Drug-Resistance Enables Selective Killing of Resistant Leukemia Cells: Exploiting of Drug Resistance Instead of Reversal MV Blagosklonny

Home , Drug resistance, Multiple drug resistance

Leukemia (1999) 13, 2031–2035  1999 Stockton Press All rights reserved 0887-6924/99 $15.00 http://www.stockton-press.co.uk/leu Drug-resistance enables selective killing of resistant leukemia cells: exploiting of drug resistance instead of reversal MV Blagosklonny

Medicine Branch, National Cancer Institute, Building 10, R 12N226, NIH, Bethesda, MD 20892, USA

Drug resistance is a well recognized problem in cancer therapy. drug, resistant cancer cells will continue to proliferate Despite the current dogma that drug resistance is always an whereas non-resistant cells will cease proliferation. Then cells obstacle for treatment, here I show that it provides opportunities for selective protection of non-resistant cells with killing can be exposed to a second drug which kills only proliferating of drug-resistant cancer cells. According to the proposed ‘two- cells. This second drug should not be subject to resistance (in drug’ strategy, the first drug should be ineffective against a the present work, non-substrate of Pgp or MRP). Drug-resistant target drug-resistant cell (ie the drug is a substrate of MRP or cells will then die, whereas non-resistant cells will be insensi- Pgp pumps). In addition, it must be cytostatic but not cytotoxic. tive to the second drug due to cell cycle arrest by the first The second drug, which is applied in sequence, must be a drug. Therefore, the second drug may selectively kill cells cycle-dependent apoptotic drug to which the target cell is not cross-resistant. Thus, low doses of adriamycin, etoposide and which are resistant to the first drug. actinomycin D, used as the first drugs, were cytostatic to par- Is this scenario possible? Or are there too many require- ental HL60 cells. Therefore, these drugs precluded Bcl-2/Raf-1 ments to be met by the drugs? I demonstrate that this situation phosphorylation, PARP cleavage and cell death which are is not only possible but can be achieved by using well-known otherwise induced by paclitaxel, a mitosis-selective apoptotic chemotherapeutic drugs which are widely used in the treat- drug for HL60 cells. In contrast, HL60/ADR cells which express ment of cancer. MRP, a transporter which pumps out the first drugs from a cell, were insensitive to the first drugs and therefore readily underwent apoptosis following the second drug. This strategy also allowed a selective killing of HL60/TX cells which express Materials and methods MDR-1, with the only difference being that the second drug, paclitaxel, was substituted for epothilones, non-Pgp sub- Cell lines and reagents strates. Lack of protection by the first drug, a Pgp substrate, resulted in HL60/TX killing by the second drug, whereas parental HL-60 cells were fully protected. Therefore, drug resistant HL60, a human leukemia cell line, was obtained from Amer- cells can be selectively killed by a combination of drugs not ican Type Culture Collection (Manassas, VA, USA). killing sensitive cells. Lack of toxicity against normal cells will HL60/ADR cells, a multidrug-resistant due to MRP clone of be clinically translated in reduction of adverse side-effects of HL-60 cells, were described previously.6 HL60/TX, a multi- chemotherapy against drug-resistant malignancies. drug-resistant due to MDR1 clone of HL-60 cells, were Keywords: chemotherapy; cancer; leukemia; resistance; cyto- obtained from Dr Bhalla (Emory University, Atlanta, GA, USA) toxicity and described previously.7,8 Vinblastine was obtained from the Development Therapeutics Program, NCI. Paclitaxel (Taxol), was a Bristol-Myers product. Epothilones A and B Introduction were provided by Dr F Lee (Bristol-Myers Squibb, Princeton, NJ, USA). The statement ‘the effect of anticancer therapy is limited by the development of drug resistance’, as a common introduction to multidrug resistance in cancer, is consistent with Immunoblot analysis tremendous attempts to reverse drug resistance. The most clear cut example is the multidrug resistance due to Cells were lysed and soluble proteins were harvested in TNES expression of the drug efflux pumps such as Pgp or MRP, buffer (50 mM TrisHCl pH 7.5, 100 mM NaCl, 2 mM EDTA, because of the highly resistant phenotypes and available 1mM sodium orthovanadate, 1% (v/v) NP40) containing pro- 1–3 approaches to circumvent it. For example, inhibitors of Pgp tease inhibitors (20 ␮g/ml aprotinin, 20 ␮g/ml leupeptin, 1 mM prevent efflux of chemotherapeutic drugs from a cell; and ran- PMSF). Proteins were resolved with 7.5% SDS-PAGE for 4,5 domized clinical trials are underway. The ultimate goal of detection of PARP and Raf-1 or with 12.5% SDS-PAGE for the conventional approach is complete reversal of multidrug detection of Bcl-2 as previously described.9 Immunoblotting resistance without toxicity of the reversal agent. However, for Raf-1 and Bcl-2 was performed using the following anti- even ‘non-resistant’ cancer cells are not easy targets because bodies: rabbit polyclonal anti-human Raf-1 (C12, Santa Cruz, there is no specific molecular target for anticancer drugs. Santa Cruz, CA, USA), and mouse monoclonal anti-human Anticancer drugs affect biological relevant molecules Bcl-2 (DAKO, Carpinteria, CA, USA) antibodies as pre- (ie DNA, microtubules, signal transduction kinases, viously described.9 oncoproteins) which are also essential for survival and/or proliferation of normal cells. Ironically, drug resistance may provide such selectivity. MTT assay One can envision that, following treatment with a cytostatic 15 000 cells were plated in 96-well flat bottom plates and then exposed to the pharmacological agents. At the indicated Correspondence: MV Blagosklonny; Fax: 301 402–0172 time (1–2 days), 20 ␮l of 5 mg/ml MTT solution in PBS was Received 5 July 1999; accepted 13 September 1999 added to each well for 4 h. After removal of the medium, Selective killing of resistant cancer cells MV Blagosklonny 2032 170 ␮l of DMSO was added to each well to dissolve the for- mazan crystals. The absorbance at 540 nm was determined using a Biokinetics plate reader (Bio-Tek Instruments, Winooski, VT, USA). Triplicate wells were assayed for each condition and standard deviations were determined.

DNA synthesis

DNA synthesis was monitored by 3H-thymidine incorporation. In brief, 2000 cells were plated in 96-well ﬂat bottom plates or 15 000 cells were plated in 24-well plates. The next day, cells were treated with drugs and incubated for 24 h. At the indicated time, cells were incubated with 1 ␮Ci methyl 3H-thymidine (Amersham, Piscataway, NJ, USA) for an additional 4–8 h and then acid-insoluble radioactivity was determined.

Results

Cytostatic doses of adriamycin prevent paclitaxel- induced apoptosis in HL60 cells

Low doses of topoisomerase inhibitors, including adriamycin, etoposide, actinomycin D, can inhibit proliferation of parental HL60 cells without cell death. Microtubule-active drugs at doses which induce mitotic arrest cause apoptosis in HL60 cells. Early biochemical events are shown in Figure 1. As expected, paclitaxel (PTX) induced Raf-1 and Bcl-2 hyperpho- sphorylation as an indicator of mitotic arrest and ensuing cell death. Cell death was preceded by the cleavage of Bcl-2 and PARP due to the activation of caspase-3,10 which is character- Figure 1 Abrogation of the biochemical events associated death istic of apoptosis. The cleavage of PARP was evidenced by of HL60 cells. (a) HL60 cells were pretreated with 50 ng/ml adriamy- disappearance of a 116 kDa band and the appearance of a cin (ADR) and then 50 ng/ml paclitaxel (PTX) were added for 18 h. Bcl-2 and Raf-1 phosphorylation, as well as Bcl-2 and PARP cleavage, truncated 85 kDa band (Figure 1). Chemotherapy-induced were evaluated by immunoblot as the markers of mitotic arrest and cleavage of Bcl-2 protein is associated with apoptosis in apoptosis. Note: as previously described,9 the appearance of a 22 kDa myeloid leukemia cells.9,11 This was followed by cell death Bcl-2 product following paclitaxel treatment could be only detected by 24–36 h, as evidenced by the inability to exclude trypan on grossly overexposed films. (b) Cells were treated as above, and blue (data not shown) and lack of metabolism (MTT assay). MTT assay was performed after 2 days. Dramatic decrease of metabolically alive cells (MTT assay), in a short-term 1–2 day assay, correlates with cell killing rather than growth arrest, lane 2 (PTX) vs lane 3 (ADR) in Figure 1b. and etoposide (VP16).12 While conferring resistance to the Pretreatment with cytostatic doses of adriamycin (ADR) Vinca compounds, MRP does not confer resistance to pacli- abrogated Raf-1 and Bcl-2 phosphorylation and significantly taxel,12 which, therefore, was chosen as a ‘second’ drug. reduced PARP cleavage, to the extent that the decrease in full- As expected, both adriamycin and paclitaxel inhibited 3H- length PARP was not apparent (Figure 1a). Accordingly, all thymidine incorporation (growth arrest) in parental HL60 cells HL60 cells were dead following exposure to PTX, whereas (Figure 3a) but only PTX killed cells at selected doses adriamycin-pretreated cells were alive (Figure 1b). Optimal (Figure 2b, c: HL60). Cytostatic doses of adriamycin prevented cytoprotection requires 12–16 h of pretreatment with adria- paclitaxel cytotoxicity and PARP cleavage in parental cells. In mycin before an addition of paclitaxel. contrast, adriamycin did not affect proliferation of HL/ADR In order to exert cytoprotection, concentrations of adriamy- (Figure 3a). Therefore, pretreatment with adriamycin did not cin should be sufficient to cause growth arrest but not be cyto- abrogate paclitaxel-induced death of HL/ADR cells toxic. Thus, doses of adriamycin below 5 ng/ml (which did (Figure 3b,c). not effectively block proliferation) did not protect against Thus, although HL/ADR cells are resistant to many drugs, paclitaxel. On the other hand, although doses of adriamycin they undergo apoptosis with the rational ‘two-drug’ approach higher than 200 ng/ml can abrogate PTX effects, they induced whereas ‘sensitive’ parental cells were fully protected. cytotoxicity comparable with paclitaxel and therefore by no means could be beneficial (Figure 2). More examples of the first drugs

Selective killing of HL/ADR cells Success of the ‘two-drug’ strategy relied on the effective cytoprotection of ‘sensitive’ cells. This cytoprotection is due to HL/ADR, a multidrug resistant clone of HL60 cells, expresses cytostatic effect without cytotoxic one. Higher cytoprotection MRP and therefore resistant to adriamycin, actinomycin D, is shifted to lower doses of adriamycin with an increase of Selective killing of resistant cancer cells MV Blagosklonny 2033

Figure 2 Dose-dependence of cytoprotection by adriamycin. HL60 cells were treated with indicated concentrations of adriamycin. After 16 h, medium or 100 ng/ml paclitaxel were added for an additional 30 h and then MTT assay was performed. Results represent percent of control (untreated cells). Figure 3 Comparison of parental HL60 and HL/ADR cells. HL60 cells and HL/ADR cells were either (1) untreated, (2) treated with 100 ng/ml paclitaxel (ADR), (3) pretreated with 50 ng/ml adriamycin time (data not shown) underscoring that, while protective drug (ADR) for 16 h, (4) pretreated with 50 ng/ml adriamycin and then treated with 100 ng/ml paclitaxel. (a) 3H-thymidine (3H-Td) was should not be cytotoxic, adriamycin at cytostatic doses is not added 20 h after of PTX. 3H-Td incorporation was performed as completely free from toxicity. described in Methods. (b) Cell survival was measured 48 h after PTX, Actinomycin D is more toxic than adriamycin, and cytopro- by MTT assay as described in Methods. (c) PARP cleavage was tection was observed at doses 1–4 ng/ml (data not shown). measured after 16 h by immunoblot as described in Methods. VP16 demonstrated cytoprotective properties with a maximal effect at 400 ng/ml (Figure 4a), whereas higher doses of VP16 were too cytotoxic. As with too low doses, too high doses of marginally resistant to epothilones. Accordingly pretreatment VP16 are ineffective. Importantly, HL/ADR cells were resistant with adriamycin, while protecting HL60 cells, did not protect to VP16, which result in lack of any protection of HL/ADR either HL/ADR or HL/TX against epothilone A (data not cells against paclitaxel cytotoxicity (Figure 4b). shown) and epothilone B (Figure 6). Emphasizing the importance of static vs toxic effects of the ‘first’ drug, neither UCN-01 nor lovastatin provided any protection because both drugs induced apoptosis at the same Discussion doses as they arrest growth of HL60 cells (data not shown). The demonstration that drug resistance can be used for selective elimination of resistant leukemia cells reveals alternative Other microtubule-active drugs as the second agents strategies in cancer treatment. This approach requires at least in paclitaxel-resistant HL/TX cells two drugs acting in sequence. First, a merely cytostatic (not cytotoxic) drug to which a target cell is resistant. Opposite to Using paclitaxel for selective killing HL/ADR cells was poss- that, a second drug to which the target cell is not cross-resist- ible because these cells express MRP which confers resistance ant is cytotoxic in a cell-cycle dependent manner. While the to the protective drugs but not to paclitaxel. However, a well first drug protects normal cells, the second drug eliminates characterized resistance mechanism in acute myeloid leuke- drug-resistant cells. This strategy is not limited to the resist- mia is mediated by MDR-1/Pgp,13 which confers resistance to ance due to drug transporters (used here as a model) but is paclitaxel. Thus, HL/TX cells, a clone of HL60 cells selected universal regardless of the particular mechanism of resistance for the paclitaxel resistance, expresses Pgp.8 Although these to the first drug. cells are also resistant to the cytoprotective (first) drug, resist- The drugs used in this study are currently employed in the ance to paclitaxel precludes any advantage (Figure 5). How- treatment of cancer. Therefore, one does not need to test their ever, like paclitaxel was effective in vincristine-resistant effectiveness in animal models or humans. For example, HL/ADR cells, another group of microtubule-active drugs can administration of very low doses of adriamycin 16 h prior to substitute for paclitaxel in paclitaxel-resistant HL/TX cells. In paclitaxel can be suggested as a clinical trial for the treatment agreement with prior demonstrations that epothilones are of the MRP-positive leukemia. The strategy is primary aimed effective against Pgp-expressing cells,14 HL/TX cells were only not to increase the effectiveness of the drugs but to protect Selective killing of resistant cancer cells MV Blagosklonny 2034

Figure 5 The ‘second drug’ failure in HL/TX cells. Cells were pretreated with 60 ng/ml adriamycin for 16 h (closed symbols) or left without pretreatment (open symbols). Then cells were treated with paclitaxel for 36 h and MTT assay was performed. Results represent percent of control (untreated cells).

Figure 4 VP16 as a ﬁrst drug in the ‘two-drug’ treatment. HL60 (a) or HL/ADR (b) cells were pretreated with indicated concentrations of (etoposide) VP16. After 16 h, medium or 100 ng/ml paclitaxel were added for an additional 24 h, and then MTT assay was performed. Results represent percent of control (untreated cells).

normal cells. This, in turn, can increase effectiveness of the treatment because a cure for cancer is limited by drug toxicity in a patient. Solving the dilemma between too low doses vs too high doses of the first drug, too high doses must be Figure 6 Epothilone B as a second drug. HL60, HL/ADR, or HL/TX avoided because they might protect resistant cells. cells were pretreated with 60 ng/ml adriamycin for 16 h and then were As a disadvantage of the clinically available drugs, the first treated with epothilone B (Epo B) for 36 h and MTT assay was perfor- drugs were not designed for a mere cytostatic purpose. In cell med. Results represent percent of control (untreated cells). culture, the cytostatic effects of these drugs (ie adriamycin) are hardly reversible. For example, although cytostatic doses of adriamycin protect parental cells from the second drug, clon- are directed to the discovery of such drugs and some of them ogenic assay cannot be performed because non-proliferating are already available. Thus, paclitaxel can serve as a cells cannot form colonies. This underscores that the identifi- ‘second/eliminating drug’ in HL/ADR cells, whereas epothi- cation of reversible pure cytostatic compounds is necessary. lones are good substitutions in HL60/TX cells. Using the HL60 What problems must be solved for success of this strategy? leukemia cell model, we certainly have an advantage that As we discussed, the second drug: (1) must induce rapid these cells readily undergo apoptosis after 16–24 h of death; (2) of cycling cells; and (3) be active in target cells exposure to microtubule-active drugs. Unfortunately, various which are resistant to the first drugs. Current research efforts cancer cells do not share this ability with leukemia cells. If Selective killing of resistant cancer cells MV Blagosklonny 2035 the ‘second’ drugs induced slow cell death, the efficacy of the 3 Zhou DC, Zittoun R, Marie JP. Expression of multidrug resistance- ‘two-drug’ approach may be diluted. Although much work associated protein (MRP) and multidrug resistance (MDR1) genes still lies ahead, one can expect that the list of apoptotic and in acute myeloid leukemia. Leukemia 1995; 9: 1661–1666. 4 Bates SE, Wilson WH, Fojo AT, Alvarez M, Zhan Z, Regis J, Robey cycle-dependent drugs will expand. R, Hose C, Monks A, Kang YK, Chabner B. Clinical reversal of Additional limitations include the expression of Pgp on multidrug resistance. Stem Cells 1996; 14: 56–63. some normal cells such as hematopoietic stem cells.15 The 5 Sikic BI, Fisher GA, Lum BL, Halsey J, Beketic-Oreskovic L, Chen two-drug strategy will not offer the protection to these parti- G. Modulation and prevention of multidrug resistance by inhibi- cular normal cells if target cancer cells also express Pgp, tors of P-glycoprotein. Cancer Chemother Pharmacol 1997; 40 although there would be no problems if mechanisms of resist- (Suppl): S13–S19. 6 Blagosklonny MV, Alvarez M, Fojo A, Neckers LM. Bcl-2 protein ance in cancer cells are different. This problem should not be downregulation is not required for differentiation of multidrug overestimated because the same question could be re- resistant HL60 leukemia cells. Leukemia Res 1996; 20: 101–107. addressed to the conventional approach to reverse drug resist- 7 Bhalla K, Huang Y, Self S, Ray S, Mahoney ME, Ponnathpur V, ance by Pgp blockers: what about normal cells with Pgp? Tourkina E, Ibrado AM, Bullock G, Willingham MC. Characteriz- Desirable properties of the ‘first drugs’ are not only opposite ation of a human myeloid-leukemia cell line highly resistant to to the ‘second drugs’, but also opposes what is considered as taxol. Leukemia 1994; 8: 465–475. 16–18 8 Huang Y, Ibrado AM, Reed JC, Bullock G, Ray S, Tang C, Bhalla ‘good’ apoptosis inducing anticancer drugs. The first drug K. Co-expression of several molecular mechanisms of multidrug should be merely cytostatic to normal cells and completely resistance and their significance for paclitaxel cytotoxicity in ineffective against resistant cells due to the drug resistance. human AML HL-60 cells. Leukemia 1997; 11: 253–257. Although adriamycin or etoposide (VP16) were used here in 9 Blagosklonny MV, Chuman Y, Bergan RC, Fojo T. Mitogen- in vitro experiments, a window of cytostatic but not toxic activated protein kinase pathway is dispensable for microtubule- effect is narrow and, even at cytostatic doses, a long-term active drug-induced Raf-1/Bcl-2 phosphorylation and apoptosis in leukemia cells. Leukemia 1999; 13: 1028–1036. effect may decrease cell survival. Identification of merely 10 Cheng EH, Kirsch DG, Clem RJ, Ravi R, Kastan MB, Bedi A, Ueno cytostatic agents may be a cornerstone for the development of K, Hardwick JM. Conversion of Bcl-2 to a Bax-like death effector the ‘two-drug’ approach. While multidrug resistance provides by caspases. Science 1997; 278: 1966–1968. numerous challenges, it also provides opportunities to achieve 11 Fadeel B, Hassan Z, Hellstr E, Lindberg M, Henter J-I, Orrenius S, a high therapeutic index by maximizing the desired effect, and Zhivotovsky B. Cleavage of bcl-2 is an early event in chemo- minimizing the undesired side-effects. therapy-induced apoptosis of human myeloid leukemia cells. Leu- kemia 1999; 13: 719–728. 12 Zaman GJR, Flens MJ, van Leusden MR, de Haas M, Mulder HS, Lankelma J, Pinedo HM, Scheeper RJ, Baas F, Broxterman HJ, Borst P. The human multidrug resistance-associated protein MRP Acknowledgements is a plasma membrane drug-efflux pump. Proc Natl Acad Sci USA 1994; 91: 8822–8826. I thank Drs Tito Fojo (NCI, NIH) and WD Figg (NCI, NIH) for 13 Legrand O, Zittoun R, Marie JP. Role of MRP1 in multidrug resist- their generous support. I thank Dr Bhalla (Emory University, ance in acute myeloid leukemia. Leukemia 1999; 13: 578–584. Atlanta) for providing HL/TX cells. I thank Dr F Lee (Bristol- 14 Chou TC, Zhang XG, Harris CR, Kuduk SD, Balog A, Savin KA, Bertino JR, Danishefsky SJ. Desoxyepothilone B is curative against Myers Squibb, Princeton) for providing epothilones A and B. human tumor xenografts that are refractory to paclitaxel. Proc Natl Acad Sci USA 1998; 95: 15798–15802. 15 Chaudhary PM, Roninson IB. Expression and activity of P-glycoprotein, a multidrug efflux pump, in human hematopoietic stem References cells. Cell 1991; 66: 85–94. 16 Fisher DE. Apoptosis in cancer therapy: crossing the threshold. 1 Fojo AT, Shen DW, Mickley LA, Pastan I, Gottesman MM. Intrinsic Cell 1994; 78: 539–542. drug resistance in human kidney cancer is associated with 17 Martin SJ, Green DR. Apoptosis as a goal of cancer therapy. Curr expression of a human multidrug-resistance gene. J Clin Oncol Opin Oncol 1994; 6: 616–621. 1987; 5: 1922–1927. 18 Kastan MB, Canman CE, Leonard CJ. P53, cell cycle control and 2 Pastan I, Gottesman M. Multiple-drug resistance in human cancer. apoptosis: implications for cancer. Cancer Metast Rev 1995; 14: New Engl J Med 1987; 316: 1388–1393. 3–15.