Synergistic cytotoxicity of bcl-2 antisense oligodeoxynucleotides and etoposide, doxorubicin and cisplatin on small-cell lung cancer cell lines.

Expression of Bcl-2 is life-sustaining for small-cell lung cancer cells and associated with drug resistance. In the present study, the interactions between the bcl-2 antisense oligodeoxynucleotide 2009 and the chemotherapeutic agents etoposide, doxorubicin and cisplatin were investigated on small-cell lung cancer cell lines to search for synergistic combinations. The cell lines NCI-H69, SW2 and NCI-H82 express high, intermediate-high and low basal levels of Bcl-2, respectively, which are inversely correlated with the sensitivities of the cell lines to treatment with oligodeoxynucleotide 2009 and the chemotherapeutic agents alone. Moreover, differences were found in the responsiveness of the cell lines to treatment with combinations of oligodeoxynucleotide 2009 and the chemotherapeutic agents. In the cell lines NCI-H69 and SW2, all combinations resulted in synergistic cytotoxicity. In NCI-H69 cells, maximum synergy with a combination index of 0.2 was achieved with the combination of oligodeoxynucleotide 2009 and etoposide. In SW2 cells, the combination of oligodeoxynucleotide 2009 and doxorubicin was the most effective (combination index = 0.5). In the cell line NCI-H82, which expresses a low basal level of Bcl-2, most of the combinations were slightly antagonistic. Our data suggest the use of oligodeoxynucleotide 2009 in combination with chemotherapy for the treatment of small-cell lung cancer that overexpresses Bcl-2.

relies on different classes of chemotherapeutic agents. including epipodophyllotoxins. anthracyclines and platinum analogtues. Although the introduction of combination chemotherapy as the principal form of treatment has led to an increase in median sunrival. only a small proportion of patients with SCLC are cured (Souhami and Las-. 1990).
Mechanisms of drucg resistance in solid tumours haxe been examined extensixel over the last 10 years. In lung cancer cell lines selected in vitro grenetic chanoes have been identified that alter drug transport and activity. such as overexpression of the Pglycoprotein or the multidrug resistance-associated protein (MRP) (Dovle. 1993: Gonzalez Manzano et al. 1996: Versantxoort et al. 1996. However. no correlation between overexpression of these drug transporters in lung, cancer cells and response to therapy has been found in patients (Lai et al. 1989) suggesting that other mechanisms are more important for chnical drug resistance of lung cancer. It is now generally accepted that resistance to cytotoxic treatments relates to failure of cells to engyage the process of apoptosis. and some of the genetic defects that antagonize apoptosis have alreadv been unravelled. Most SCLC cell lines and tumour tissues overexpress the Bcl-2 oncoprotein (Ben Ezra et al. Modulation of gene expression by antisense oligodeoxynucleotides (ODNs) is a promising approach because of its target specificity and potential applicability to any sequenced gene The mechanisms implicated in the action of antisense ODNs relate to RNase H-mediated hydrolysis of the target mRNA or to translational arrest anrsing from steric hindrance by the RNA-DNA heteroduplex (Stein andCheng. 1993: Ho andParkinson. 1997). These mechanisms differ greatly from those exerted by chemotherapeutic arents and thus might operate also in drum-resistant tumour cells. Antisense ODNs targeting the first six codons of the bcl-2 mRNA have been show-n to increase the sensitivity of lymphoma cells to chemotherapeutic acents (Kitada et al. 1994).
However. althouth the combination of bcl-2 antisense ODNs and chemotherapy seems to be appealing. the cytotoxic interaction of these treatments on tumour cells has not yet been addressed in detail. Recently. we has-e identified an antisense sequence (ODN 2009) targeting the bcl-2 coding region that effectixelv downregulated bcl-2 expression and induced apoptosis in SCLC cells (Ziegler et al. 1997). In the present study. we examined the cytotoxic effects of ODN 2009 in combination with etoposide. doxorubicin and cisplatin on SCLC cell lines with different Bcl-2 levels.
Analysis of potential synergy w-as performed by the median effect method described by Chou and Talalay ( 1984).
Delivery of ODNs to SCLC cells ODN 2009 and the control ODN sc-21 were delivered to cells in the form of complexes w-ith the cationic lipid N-[ 1-(2.3dioleoy loxy )propx l]-AN.N.V-trimethv-lammonium methylsulphate (DDOTAP: Boehrinrer MIannheim. Germany) essentiallx as described prexiously (Ziegler et al. 1997). Briefly. equal volumes of ODNs (24 jm) and DOTAP (400 Im) in 20nmm HEPESbuffered saline were mixed and allowed to complex for 10 min at room temperature. The mixtures w ere diluted into nine volumes of cell culture medium to achieve solutions of 1.2 jmI ODN. For use in experiments. ODN stock solutions were further diluted serially into cell suspensions containing medium alone or medium with various dilutions of the chemotherapeutic agents.

Western blot analysis
Western blottingxwas performed as described by Ziegler et al (1997). Briefly. 1 ml of cells/ODN mixture wxas plated in a 24-xxell plate and incubated for 24 h at 37TC. ODN concentrations and cell densities were as above. Ten iga of soluble protein extract per sample was separated on a 12% polvacrxlamide sodium dodecy l sulphate (SDS) cgel at 150 V for about 3 h. and transfer to a polyvinylidene fluoride membrane (Immobilon-P. Millipore) was performed in a semi-drx blotting chamber (Schleicher and Schuell) at 1 mA cm-for 1 h. The blots x-ere blocked in Trisbuffered saline (TBS) containino 5%'7 bovine serum and 5%7c nonfat drx milk. and then incubated oxernight at 4 C with mouse anti-human Bcl-2 monoclonal antibodx (Dako Diagnostics. Glostrup. Denmark). To detect the primarx antibody. blots wxere incubated wkith a rabbit antimouse immunoglobulin peroxidase conjugate (Sigma Chemical. St. Louis. MO. USA) for 2 h at room temperature. Visualization of the immunocomplex xx-as performed by enhanced chemiluminescence using the ECL kit (Amersham). followed by exposure to radiographic films (Fuji RX) for different time periods dependent on the cell line. Relative protein levels were quantified after scanning of the films using a flat bed scanner (Hewlett Packard ScanJet Hcx) and the ImageQuant software (Molecular Dv namics.) Chemotherapeutic agents Doxorubicin (Adnrblastine) was obtained from Farmitalia Carlo Erba (Zug. Switzerland). Etoposide (Vepesid) and cisplatin (Platinol) were obtained from Bristol-Myers Squibb (Baar. Switzerland). All agents were clinical grade and diluted with cell culture medium before use.

Measurement of cell viability
The cytotoxic effects of ODN 2009 and chemotherapeutic agents on SCLC cell lines were determined by use of the colorimetric WST-1Iiabilitx assav as described previously (Ziegler et al. 1997). Once the optimal growth conditions have been established for each cell line. the WST-1 viabilitv assav provides reproducible results which correlate well with the actual number of X iable cells determined by propidium iodide exclusion (Ziegler et al. 1997). Briefly.
for each experiment. 100 gl each of cell suspensions containing ODN 2009 or control ODNs. the chemotherapeutic agents. or a combination of both types of agents were plated in triplicates in 96well plates. Cell densities were 0.5 x 105 cells ml (SW2. NCI-H82). or I0cells ml' (NCI -H69). Cells were incubated for4 days at 37 C. and then 10 gl of WST-1 reagrent (Boehringer Mannheim. Germanv) was added per well and allowed to react for 3-5 h at 37"C. Absorbance at 450 nm was measured by use of an enzx-melinked immunosorbent assav reader (2550 EIA reader. Bio Rad Laboratories. Hercules. CA. USA).

Determination of viable cell numbers based on propidium iodide exclusion
To determine the number of x-iable cells after cytotoxic treatment. 4 ml of cell suspensions containing ODN 2009. chemotherapeutic agents or a combination of both types of agents were plated in sixwell plates. The same cell densities as for the WST-1 viabilitx assay described above w ere used. At different time points of incubation. cells were harnested. briefly trvpsinized and resuspended in phosphate-buffered saline (PBS). Immediately before measurement. propidium iodide was added to a final concentration of 1.25 jge ml . The number of cells in the cultures w as quantitated at a constant flow rate of 12 ' a mmn-I by use of a FACSCalibur cvtofluorometer (Becton Dickinson. Mountain View. CA. USA). Only' cells that excluded propidium iodide w ere considered Xiable. Apoptotic cell death after cvtotoxic treatment was confirmed by forward and side light scatter analy-sis as described prexviouslV (Cotter et al. 1992: Ziegler et al. 1997).
Analysis of combined ODN/drug effects The median-effect method described by Chou and Talalay (1984)  Ten micrograms of soluble protein were analysed per sample and Westem blotting was performed as described in the Materials and methods section. Blots were exposed to radiographic films for 5 min (NCI-H69 and SW2 cells). or 40 min (NCI-H82 cells) 2009 and the chemotherapeutic agents. This analysis is based on the median-effect principle of the mass action law and relies on linear regression as a well-accepted statistical approach. In each experiment, cells were treated with serial dilutions of ODN 2009 and drugs individually. and with fixed ratios of ODN 2009 and drugs simultaneously at doses in the range of the indi-idual concentrations at w hich cell viabilitv was inhibited bN 50% (IC*).
The fraction affected (f) was calculated by dividing the per cent v-iabilits in ODN 2009 and drug-treated wells by the viability in untreated wells. and data were analysed by the median-effect method (Chou and Talalay. 1984 by a CI less than 1. additix itv by a CI equal to 1. and antaconism by a Cl greater than 1. For each combination. CI-values w-ere calculated based on the assumption that drug interaction w-as mutually exclusix e and mutually non-exclusive (cuhen drugs have different modes of action or act independently). Because wve did not map the entire response surface (Greco et al. 1995). the data of interaction of ODN 2009 with the chemotherapeutic agents calculated for each cell line rely on the fixed ratios of the agents.

RESULTS
Bcl-2 levels in SCLC cell lines and effect of treatment with ODN 2009 To demonstrate the abilitv of ODN 2009 to doxx n-regulate Bcl-2 expression in the SCLC cell lines. Westem blot analy sis was performed. Cells were incubated for 48 h with 75 nr\I or 150 nmt ODN 2009. Untreated cells were used to determine the basal levels of Bcl-2 in the cell lines. The relative basal levels of Bcl-2 which w-ere quantitated from the Western blots are shown in Figure 1.
Bcl-2 was abundantly expressed in the cell lines NCI-H69 (174%7c) and SW2 (100%c). but A-as barely detectable in the NCI-H82 cell line (23%   Figure 4A. the combination of ODN 2009 (75 nM) and doxorubicin (6.3 nm) reduced the number of viable SW2 cells to 10% of the untreated control during a 96-h treatment. Although synergistic cytotoxicity cannot be directly deduced from the viability curves. it is obvious that this combination was significantly more cytotoxic to the cells than treatment with equivalent concentrations of ODN 2009 or doxorubicin alone (P < 0.003). In contrast. the effects of a combination of ODN mismatch and doxorubicin. and doxorubicin alone were not significantly different (P > 0.2). This indicates that potentiation of the cytotoxicity of doxorubicin was specific for ODN 2009 and its ability to down-regulate bel-2 expression.
To demonstrate morphological changes of the cells typical for apoptosis. SW2 cells were subjected to forward and side light scatter analysis 72 h after treatment with 75 nM ODN 2009 in combination with 6.3 nm doxorubicin. As shown by the contour plots in Figure 4B. the treatment caused an obvious increase in side light scattering and a slight reduction in forward light scattering of the cells. Microscopic analysis of the treated cells revealed shrinkage, extensive plasma membrane blebbing and nuclear condensation (not shown).

Interaction between ODN 2009 and chemotherapeutic agents on SCLC cells expressing high levels of Bcl-2
Compared with the other cell lines used in this study. the NCI-H69 cell line has the highest level of Bc1-2 (Fioure 1. Table 1) Compared with the cell lines SW2 and NCI-H69. the NCI-H82 cell line has barely detectable Bcl-2 levels ( Figure 1. Table 1). The different slopes of the median-effect curves (data not show-n) suggest that on this cell line all combinations resulted in mutuall1 non-exclusive interactions. As shown in Finure 3. the combination of either ODN 2009 and etoposide at a ratio of 1:43 or ODN 2009 and doxorubicin at a ratio of 4:1 % as slightiv less than additive and resulted in antagonistic effects (CI > 1 ) at all lesels of toxicity.

Synergistic cytotoxicitv on NCI-H82 cells A-as obtained only if
ODN 2009 was combined with cisplatin (1:43) at the IC,0 and the ICR) of the combination with CI salues of 0.72 ± 0.08 and 0.95 ± 0.1 respectivelx (Figure 3.)

DISCUSSION
Etoposide. doxorubicin and cisplatin are routinel1 used for the treatment of SCLC. Thev A orkbv damaging, DNA w-hich tric-aers a common death programme called apoptosis (Strasser et al. 1994). The Bcl-2 oncoproteim can counteract drug-induced apoptosis and its expression has been associated xxith multidrucg resistance in a varietv of tumour cells (Mix ashita and Reed. 1993: Ohmori et al. 1993: Dole et al. 1994: Strasser et al. 1994. Here. we report on the cytotoxic effect of antisense-mediated down-regulation of bcl-2 expression in combination w-ith etoposide. doxorubicin and cisplatin on three SCLC cell lines. Bcl-' is abundantlx expressed in the cell line NCI-H69. whereas intermediate-high and low levels are present in the cell lines SW2 and NCI-H82 respectively (Ziegler et al. 1997). We obser ed an inverse correlation of the Bcl-2 levels of these cell lines w-ith the sensitivities to the chemotherapeutic agents tested. This suggests that Bcl-2 is critical for inhibiting drug-induced apoptosis in SCLC cells. Antisense ODNs have been used to disrupt the expression of various cancer related genes and to inhibit tumour cell growth in preclinical studies (Dosaka Akita et al. 1995: Monia et al. 1996: Kitada et al. 1994: Szczvlik et al. 1991. and first results of antitumour actixitv are also available from clinical studies (Webb et al. 1997). ODN 2009 is a 20-mer phosphorothioate that targ-ets the coding region of the bcl-2 mRNA (Ziecler et al. 1997). In SCLC cells it effectively down-regulated bcl-2 expression and induced apoptosis to a degree inversely correlated with the level of expression. This suggests Bcl-2 to be a critical sur-ix al factor for SCLC cells.
Combination chemotherapy has become the standard treatment for SCLC (Souhami and Law. 1990). The cytotoxic interactions of various chemotherapeutic agents have been analysed in xitro bv different calculation methods (Kaufmann et al. 1996: Photiou et al. 1997. and it has been show-n that even clinicallx approved drug combinations may result in less than additive effects (Kaufmann et al. 1996). These findings imply that such in vitro studies migCht be useful for the selection and design of optimal druga combinations for clinical application. Combinations of antisense ODNs targeting specific oncogenes or gyenes involved in drug, resistance. such as bcl-2. and less toxic doses of chemotherapeutic agents represent a rational therapeutic strategy to pursue. In models of human leukaemia and colon carcinoma xenografts in mice. it has been shown that antisense ODNs targeting the oncogenes berlabl (Skorski et al. 1997) or c-mvb (Del Bufalo et al. 1996. or the multiple drug resistance gene mdrl (Cucco and Calabretta. 1996) can indeed enhance the anti-tumour effect of chemotherapeutic agents. However. these studies did not directly insvestioate the interaction of the antisense ODNs with chemotherapeutic agents in vitro and did not search for synergistic combinations.
In the present study. w-e demonstrate for the first time that a combination of bcl-2 antisense ODN with etoposide. doxorubicin I from a solid tumour in which Bcl-2 is prevalent (Ben Ezra et al. 1994). Synergy was particularly pronounced on the cell lines NCI-H69 and SW2 that have high and intermediate-high Bcl-2 levels respectively. The cell line NCI-H82 which expresses barely detectable Bcl-2 levels was extremely sensitive to treatment with ODN 2009 and also to the chemotherapeutic agents alone. This miaht explain why. on this cell line. ODN 2009 and the chemotherapeutic agents did not interact synergistically.
Because the sulphur backbone of phosphorothioate ODNs nonspecifically interacts with proteins and nucleic acid targets (Stein andKrieg. 1994: Stein. 1995). it is difficult to determine w-hich of their bioloaical effects are truly antisense in nature. Therefore. in the present study. a series of control phosphorothioate ODNs was used for comparison to ODN 2009. At concentrations less than 150 n-. only the ODNs mismatch and 5'-3' reversed caused a slight reduction in cell viability. In combination experiments. none of the control ODNs potentiated the cytotoxic effects of the chemotherapeutic agents. This unequivocally excludes the possibilitv that the increase in cvtotoxicitv achieved w-ith the combinations of ODN 2009 and chemotherapeutic agents was the result of enhanced cellular uptake of the latter in the presence of cationic lipids (Bennett et al. 1992).
Although the intracellular targets and the mechanisms of action of ODN 2009 and the chemotherapeutic agents are different. their effects eventually merge in a common final death pathw-ay. In this case. one would expect the interactions between ODN 2009 and etoposide. doxorubicin or cisplatin to be mutually exclusive. For example. Bcl-2 has been shown to inhibit apoptosis induced by etoposide through effects on events early after topoisomerase IIinduced DNA damage (Kamesaki et al. 1993). According to Chou and Talalay ( 1984). agents are claimed to be mutually exclusive if the slopes of the median-effect lines of the single-agents and their combination are identical. Based on this assumption. however. on S'W2 cells a mutually exclusive interaction occurred betueen ODN 2009 and etoposide. but not between ODN 2009 and doxorubicin or cisplatin. On NCI-H69 cells. only the interaction of ODN 2009 and doxorubicin is suggested to be mutually exclusive. This indicates that: (1) cells can differ in the way they handle the apoptotic signals provided by ODN 2009 and the chemotherapeutic agents: (2) the assumption whether agents act independently or not cannot be made solely based on our current understanding of the initial steps in their mechanisms of action. Considerinc this uncertaintv. in the present study all data were analy-sed assuming both mutually exclusive and non-exclusive interactions of the agents.
In considerinc the potential implication of this study. one limitation must be kept in mind. The data for each cell line were generated using fixed ratios of the agents. If a different ratio had been evaluated. a different CI plot would have resulted. Thus. because the entire response surfaces (Greco et al. 1995) were not mapped. the conclusions of the study are limited to the ratios of ODN 2009 and chemotherapeutic agents that w ere actually used. Nevertheless. our data provide clear evidence that synergistic interactions of ODN 2009 and the chemotherapeutic agents can be expected to be more pronounced on cells expressino higher levels of Bcl-2. It will be of interest to see to what extent the differences bet-een synergistic and antagonistic effects observed in vitro translate to in vivo responses.
Our data suggest the use of ODN 2009 in combination with conventional chemotherapy as a novel approach to more effective treatment of SCLC and other types of tumours in which Bcl-2 is prevalent.