Arsenic trioxide sensitivity is associated with low level of glutathione in cancer cells

Arsenic trioxide (As2O3) is a novel anticancer agent, which has been found to induce remission in acute promyelocytic leukaemic patients following daily intravenous administration. The therapeutic value of As2O3 in other cancers is still largely unknown. Cytotoxic tests in a panel of cancer cell lines showed that bladder cancer, acute promyelocytic leukaemic and gastrointestinal cancer cells were the most sensitive to As2O3 among 17 cell lines tested. Cellular glutathione (GSH) system plays an important role in arsenic detoxification in mammalian cells. Cancer cells that were intrinsically sensitive to As2O3 contained lower levels of GSH, whereas resistant cancer cells contained higher levels of GSH. On the other hand, there was no association of glutathione-S-transferase-π or multidrug resistance-associated protein 1 levels with arsenic sensitivity in these cancer cells. Multidrug-resistant cancer cells that were cross-resistant to arsenic contained higher levels of GSH or multidrug-resistance-associated protein 1 than their drug-sensitive parental cells. Cancer cells become more sensitive to arsenic after depletion of cellular GSH with L-buthionine sulphoximine. We concluded that cellular GSH level is the most important determinant of arsenic sensitivity in cancer cells. Cellular GSH level and its modulation by buthionine sulphoximine should be considered in designing clinical trials using arsenic in solid tumours. © 1999 Cancer Research Campaign


Cellular GSH content
Cellular GSH content was determined by Bioxytech GSH-400 colourimetric assay kit (Oxis International, Portland, OR, USA). Cells (10 6 -10 7 ) were trypsinized, centrifuged and washed with phosphate-buffered solution. Cells were then re-suspended in 500 µl of ice-cold metaphosphoric acid. After homogenization, the solution was centrifuged at 3000 g, 4°C for 10 min. The clear supernatant was collected at 4°C for further assay. Reagent R1 and sodium hydroxide were added to the solution. After incubation at 25°C for 10 min in the dark, the absorbance of the solution was measured at 400 nm. GSH concentrations in the solution were calculated from the absorbance. Cellular GSH content is expressed as µg of GSH mg -1 of protein.

Western blot of GST-π and MRP1
Total protein of the cells was separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidine difluoride (PVDF) membrane. The membrane was blocked in 5% skim milk in Tris-buffered saline, containing 0.15% Tween for 1 h before washing three times in the same solution containing 0.01% Tween. The membrane was then incubated with a 1:1000 dilution of polyclonal antibody against human GST-π (Medical and Biological Laboratory, Nagoya, Japan) for 1 h. Membrane was washed and incubated with HRPconjugated secondary antibody. The immunolabelled protein was detected using a chemiluminescence kit (NEN Life Science, Boston, MA, USA).
Membrane protein was isolated from cancer cells. MRP1 levels were measured by Western blot using 1:200 dilution of monoclonal antibody MRPm6 (Sanbio, Uden, The Netherlands) as described previously (Flens et al, 1994). The immunolabelled protein was visualized with a chemiluminescence kit (NEN Life Science, Boston, MA, USA).

Cytotoxicity of As 2 O 3 in cancer cells
Concentrations of As 2 O 3 that inhibit 50% of cell growth (IC 50 s) are listed in Table 1. Several bladder cancer (NTUB1, BFTC905, T24 and HTB-9) and gastrointestinal cancer cell lines (SW620 and AGS) were relatively sensitive to As 2 O 3 in addition to acute promyelocytic leukaemic cells (NB4). Table 1. To correlate GSH contents to IC 50 s of nine tested cell lines, Spearman's rho correlative coefficient was 0.661 (P = 0.026, one-tail). Five cell lines that were intrinsically sensitive to arsenic (IC 50 s < 1.5 µM) all contained a low level of GSH (GSH < 10 µg mg -1 protein), whereas four cell lines that were intrinsically resistant to arsenic (IC 50 s > 1.5 µM) all contained a high level of GSH (GSH > 10 µg mg -1 protein).

GST-π protein expression in cancer cells
Overexpression of GST-π may facilitate conjugation of trivalent arsenic to GSH. GST-π protein expression in the cancer cells was measured by Western blot as shown in Figure 1. Arsenic-sensitive NB4 cells contained very low levels of GST-π protein. However, several arsenic-sensitive cells, such as BFTC905 and SW620 cells, expressed high levels of GST-π protein, whereas arsenic-resistant H460, Hep3B and BFTC909 cells expressed low levels of GST-π protein. There was no correlation of GST-π levels to As 2 O 3 IC 50 s in cancer cells. Multidrug-resistant NTU-B1/P14 cells overexpressed GST-π compared to their drug-sensitive parental NTU-B1 cells.

MRP1 expression in cancer cells
MRP1 may facilitate export of conjugated GSH out of the cells (Rappa et al, 1997) and thus, may affect arsenic resistance in cancer cells. MRP1 expression in the membrane protein of cancer cells was measured by Western blot as shown in Figure 1. H460, BFTC909 and one multidrug-resistant MCF7/VP cell contained measurable levels of MRP1, whereas MRP1 expression was very low in other cancer cells. MRP1 seemed to confer resistance to arsenic; however, not all arsenic-resistant cancer cells expressed high levels of MRP1.

Cross-resistance of arsenic in multidrug-resistant cancer cells
IC 50 s of two multidrug-resistant cancer cells are listed in Table 1. Cisplatin-resistant NTU-B1/P14 was 5.5-fold resistant to arsenic. Etoposide-resistant MCF7/VP cells were 4.8-fold resistant to arsenic. Glutathione content of NTU-B1/P14 was 6.7-fold higher than that of NTU-B1 cells. On the other hand, there was no difference of GSH content between MCF7/WT and MCF7/VP cells. MCF7/VP expressed high levels of MRP1, which may account for its arsenic resistance, whereas NTU-B1/P14 expressed no measurable level of MRP1 ( Figure 1).

Modulation of GSH content in cancer cells by BSO
BSO is known to deplete cellular GSH via inhibition of γ-glutamylcysteine synthetase, which is required for GSH biosynthesis.
NTU-B1, NTU-B1/P14, MCF7/WT and MCF7/VP cells were incubated with various concentrations of As 2 O 3 and 10 µM of BSO for 4 days. Ten micromolars of BSO were not toxic to these cancer cells (IC 10 s of BSO in NTU-B1, NTU-B1/P14, MCF7/WT and MCF7/VP cells were 37 µM, > 50 µM, 27 µM and 24 µM respectively). The representative cytotoxicity curves of NTU-B1 and NTU-B1/P14 cells in As 2 O 3 with or without co-incubation with 10 µM of BSO are shown in Figure 2. IC 50 s of As 2 O 3 and GSH contents in BSO-treated treated GSH-depleted cells (drug-sensitive and -resistant NTU-B1 and MCF7/WT cells) are shown in Table 1. All four cancer cells became very sensitive to arsenic (IC 50 s 0.1 µM to 0.4 µM) when GSH was depleted by BSO.

DISCUSSION
Arsenic has been used widely for a long time in both Western and Chinese medicine. The definitive role of arsenic as an anticancer agent was not clear until a recent report for treatment of acute promyelocytic leukaemia (Shen et al, 1997). As 2 O 3 , given as a daily 10 mg intravenous infusion seemed to be an effective and tolerable regimen for refractory acute promyelocytic leukaemic patients. Induction of partial differentiation or induction of apoptosis have been proposed as the primary mechanism of cytotoxicity of arsenic to acute promyelocytic leukaemic cells . Due to the unique mechanism of action of arsenic to cancer cells, the attempt to use arsenic in malignancies other than acute promyelocytic leukaemic patients is clearly warranted (Gallagher, 1998).
In this study, bladder cancer cells NTU-B1 and BFTC905 were most susceptible to As 2 O 3 . Apoptosis can be induced in NTU-B1 cells at 1 µM As 2 O 3 (data not shown). IC 50 s of the bladder cancer cell lines was substantially lower than reported in As 2 O 3 peak plasma levels (4-6 µM) in patients (Shen et al, 1997). Thus, it is conceivable that As 2 O 3 may be effective in the treatment of bladder cancer and other solid tumours that show similar sensitivity to arsenic.
Multiple mechanisms account for arsenic resistance in bacteria and mammalian cells. Trivalent arsenic was shown to directly react to reduced GSH in solution (Scott et al, 1993). The cellular  arsenic content was reduced by GSH pretreatment and increased in BSO-treated Chinese hamster ovary cells (Huang et al, 1993). It is conceivable that cellular GSH content affects sensitivity of cancer cells to As 2 O 3 . In this study, we have shown clearly that cancer cells that contained low levels of GSH were all sensitive to arsenic exposure, whereas resistant cancer cells such as lung and liver cancer cells contained the highest amounts of GSH. Furthermore, sensitivity of NTU-B1 and MCF7 cells to arsenic can be increased when GSH was depleted by pretreatment with 10 µM BSO. Cisplatin-resistant NTU-B1/P14 cells were cross-resistant to arsenic, the GSH content of these cells was higher than in steadystate parental cells. When GSH was depleted by BSO, resistant cells became sensitive to arsenic treatment. BSO may also enhance arsenic toxicity in wild-type MCF7 cells and multidrug-resistant and MCF7/VP cells. When cellular GSH was depleted, all drugsensitive and multidrug-resistant cancer cells became very sensitive to arsenic. GST-π overexpression was noted in a Chinese hamster ovary cells resistant to As 2 O 3 (CHO/SA7) (Lo et al, 1992). It is conceivable that GST-π protein levels may also affect intrinsic sensitivity to arsenic. The GST-π level seems to play very little role, however, in the determination of arsenic sensitivity of cancer cells in this study.
MRP1-transfected HeLa cells were resistant to several heavy metal anions, including trivalent arsenic (Cole et al, 1994). In this study, multidrug-resistant cancer cells such as MCF7/VP and NTU-B1/P14 cells were cross-resistant to As 2 O 3 . MRP1 was overexpressed in several arsenic-resistant cancer cells such as BFTC909, H460 and MCF7/VP cells. However, MRP1 was not expressed in meaningful amounts in intrinsically resistant MCF7, HepB3 cells or acquired resistant NTU-B1/P14 cells. We demonstrated complete reversal of arsenic resistance in MRP1-overexpressing MCF7/VP cells when GSH was depleted by BSO. The result suggests that MRP1 overexpression may not protect cancer cells from arsenic toxicity when GSH was depleted. Overexpression of MRP1 may contribute to arsenic resistance, but MRP1 expression is not the main determinant of arsenic sensitivity in cancer cells.
Use of As 2 O 3 in solid tumour clinical trials is clearly warranted. Our study suggests that GSH content in tumour cells may be the main determinant of arsenic sensitivity. Attempts should be made to measure tumour GSH content and correlate to arsenic response in clinical trials. BSO was used to deplete GSH and enhance chemosensitivity of alkylating agents (Bailey et al, 1994). The peak plasma level of BSO in patients (4-6 mM) was much higher than levels needed to enhance arsenic toxicity in arsenic resistant cancer cells (10 µM). Therefore, adding BSO to arsenic treatment may potentially be useful to reverse acquired arsenic resistance in acute promyelocytic leukaemic patients or to treat tumours that are intrinsically resistant to arsenic.
In conclusion, GSH content correlates well with arsenic resistance in cancer cells. Depletion of cellular GSH by BSO enhanced arsenic toxicity in both arsenic-sensitive and -resistant cancer cells. Further animal studies and human trials evaluating arsenic as an anticancer drug are warranted. Our study suggests that As 2 O 3 should be tested in solid cancers, especially patients with bladder and gastrointestinal cancer. This study suggests that measurement and modulation of cellular GSH content in cancer cells should be deployed in designing future clinical trials.