubated for one hour with Adriamycin (ten micrograms per milliliter) and verapamil (three micrograms per milliliter), washed thoroughly with phosphate-buffered saline solution, and incubated for one hour with drug-free medium. The fluorescence of Adriamycin within each cell was measured with epi-illumination cytofluorometry.
Alkaline Phosphatase Activity
Cells were obtained by trypsinization; suspended in ten-millimolar Tris hydrochloride (pH 7.4) containing one-millimolar magnesium chloride, twenty-micromolar zinc chloride, 0.02 per cent (weight per volume) sodium azide, and 0.1 per cent Triton X-199; and sonicated for thirty seconds. The sonicates were centrifuged at 4 degrees Celsius for fifteen minutes at 8000 times gravity. Alkaline phosphatase activity was determined by the measurement of the release of p-nitrophenol from p-nitrophenylphosphatase on a spectrophotometer (410 nanometers at 37 degrees Celsius), according to the method of Lowry30. Protein was measured with a protein assay kit (BCA; Pierce, Rockford, Illinois). The results were expressed in micromoles per minute per milligram of protein.
Inoculation of the Cells into Mice
The cells were released by trypsinization and were resuspended in a small volume of medium, which was injected subcutaneously into C3H/Sed mice. The tumors were resected under sterile conditions. One-half of each tumor was cultured to study the accumulation of Adriamycin and to stain for P-glycoprotein, and the other half was processed for histological examination.
Multidrug-Resistant Phenotype
The multidrug-resistant phenotype of the cell lines was determined with the tetrazolium colorimetric assay described by Hansen et al.24. In this assay, cell growth or death is indicated by the conversion of the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide to the colored formazan, the concentration of which can be measured with spectrophotometry. Cells (2 x 103 in 100 microliters of medium) were seeded in ninety-six-well plates. After thirty-six hours, the culture medium was replaced with media containing a variety of concentrations of anticancer drugs. The cultures were incubated for an additional seventy-two hours. Twenty-five microliters of the tetrazolium reagent (Sigma Chemical, St. Louis, Missouri), at a concentration of five milligrams per milliliter in phosphate-buffered saline solution, was added to the cultures; after two hours, 100 microliters of the extraction buffer N,N-dimethylformamide and sodium dodecyl sulfate (Sigma Chemical), at pH 4.7, also was added. The cultures were incubated for twenty-four hours, and the optical densities at 570 nanometers were measured with use of a microplate reader (model 700; Cambridge Technology, Cambridge, Massachusetts). Inhibition of growth by anticancer drugs was calculated with the formula: cytostasis (%) = (1 - [A/B]) x 100, where A is the absorbance of the treated cells and B is the absorbance of the control cells. The concentrations of the drug that produced 50 per cent growth inhibition then were determined from the dose-response curves by plotting the cytostasis versus the concentration of the drug.
Reversal of Adriamycin Resistance by Resistance-Modifying Agents
We tested the effect of three different types of resistance-modifying agents: verapamil, a calcium channel blocker; trifluoperazine, a calmodulin inhibitor; and cyclosporin A, an immunosuppressor. First, growth inhibition by each of the resistance-modifying agents alone was examined, in order to determine the non-toxic doses of the drugs. Cells were incubated for seventy-two hours in various concentrations of each of the resistance-modifying agents, and the cytotoxic concentration for each agent was determined with use of the tetrazolium colorimetric assay. To assess the effect of each resistance-modifying agent, the concentration of Adriamycin producing 50 per cent growth inhibition was compared in the absence and presence of non-toxic doses.
Expression of P-Glycoprotein
As mentioned, immunofluorescence staining for P-glycoprotein was carried out with use of the murine monoclonal antibody C219 (Figs. 1-A, 1-B, and 1-C). Staining of the parent cell line by the antibody C219 was not detectable. Of fifteen subclones produced by single-cell culture after pulse treatment of the MOS cells with Adriamycin, two clones showed bright green fluorescence of P-glycoprotein; they were designated MOS/ADR1 and MOS/ADR2.
Accumulation of Adriamycin
After the cells had been incubated for one hour with Adriamycin (ten micrograms per milliliter), the nuclei of the parent (MOS) cells revealed strong red fluorescence. In contrast, very weak fluorescence was seen in most of the nuclei of the Adriamycin-resistant clones (MOS/ADR1 and MOS/ADR2) (Figs. 2-A, 2-B, and 2-C). To confirm these histological observations, we measured the intensity of the fluorecence of Adriamycin with cytofluorometry. The accumulation of Adriamycin was much lower in the two P-glycoprotein-positive cell lines than in the parent cell line (Fig. 3).
Efflux of Adriamycin
Preloading of the cells with Adriamycin (ten micrograms per milliliter) and verapamil (three micrograms per milliliter), followed by incubation in drug-free medium, revealed a pronounced decrease in intracellular Adriamycin in the MOS/ADR1 and MOS/ADR2 cells within thirty minutes after the extracellular Adriamycin had been removed (Fig. 4). This result indicated active extracellular transport of Adriamycin by these cells.
Osteoblastic Phenotype
The cellular alkaline phosphatase activity was approximately 2.4-fold and 2.0-fold greater, respectively, in the MOS/ADR1 and MOS/ADR2 cells than in the MOS cells (Fig. 5). Histological evaluation of the tumors that had resulted from the inoculation of MOS cells into mice showed a homogeneous population of cells with relatively immature osteoid (Fig. 6-A). In contrast, the tumors produced by the MOS/ADR1 cells showed a considerable amount of osteogenic activity, with abundant formation of bone and osteoid (Fig. 6-B). The tumors produced by the MOS/ADR2 cells also had a substantial amount of osteogenic activity (Fig. 6-C). The analysis of cells from the tumors produced by the inoculation of MOS/ADR1 or MOS/ADR2 cells into mice demonstrated positive staining for P-glycoprotein and reduced accumulation of Adriamycin, as expected (data not shown).
Spectrum of Cross Resistance
The sensitivity of the cell lines to anticancer drugs was evaluated with use of the tetrazolium colorimetric assay, and the concentrations of the drugs producing 50 per cent inhibition of cell growth were determined for each of the cell lines with linear regression analysis (Table I). The ratios of the concentrations determined for the resistant cell lines to those determined for the parent cell line were calculated (Fig. 7). The MOS/ADR2 cells were more than tenfold more resistant to Adriamycin and vincristine; fivefold to tenfold more resistant to vinblastine, dacarbazine, and etoposide; and twofold to fivefold more resistant to bleomycin, mitomycin C, actinomycin D, cisplatin, and carboplatin. The MOS/ADR2 cells, however, were not resistant to cytosine arabinoside, carmustine, cyclophosphamide, ifosfamide, methotrexate, and 5-fluorouracil. The MOS/ADR1 cells had a similar spectrum of cross resistance, except for their sensitivity to dacarbazine and cisplatin, but, for the most part, their resistance was lower than that of the MOS/ADR2 cells.
Reversal of Adriamycin Resistance by Resistance-Modifying Agents
As a first step, we examined growth inhibition by resistance-modifying agents alone in order to determine the non-toxic doses of these drugs. The concentrations that caused more than 10 per cent growth inhibition were ten micrograms per milliliter for verapamil, three micrograms per milliliter for trifluoperazine, and one microgram per milliliter for cyclosporin A (Fig. 8). Therefore, we used the non-toxic doses of the resistance-modifying agents (one and three micrograms per milliliter for verapamil, 0.3 and one microgram per milliliter for trifluoperazine, and 0.1 and 0.3 microgram per milliliter for cyclosporin A). Verapamil and trifluoperazine substantially reversed the resistance to Adriamycin in the MOS/ADR1 and MOS/ADR2 cell lines, but cyclosporin A was relatively ineffective (Fig. 9).
This study describes two newly developed murine osteosarcoma cell lines that were resistant to Adriamycin, MOS/ADR1 and MOS/ADR2, which demonstrate the classic multidrug-resistant phenotype in terms of overexpression of P-glycoprotein, increased efflux of drugs, and decreased intracellular accumulation of drugs. The cell lines were used to establish tumors in mice, making the lines useful as in vivo and in vitro models of drug-resistant osteosarcoma. The cell lines also have several other features of interest.
First, although most multidrug-resistant cell lines have been induced by continuous exposure of the parent cell line to a drug, we produced the multidrug-resistant cell lines by repetitive short-term exposure of the parent cell line to Adriamycin, which may have resulted in a drug resistance that is more similar to that encountered in a clinical setting. Consequently, the degree of drug resistance in the MOS/ADR1 and MOS/ADR2 cell lines was not as great as that in cell lines that were produced by long-term continuous exposure.
Second, both the MOS/ADR1 and the MOS/ADR2 cell lines were cloned from a single cell, which makes them suitable for genetic analysis of the multidrug-resistant phenotype. The cell lines were heterogeneous with respect to expression of P-glycoprotein (Figs. 1-A, 1-B and 1-C) and accumulation of Adriamycin (Figs. 2-A, 2-B and 2-C), which indicates that the drug-resistant phenotype may be unstable. This may be a useful model for the investigation of the mechanisms of spontaneous reversal of drug resistance.
The third observation is that both MOS/ADR1 and MOS/ADR2 cell lines displayed greater alkaline phosphatase activity in vitro and greater osteogenic activity in vivo than the parent cell line; this indicates that the development of drug resistance can be accompanied by changes in cellular differentiation. Such phenotypic alteration has also been shown in multidrug-resistant human neuroblastoma cell lines3,28.
Tumor cells are believed to become resistant by a variety of mechanisms, one of which is the P-glycoprotein pump mechanism that is encoded by the multidrug-resistant gene 1. Cells that overexpress P-glycoprotein are resistant to anthracyclines, vinca alkaloids, and podophyllotoxins but not to alkylating agents, antimetabolites, or cisplatin21. The MOS/ADR1 cell line displayed this classic multidrug-resistant phenotype. The MOS/ADR2 cell line demonstrated a different phenotype, with a high level of resistance to dacarbazine, one of the alkylating agents, and moderate resistance to cisplatin. Other drug-resistance mechanisms have been postulated, such as decreased influx of the drug to the cell41, reduced activity of topoisomerase II13,16, an increased level of glutathione S-transferase9,12 or metallothionein27, enhanced activity of DNA repair mechanisms17, altered intracellular distribution of the drug26, and altered nuclear import and export of the drug33. The demonstration of an atypical resistance pattern in MOS/ADR2 indicates that a cell line made from single-cell cloning can display both P-glycoprotein-mediated drug resistance and non-P-glycoprotein-mediated drug resistance. This observation is consistent with the concept that a tumor may express resistance by more than one mechanism.
In theory, the P-glycoprotein-related drug-resistant cells can be killed by anticancer drugs that are not exported from the cell by this mechanism. In the present study, both the MOS/ADR1 and the MOS/ADR2 cells were sensitive to cytosine arabinoside, carmustine, cyclophosphamide, ifosfamide, methotrexate, and 5-fluorouracil, many of which have known activity against human osteosarcoma. These in vitro findings are consistent with the observation that multidrug chemotherapy is more effective than single-drug treatment of human osteosarcoma. Prospective identification of P-glycoprotein-positive tumors may be helpful in the identification of specific agents that are likely to be active against a given tumor.
Alternatively, for patients who have multidrug-resistant osteosarcoma, management with resistance-modifying agents may enhance the effectiveness of chemotherapeutic agents exported by the P-glycoprotein mechanism. Resistance-modifying agents circumvent multidrug resistance by inhibiting the pump mechanism of P-glycoprotein, usually by competing with the chemotherapeutic agent for binding to P-glycoprotein. Of the three different types of resistance-modifying agents tested in this study, verapamil and trifluoperazine substantially potentiated the cytotoxicity of Adriamycin, but cyclosporin A did not exert a dramatic effect. All three drugs have been used in the clinical setting for certain multidrug-resistant cancers, in an attempt to enhance the effect of cytotoxic drugs1,15,31,32,34,36.
The usefulness of this approach remains to be proved, however, because of the toxicity of resistance-modifying agents and the effect they have on the protective role of P-glycoprotein in normal tissues. Recently, several new drugs have been developed that appear to be less toxic and more effective in reversing resistance related to P-glycoprotein4,37,38. Although this approach is promising, a final caveat is that these drugs enhance the cytotoxicity of anticancer drugs to normal tissue, as P-glycoprotein is present in such normal tissue as the adrenal cortex, renal proximal tubules, and blood-brain barrier14,19,39,40. Cell lines such as the ones described here may be useful in the further study of the efficacy of these agents.
The phenomenon of drug resistance is a major factor in the failure of current adjuvant drug protocols. We believe that the two cloned multidrug-resistant osteosarcoma cell lines may be useful for additional investigations of drug resistance in osteosarcoma. In addition to enabling investigators to study the effect of resistance-modifying agents, the cell lines are of value in studies regarding the genetic stability of the multidrug-resistant phenotype, the relationship between osteoblastic and multidrug-resistant phenotype, and the differences in metastatic and invasive potential between drug-resistant and drug-sensitive cells. They also offer a potential method with which to search for new, effective chemotherapeutic agents and to optimize the use of known drugs for the management of patients who have osteosarcoma.